Advertisement

Patient Preferences for Anxiety and Depression Screening in Cancer Care: A Discrete Choice Experiment

Open ArchivePublished:August 31, 2021DOI:https://doi.org/10.1016/j.jval.2021.05.014

      Abstract

      Objectives

      Screening for anxiety and depression in cancer care is recommended, as identification is the first step in managing anxiety and depression. Nevertheless, patient preferences for anxiety and depression screening in cancer care are unknown. The objective of this study was to investigate and identify the aspects of an anxiety and depression screening program cancer patients value most, to inform decision-makers about ways to improve patient uptake and ultimately, the provision of patient-centered care.

      Methods

      A discrete choice experiment was designed and implemented within an Australian cancer population sample. Participants were presented with a series of hypothetical screening programs labeled as “screening program 1” and “screening program 2” and were asked to choose their preferred one. The discrete choice experiