Patient preference to use a questionnaire varies according to attributes

Patient preference to use a questionnaire varies according to attributes

Patient Education and Counseling 84 (2011) 191–199 Contents lists available at ScienceDirect Patient Education and Counseling journal homepage: www...

395KB Sizes 0 Downloads 32 Views

Patient Education and Counseling 84 (2011) 191–199

Contents lists available at ScienceDirect

Patient Education and Counseling journal homepage:

Patient Perception, Preference and Participation

Patient preference to use a questionnaire varies according to attributes Na Yae Kim a,b,*, Lyndsay Richardson c, Weilin He a, Glenn Jones b,d a

University of Toronto, Canada Credit Valley Hospital, Canada c McGill University, Canada d McMaster University, Canada b



Article history: Received 26 February 2010 Received in revised form 30 July 2010 Accepted 29 August 2010

Objective: Health care professionals may assume questionnaires are burdensome to patients, and this limits their use in clinical settings and promotes simplification. However, patient adherence may improve by optimizing questionnaire attributes and contexts. Methods: This cross-sectional survey used Contingent Valuation methods to directly elicit patient preference for conventional monitoring of symptoms, versus adding a tool to monitoring. Under explicit consideration was the 10-question Edmonton Symptom Assessment System (ESAS). In the questionnaire, attributes of ESAS were sequentially altered to try and force preference reversal. A separate group of participants completed both questionnaire and interviews to explore questionnaire reliability, and extend validity. Results: Overall, 24 of 43 participants preferred using ESAS. Most important attributes to preference were frequency, specificity, and complexity. Where preference is initially against ESAS, it may reverse by simplifying the tool and its administrative processes. Interviews in 10 additional participants supported reproducibility and validity of the questionnaire method. Conclusions: Preference for using tools increases when tools are made relevant and used more appropriately. Practice implications: Questionnaires completed by patients as screening tools or aids to communication may be under-utilized. Optimization of ESAS and similar tools may be guided by empirical findings, including those obtained from Contingent Valuation methodologies. ß 2010 Elsevier Ireland Ltd. All rights reserved.

Keywords: Edmonton Symptom Assessment System (ESAS) Preference Contingent Valuation (CV) Breast cancer Radiation therapy

1. Introduction Conventional clinical encounters depend on verbal communication between patients and health care professionals, but interactions are constrained by staff time and this may contribute to low patient satisfaction. One strategy for improvement is to enhance visits by adding standardized questionnaires completed in advance by patients. Patient-completed tools provide a structured framework for screening or aiding communication, to guide clinical visits and decision-making while encouraging patients to be more actively engaged. Tools already provide greater understanding of patient experience in research trials [1], and generate data for clinical audit and program development [2]. However, few institutions have adopted even very short tools (e.g. Distress Thermometer) [3] for clinical visits. Staff may assume

* Corresponding author at: Department of Radiation Oncology, Credit Valley Hospital, Peel Regional Cancer Centre, 2200 Eglinton Ave West, Mississauga, ON, Canada L5M 2N1. Tel.: +1 905 813 2200; fax: +1 905 813 3962. E-mail address: [email protected] (N.Y. Kim). 0738-3991/$ – see front matter ß 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.pec.2010.08.018

large patient burden, which is supported by response rates to questionnaires in healthcare contexts that seldom exceed 70% [4]. Consequently, many tools have been revised down to ‘short-forms’ (e.g. Cancer Rehabilitation Evaluation System Short-Form) [5] or ‘ultra-short’ tools [6]. However, shorter tools ‘‘privilege’’ a few domains, over others, and may limit a tool’s relevance to fewer patients and clinical visits. Finding optimum characteristics to minimize patient (and staff) burden while maximizing relevance for a majority of visits is crucial for there to be wider acceptance of tools; patients must perceive an increase in personal welfare when using a tool, as compared with not using it, for it to be used regularly. Five years ago, Cancer Care Ontario (CCO) adopted the Edmonton Symptom Assessment System (ESAS) [2], a patientcompleted tool for use during visits with health care professionals at regional centres and with allied community services. Detailed reviews have demonstrated that most ESAS research studies pertain to palliative patients [7,8], the population for which this tool was developed [2]. Approximately 85% of patients find ESAS instructions to be clear, and ESAS to be easy to complete [9], though it is not known whether using ESAS improves patient satisfaction


N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199

or self-perceived welfare—these are broader concepts and experiences. In Radiation Oncology, a majority of patients receiving treatment are managed for cure with radiation administered 5 days per week for several weeks. ESAS has not been methodologically or statistically validated in this context [8]. Further, many radiotherapy patients are women with breast cancer who receive a 3- to 6-week course of radiation to a breast and who will experience local side effects. These are monitored and managed at a regular weekly visit with nurse and physician, and on a daily basis by radiotherapists. ESAS covers only nine general physical and emotional symptoms which are not specific to radiation. In short, the CCO mandate to administer ESAS in this context could be a burden to women with breast cancer undergoing radiotherapy, and such a wide implementation of ESAS forces the question of how best to optimize such tools to improve patient welfare. The crucial question is whether patients receiving curative treatments prefer using tools embedded within clinical processes, in contrast to continuing with conventional encounters that do not include tools. The properties of tools (attributes), tool-related clinical processes, and clinical culture (e.g. staff and patient expectations about methods of communication and ways of behaving) can influence patient preference for tools. Understanding such factors can guide optimization of tools, to improve use and clinical care. Accommodating patient values, perspectives and preferences for, and in, clinical visits, is an important component of Evidence-Based Medicine [10,11]. However, methodology of directly eliciting preference from patients is in its infancy. Possibly the best method to directly elicit preference is Contingent Valuation (CV), a set of procedures to measure a change in an individual’s welfare based upon a subjective valuation of a good or service. These procedures are especially well-established in non-medical contexts, with published guidelines to help assure reliability and validity [12–17]. Respondents imagine a market (not necessarily monetary) for accessing a good or service and construct preferences, analogous to the process of medical informed consent. The valuation of options is contingent upon their being possible in the real world. Guidelines help minimize bias and the play of chance in interviews and questionnaire surveys, so that ex ante declared preference should be predictive of behavior in the real world, should options become available [18]. Objectives of this study were to: (1) elicit preference for using two tools through the methods of CV; (2) determine the influence of ESAS attributes on preference, thereby estimating relative strengths of preference across attributes; (3) uncover contextual factors that might influence preference; and (4) methodologically compare questionnaire-CV and interview-CV.

comparison with ESAS (Appendix C). The PTA was compared to ESAS using previously reported methods [18–20] to assess withinperson consistency in how attributes in-the-aggregate predict preference when several attributes are simultaneously adjusted. Participants were asked to choose between tools and further to decide whether they would use both. Last, the questionnaire elicited socio-demographic and past health information. Medical data were extracted from hospital electronic charts (Appendix D). Patients eligible for phase one of this study were women with biopsy-confirmed breast cancer, who were attending the Peel Regional Cancer Centre (PRCC) for adjuvant breast radiotherapy with curative intent after completing all surgeries and any chemotherapy. Women with metastatic breast cancer were not eligible. All consecutive patients meeting these criteria between December 2008 and January 2009 were approached for study. A majority (61%) of patients were approached within the first quarter of their respective courses of radiation; others were approached later in treatment to assess influence of experience on responses. Language was not an exclusion criterion, though questionnaires were available only in English—only 3% of patients at PRCC are not able to communicate in English as a primary or secondary language. When providing consent and undergoing orientation and training, participants were not aware that the questionnaire would refer to ESAS. Consent and completed questionnaires were returned by participants within 24 h of being approached. Responses were reviewed for missing or inconsistent findings and women were contacted for completion or clarification as necessary. A separate cohort of consecutive patients meeting eligibility criteria was approached for phase two of this study between February and March 2009, to consent to both questionnaire and a 30-min interview. The ordering of these was randomized but content was identical. Interviews were conducted by a trained research assistant. All phase two participants were within their first week of radiotherapy, the time interval between interview and questionnaire was inside 1 week, and participants were financially compensated ($5). Data were analyzed with Stata 10 [21]. Variables of interest include all answers to CV questions and variables of patient, disease and treatment nature (as listed in Table 1 and Appendix D). Associations were explored by tabulation, T-tests, Chi-square, Fisher’s Exact, Spearman’s correlation, Analysis of Variance and Logistic regression. Baseline variables were compared across participation and according to initial direction of preference (for options one and two) (Table 1). Alpha was 0.05 with all p-values two-sided. Proportions of women holding a given preference are summarized as a per cent (%) along with a Clopper–Pearson 95% confidence interval (CI). Our focus is estimation of overall preference and how it changes in relation to attributes.

2. Methods 3. Results Ethics approval was provided by the Credit Valley Hospital Research Review Committee. All participant women provided written consent. In accordance with CV guidelines, we developed a detailed questionnaire that described options regarding using ESAS. Option one means care without ESAS; option two means care with ESAS. Pros and cons for the options were provided along with a copy of the 10-question ESAS with its 11-point scales (Appendix A). Depending upon initial choice, participants completed corresponding sub-sections of the questionnaire, which sequentially asked how tool attributes might influence preference. Text and interrogative sentences were designed to elicit at what point a participant would no longer adhere to initial choice and would invert preference. We targeted six attributes (Appendix B). Notably, participants were not asked to monetize benefits and costs of choices, so as to preserve the realism of the scenario. We also developed a Prototype Toxicity Assessment (PTA) for direct

Eighty-six consecutive women were approached and their characteristics are presented (Table 1). There were no significant differences between the 43 non-participants and the 43 participants. Nineteen participants selected option one (44%, CI 29–60%) and 24 selected option two (56%, CI 40–71%). Patient characteristics (Appendix D) that might influence choice were explored. Choosing option two was only associated with two variables: a higher proportion of completed-radiation-fractions when entering the study (p = 0.02); and claiming self-awareness of body sensation and symptoms (p = 0.04). Those who selected option one received questions about making ESAS simpler and easier. The purpose was to elicit at what point women would change their preference (away from not using ESAS). Of 19 participants there were five ‘resisters’ who chose to never fill out ESAS, no matter how simple and easy it

N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199


Table 1 Baseline characteristics by study phase and sub-groups of women. Phase 1 study

Phase 1 study participants (n = 43)a

Phase 2 study

Study decliners (n = 43)

Study participants (n = 43)

Option 1 no ESAS as CA (n = 19)

Option 2 ESAS as CA (n = 24)

Interview decliners (n = 9)

Interview participants (n = 10)

Median age (range) Marital status Without partner With partner Unknown Primary language English Non-English Unknown Ethnicity Caucasian Asian Southeast Asian African American Hispanic Not reported Education level Below high school High school College/university Post graduate Not reported Family yearly income <$35,000 $36,000–$55,000 $56,000–$90,000 >$90,000 Breast treatment Chemotherapy No prior chemotherapy Prior chemotherapy Radiation treatment fields 2 4 Radiation course Short (21 fractions) Long (25 fractions)

61 (27–79)

57 (29–77)

57 (40–77)

59 (29–76)

56 (43–78)

59 (35–68)

10 (23%) 27 (63%) 6 (14%)

9 27 7

(21%) (63%) (16%)

2 15 2

(11%) (79%) (11%)

7 12 5

(29%) (50%) (21%)

1 (11%) 8 (89%) 0 (0%)

1 (10%) 9 (90%) 0 (0%)

35 (81%) 5 (12%) 3 (7%)

37 3 3

(86%) (7%) (7%)

17 2 0

(89%) (11%) (0%)

20 1 3

(83%) (4%) (13%)

8 (89%) 1 (11%) 0 (0%)

10 (100%) 0 (0%) 0 (0%)


31 4 4 2 1 1

(72%) (9%) (9%) (5%) (2%) (2%)

10 3 2 2 1 1

(53%) (16%) (11%) (11%) (5%) (5%)

21 1 2 0 0 0

(88%) (4%) (8%) (0%) (0%) (0%)


9 0 0 1 0 0

(90%) (0%) (0%) (10%) (0%) (0%)


5 11 25 1 1

(12%) (26%) (58%) (2%) (2%)

3 7 8 0 1

(16%) (37%) (42%) (0%) (5%)

2 4 17 1 0

(8%) (17%) (71%) (4%) (0%)


0 4 6 0 0

(0%) (40%) (60%) (0%) (0%)


6 7 11 13

(14%) (16%) (26%) (30%)

2 4 4 3

(11%) (21%) (21%) (16%)

4 3 7 10

(17%) (13%) (29%) (42%)


1 4 4 0

(10%) (40%) (40%) (0%)

21 (49%) 22 (51%)

21 22

(49%) (51%)

11 8

(58%) (42%)

10 14

(42%) (58%)

3 (33%) 6 (67%)

5 (50%) 5 (50%)

28 (65%) 15 (35%)

30 13

(70%) (30%)

16 3

(84%) (16%)

14 10

(58%) (42%)

8 (89%) 1 (11%)

9 (90%) 1 (10%)

18 (42%) 25 (58%)

21 22

(49%) (51%)

10 9

(53%) (47%)

11 13

(46%) (54%)

3 (33%) 6 (67%)

5 (50%) 5 (50%)

Median fractions completeda (range)

6 (1–29)

4 (1–21)

2 (1–8)

4 (1–6)


3 (1–13)

7 (1–21)

Median radiation fractions completed differed (p = 0.02). All other p > 0.05, NS.

might become. There were no characteristics (Appendix D) that identified the 5/19 ‘resisters’. The other 14/19 reported they might use ESAS if attributes were changed favorably. Those who selected option two received questions about making ESAS more elaborate and harder to complete. The purpose was to elicit at what point women would change their preference (away from using ESAS). Of 24 participants there were four ‘supporters’ who chose to fill out the ESAS even if all six attributes were made more elaborate. No characteristics (Appendix D) identified the 4/24 ‘supporters’. The following paragraphs provide details regarding six attributes. Specificity, frequency and complexity had clearer impact on preference and are reported first. Findings with remaining attributes were weaker, and are subsequently summarized. Specificity: We asked how many questions out of the 10 in ESAS had to be specific to radiation side-effects to reverse initial preference. Of women who selected option one, 53% (CI 29–76%) would change preference and use ESAS if more specific questions replaced existing questions. The median number of specific questions required to reverse initial preference was four questions. Of women who selected option two, only 8% (CI 1–27%) said they would change preference and not use ESAS if more specific questions replaced existing questions. However, there was a limit on how many questions could be specific for some of these women; the median number to change preference and stop using ESAS was seven specific questions, leaving only three questions for general

symptoms. However, 10/24 women indicated preference to use ESAS even if all 10 questions were radiation-specific. Frequency: We asked what frequency of administration would reverse initial preference. Of women who selected option one, 42% (CI 20–67%) would change preference and use ESAS if they only had to complete ESAS every other week. Of women who selected option two, 75% (CI 53–90%) said they would change preference and not use ESAS if asked to complete ESAS more than once a week. Only six women indicated a willingness to use ESAS more than once a week—four chose every 3 days; two chose daily. Complexity: We asked how changes to ESAS scales might reverse initial preference. Of women who selected option one, 58% (CI 33–80%) would change preference and use ESAS with simplification of scales, and a four-tick-box scale with verbal descriptors for each box was the most commonly selected replacement for the 11-point numeric rating scale of ESAS. Of women who selected option two, only 17% (CI 5–37%) would change preference and not use ESAS if it utilized different scales for different questions (i.e. variety). Remaining attributes (Length, Scope, and Medium): For the attribute of length, 53% (CI 29–76%) of women who selected option one would reverse preference and use ESAS if it were half as long (i.e. five questions). Of women who selected option two, 38% (CI 19–59%) would abandon ESAS if it were exactly doubled in length. Responses were typically rounded (nearest five or ten) by


N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199

respondents even though discrete numbers were provided in the answer sets and space was provided to write a specific number for a preference of greater than 20 questions. For the attribute of scope, women preferred a combination of questions about physical and emotional symptoms. For the 19 women who chose option one, there was a preference for more than seven questions to be about physical symptoms, while only one of all 43 women preferred more questions about emotional symptoms, than physical symptoms. For the attribute of medium of presentation, 67% (CI 51–81%) of all 43 women preferred paper-and-pencil over electronic media, relating to greater age (p = 0.04), with no difference in preference for medium by initial preference to not use, or to use ESAS. Two questions were asked regarding the Prototype Toxicity Assessment (PTA). Women directly chose between ESAS and PTA, assuming weekly use. Three-quarters of all 43 women selected PTA over ESAS. There were no significant associations between this preference and patient-disease-treatment characteristics (Appendix D). However, those who preferred the PTA had preferred more radiation-specific questions for ESAS (p = 0.01). Women were also asked if they would complete both PTA and ESAS together and 54% (CI 38–69%) were not willing to complete both every week. Willingness to complete both tools (a combined 24 questions) was associated with accepting longer tools (p = 0.01) and having initially selected option two (to use ESAS) on earlier questioning (p < 0.0005). Nineteen additional women were approached to participate in both interviews and questionnaires. Characteristics of the 10 participants and the nine non-participants did not differ significantly (Table 1 and Appendix D) except in that participants had more typically written out symptom lists prior to seeing their doctors in past visits (p = 0.04). Within the 10 participants, those randomized first to complete the interview differed only in the number of ESAS they had completed prior to study entry (p = 0.03). A direct comparison of the within-participant answers obtained through interview and paper-based questionnaire methods evidenced only minor differences. Responses were an exact match within-subject for 85% of all questions. Importantly, there were no differences in initial preference and when comparing PTA and ESAS. For the 15% of responses where there were differences between questionnaire and interview results, a majority (8% of this 15%) of these were concerning point-estimates where initial preference would reverse based on the relative balance of physicalto-emotional items (scope) and the relative balance of specific-togeneral items (specificity). Notably, emotions and radiation sideeffects can evolve within 1 week of radiotherapy so the perceived importance of scope and specificity may be dynamic as compared with other attributes. Of the remaining differences between interview and questionnaire responses, women who experienced the 30-min interview prior to doing the questionnaire indicated a greater willingness in the questionnaire to use lengthier tools (p = 0.01). Therefore, the overall pattern of findings implies that the paper-based self-reporting questionnaire yielded similar results to interviews in this particular application of CV. 4. Discussion and conclusion 4.1. Discussion Our purpose was to explore strategies optimizing patient selfreport tools about symptoms. We selected women with breast cancer, a very common cancer with annual worldwide incidence exceeding 1.15 million [22]. We focused on attributes that might influence preference for using tools, using the method of Contingent Valuation (CV) to assess personal welfare. Methods of CV may adhere to economic theory [23,24]. Economics is the science of choice [25] and selecting prospects that improve

personal welfare can be consistent with neo-classical welfare economics [26,27] and several of its extensions [18,28]. Validity and reproducibility in our study are supported by adherence to CV guidelines, internal consistency of findings in phase one, concordance of results across both phases of this study, and 85% complete agreement between responses obtained with interviews and questionnaires. In our study 24/43 (56%, CI 40–71%) women receiving curative radiotherapy selected ESAS (as in Appendix B) for weekly use. This increased somewhat when tool attributes were improved from the patient perspective, including a reasonable frequency of completion, greater relevance, and a more sensical scaling of answer options. That ideal tools are brief and relevant is consistent with the literature [29], but our findings imply that emphasizing only short and ultra-short tools [5,6] may be self-defeating for health care professionals. Staff may assume a tool is burdensome for patients, but patients may perceive that using a tool improves personal welfare in excess of burdens and costs. In our study, women appeared to be acting as rational decision-makers when making this choice [22,24,26]. Other interesting attributes are the balance between physically-oriented items and items oriented to other domains (e.g. psychosocial and supportive items), and whether the tool is specific or more relevant to the immediate context. Women preferred more questions targeting physical symptoms and for some questions to be specific to radiation. Only 20/43 expressed interest in using both PTA and ESAS, suggesting that the two tools (combined total of 24 questions) are complementary. In addition, the relevance of a questionnaire for patients commencing treatment may be increased by working with patient expectations for treatment side-effects, and by adjusting scales to capture mild symptoms as non-zero values. In our study, having completed more treatment at study entry (by which time side-effects may have emerged) was associated with preference for using ESAS. However, ESAS is not yet established for use in curative contexts, from a design or psychometric perspective [8]. In particular, prevalence and severity of symptoms identified in ESAS are low with ‘‘floor-effects’’ in response distributions, even in palliative or chemotherapy situations [30–34]. We conclude, then, that ESAS in its present form [2,7,8] is not optimized in attributes or contexts for use in Ontario. More research is required to determine whether ESAS can be altered to improve preference while improving its psychometrics (reliability, validity and responsiveness). Research is also needed to compare optimized static tools like ESAS with dynamic tools where responses to initial questions select next-questions from a question bank, leading to more precise symptom estimation. For example, the PatientReported Outcomes Measurement Information System being developed by the National Institute of Health [35] applies itemresponse theory and question banks for estimating a broad range of symptoms and functionality. Besides intrinsic psychometric properties and patient preference for tool content (e.g. relevance, focus) and structure (e.g. scales), the contexts around offering a tool are important. First, preference may depend on expectations regarding staff responses. Given limited time for clinical encounters, responses to toolidentified problems ideally would be established in advance as clinical policies, and would be applicable to most patients. For symptoms like nausea and pain this is relatively simple—an already well-established practice. However, it may not be realistic to implement specific responses for every item in a tool. Issues are rarely isolated, nor warrant a singular response. Items like ESAS ‘well-being’ have no obvious clinical response [8,32]. Well-being, anxiety and depression may require complex assessment and responses. Without clarity for all items in a tool, it may be difficult to ask patients to regularly answer all items. Second, preference for ESAS may depend on patients understanding its broader purpose

N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199

or intent. After development, ESAS was identified more as a screening tool for some common cancer symptoms [2,8] as well as to follow them over time [2]. Both applications require frequent, perhaps regular and systematic administration. In contrast, ESAS could be construed as an aid to facilitate communication with staff— i.e. a Communication Aid) (CA). This terminology to describe tools places greater emphasis on context. A tool might be made available for use when needed, as when patients feel there is something significant to communicate. Whether tools should be evolved in the direction of screening or as CAs depends on patients accepting the intent, the processes in which a tool is embedded, and the likelihood and expected value of possible clinical responses by staff (assessment, treatment or referral). Whether a single tool can be used for routine staff-driven screening as well as an optional patientinitiated CA requires more research. A clinical program or organization typically wants only one tool maximizing relevant communication and minimizing burden, but our study suggests this may be difficult to achieve. Women who initially chose to use ESAS were more homogeneous in how attributes influenced their preference reversals. But women who chose option one gave more varied responses—some were absolute decliners while others accepted altered ESASs but differed in how attributes should be changed to gain their approval. Careful program planning for optimally implementing tools has to address patient diversity. There are several study limitations. First, we focused on ESAS which is only one of many available tools differing across attributes. However, the methods of CV can be used to explore preference for other tools. Second, ESAS was being used in our clinics. Many of our participants had experience with it as a screening tool, which might have made it more difficult to consider a hypothetical choice to use it or not. Third, 50% of the women we approached did not participate. However, non-participants were similar in their characteristics to participants. Fourth, our sample size is only 43 women, with a majority of these being Caucasian with college or university educations and mid- to high-income ranges. These features compromise the external validity of our findings. Fifth, face-to-face interviews of 30 to 60 min are recommended by CV guidelines, otherwise using paper questionnaires should be validated against some interviews. We conducted only 10 interviews, but these gave very similar results to those of paper questionnaires, and the order of completion was randomized, to strengthen validity. We found that the relative benefit of an all-interview strategy would be insufficient to offset the cost in resources. Last, the PTA is not a validated tool. Created for comparison and to explore respondent consistency, it is not recommended for clinical practice.


When implementing ESAS, CCO focused on screening for general symptoms. It became apparent that ESAS is more useful to enhance communication with health care professionals, allied health providers in the community, and family and friends who may be care-givers also. It is unknown whether patients prefer one tool across these three contexts. A randomized trial is needed to compare adherence to completing tools that differ in attributes and contexts of use. This might confirm the presence of attribute-boundaries where preference reversals occur. Research could explore whether attribute-boundaries are flexible, changing with cultures of medical care including expectations about confidentiality, and comfort using a questionnaire to communicate personal information, as well as changing with the physical environment where a questionnaire is used. Moving attribute-boundaries could expand the ranges of those attributes within which patients prefer to use tools, increasing patient utilization. This is particularly important when transferring a complex research instrument to a clinical context. Tools must be optimized in both psychometrics (reliability, validity and responsiveness) and acceptability to patients (values, welfare and preference). These criteria may require trade-offs. Empirical data provided by CV methods, and randomized trials assessing behavioral outcomes, may help identify modifications, minimizing trade-offs and maximizing patient utilization. 4.2. Conclusion Methods of Contingent Valuation to directly elicit patient preference and strength of preference can provide guidance from patients regarding optimizing tools for self-reporting of symptoms. It is useful to delineate the values of a tool’s attributes at which patients perceive a maximum of personal welfare relative to burden when using a tool, either for screening or as an aid to communication at visits with staff. A reliable and valid tool that best corresponds to a consensus of acceptable attributes and contextual factors should receive broadest support from patients and staff. 4.3. Practice implications Professional health providers should not make unwarranted assumptions about tools for self-reporting of symptoms, and to what extent a particular tool is a burden for patients. Burden is only one half of the calculus; improved welfare is the other half. Systematic questioning of patients defines how best to optimize a tool from their perspectives, and determines in what contexts it makes sense for them to choose to use it. Staff can reinforce use with meaningful and timely responses to patient concerns.


N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199

Appendix A. Edmonton Symptom Assessment System (ESAS)

N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199


Appendix B. Listing of core interrogative sentences1 Interrogative sentences given to participants who selected option one, care without ESAS Q1: Would you choose option 2 and use ESAS once a week if it were half as long, meaning only 5 questions? Q2: Would you choose option 2 and use the 10-item ESAS if you were asked to fill it out no more than once every 2 weeks? Q3: Would you choose option 2 and use the 10-item ESAS once a week if it had more than 6 physical symptom questions and fewer than 3 emotional symptom questions? Q4: Would you choose option 2 and use the 10-item ESAS once a week if it were more specific, targeting local and regional side effects from radiation treatment? Q5: Would you choose option 2 and use the 10-item ESAS once per week if it were simpler or easier to fill out? Q6: Would you choose option 2 and add the 10-item ESAS if it were available in electronic format, rather than as a paper and pencil questionnaire? Q7: Would you now choose option 2, using this NEW questionnaire once a week, rather than stick with option 1, and not using any questionnaire? Q8: If you have to choose one of the two forms to use once a week, which would you prefer to do before you see your nurse or doctor? Interrogative sentences given to participants who selected option two, care with ESAS Q1: Would you still complete ESAS once a week if it were twice as long, meaning 20 questions? Q2: Would you fill out the 10-item ESAS more than once per week? Q3: Would you still use the 10-item ESAS once a week if it had more than 6 physical symptom questions and fewer than 3 emotional symptom questions? Q4: Would you fill out the 10-item ESAS if it were more specific for radiation treatment, targeting local and regional side effects from radiation? Q5: Would you still use the 10-item ESAS once per week if it were harder to fill out? Q6: Would you still be willing to complete the 10-item ESAS if it were available in electronic format, rather than as a paper and pencil questionnaire? Q7: Would you use this NEW questionnaire once a week instead of using ESAS? Q8: If you have to choose one of the two forms to use once a week, which would you prefer to do before you see your nurse or doctor?

1 The CV questionnaires and interviews include preamble sentences, paragraphs, and secondary interrogative sentences. These, combined with subject training, made it clear that after stating an initial preference for option one or two, the questions were intended to identify if, and at what point, preference would change from one option to the other. Q7 and Q8, instead, asked subjects to directly compare two questionnaires (Q7) and then decide whether to do both every week (Q8).


N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199

Appendix C. Prototype Toxicity Assessment (PTA, does not include diagrams)

N.Y. Kim et al. / Patient Education and Counseling 84 (2011) 191–199


Appendix D. Patient, disease and treatment related characteristics (62 total variables) Patient characteristics Socio-demographics

Age, height, weight, language, ethnicity, education, family income, marital status

Skin and breast histories

Personal and family histories for skin cancers, burns, skin allergies, keloids, prior cysts, mastitis, menstrual changes, Bra and cup size Surgery to breast, cosmetic result from breast surgery for cancer, skin effects from any chemotherapy

Treatment variables

Complications (type, duration) from breast surgery for cancer Prior chemotherapy Radiation plan (fields, duration), fractions completeda Radiation team

General symptoms

Quality of rest, feeling refreshed with rest/sleep

Communication experiences

Being aware of body sensation and symptomsa Ability to describe sensations and symptoms, and interpret them Problems with listing, describing, or explaining side effects and symptoms Forgetting to mention the side effects and symptoms in routine visits Making a list and referring to it during a clinical visit How a list influenced the visit (duration, tests, greater anxiety)


For 60 variables, p > 0.05. For fractions-completed p = 0.03, and for being aware of body sensation and symptoms, p = 0.04. (The question was: ‘‘How aware are you of your body and its sensations and symptoms?’’ The five possible answers ranged from ‘‘very aware’’ to ‘‘not at all’’).

References [1] Davidson SE, Trotti A, Ataman OU, Seong J, Nee Lau F, Motta NW, et al. Improving the capture of adverse event data in clinical trials: the role of the international atomic energy agency (IAEA). Int J Radiat Oncol Biol Phys 2007;69:1218–21. [2] Bruera E, Kuehn N, Miller MJ, Selmser P, MacMillan K. The Edmonton Symptom Assessment System (ESAS): a simple method for the assessment of palliative care patients. J Palliat Care 1991;7:6–9. [3] Jacobsen PB. Screening for psychological distress in cancer patients: challenges and opportunities. J Clin Oncol 2007;25:4526–7. [4] Sitzia J, Wood N. Response rate in patient satisfaction research: an analysis of 210 published studies. Int J Qual Health Care 1998;10:311–7. [5] Schag AC, Ganz PA, Heinrich RL. Cancer rehabilitation evaluation system-short form (CARES-SF). A cancer specific rehabilitation and quality of life instrument. Cancer 1991;68:1406–13. [6] Mitchell AJ. Pooled results from 38 analyses of the accuracy of distress thermometer and other ultra-short methods of detecting cancer-related mood disorders. J Clin Oncol 2007;25:4670–81. [7] Nekolaichuk C, Watanabe S, Beaumont C. The Edmonton Symptom Assessment System: a 15-year retrospective review of validation studies (1991–2006). Palliat Med 2008;22:111–22. [8] Richardson LA, Jones GW. A review of the reliability and validity of the Edmonton Symptom Assessment System. Curr Oncol 2009;16:55. [9] Watanabe S, Nekolaichuk C, Beaumond C, Mawani A. The Edmonton Symptom Assessment System—what do patients think? Support Care Cancer 2009;17:675–83. [10] Victor MM, Guyatt GH. Progress in evidence-based medicine. J Am Med Assoc 2008;300:1814–6. [11] Sackett DL, Rosenberg WM, Gray JAM, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn’t. Brit Med J 1996;312:71–2. [12] Appendix I Report of the NOAA on contingent valuation. Fed Regist 1993;58(January):4602–14. [13] Brown RA, Green CH. Threats to health or safety: perceived risk and willingness-to-pay. Soc Sci Med 1981;15C:67–75. [14] Carson RT, Mitchell RC. Sequencing and nesting in contingent valuation surveys. J Environ Econ Manage 1995;25:155–73. [15] Gafni A. Willingness-to-pay as a measure of benefits: relevant questions in the context of public decision making about health care programs. Med Care 1991;29:1246–52. [16] McFadden D, Leonard GK. Issues in the contingent valuation of environmental goods: methodologies for data collection and analysis. In: Hausman JA, editor. Contingent valuation. A critical assessment. New York, NY: Elsevier Science Publishers; 1993. p. 165–215.

[17] Morrison GC, Gyldmark M. Appraising the use of contingent valuation. Health Econ 1992;1:233–43. [18] Jones GW. A justification and adaptation in a methodology of contingent valuation (willingness-to-pay) to measure social preferences for allocating Medicare services under a binding budget with an example from radiation oncology. M.Sc. Thesis, McMaster University, Hamilton, ON, Canada; 1995. [19] Wright J, Jones G, Whelan TH. Patient preference for high or low dose rate Brachytherapy in carcinoma of the cervix. Radiother Oncol 1994;33:187–94. [20] McCombs K, Chen SC. Patient preference quality of life measures in dermatology. Derm Ther 2007;20:102–9. [21] Statacorp. Stata Statistical Software: Release 10. College Station, TX: StataCorp LP; 2007. [22] Coughlin SS, Ekwueme DU. Breast cancer as a global health concern. Cancer Epidemiol 2009;33:315–8. [23] Mitchell RC, Carson RT. Using surveys to value public goods: the contingent valuation method. Washington, DC: Resources for the Future; 1989. [24] Hausman JA, editor. Contingent valuation. A critical assessment. New York, NY: Elsevier Science Publishers; 1993. [25] Lipsey RG, Harbury C. First principles of economics. London: Weidenfeld and Nicolson; 1988. p. 7. [26] Gafni A, Birch S, Mehrez A. Economics, health and health economics: HYEs versus QALYs. J Health Econ 1993;12:325–39. [27] Wagstaff A. Health care: QALYs and the equity-efficiency tradeoff. J Health Econ 1991;10:21–41. [28] Pattanaik PK. Essays on Individual Decision-Making and Social Welfare. New Delhi, India: Oxford University Press; 2009. [29] Fayers PM, Machin D. Quality of life: the assessment, analysis and interpretation of patient-reported outcomes, second ed., Chichester: John Wiley and Sons; 2007. [30] Bradley N, Davis L, Chow E. Symptom distress in patients attending an outpatient palliative radiotherapy clinic. J Pain Symptom Manage 2005;30:18–23. [31] Chang VT, Hwang SS, Feuerman M. Validation of the Edmonton Symptom Assessment Scale. Cancer 2000;88:2164–71. [32] Moro C, Brunelli C, Miccinesi G, Fallai M, Morino P, Piazza M, et al. The Edmonton Symptom Assessment Scale: Italian validation in two palliative care settings. Support Care Cancer 2006;14:30–7. [33] Salminen E, Clemens KE, Syrja¨nen K, Salmenoja H. Needs of developing skills of palliative care at the oncology ward: an audit of symptoms among 203 consecutive cancer patients in Finland. Support Care Cancer 2008;16:3–8. [34] Vignaroli E, Pace EA, Willey J, Palmer JL, Zhang T, Bruera E. The Edmonton Symptom Assessment System as a screening tool for depression and anxiety. J Palliat Med 2006;9:296–303. [35] Home website: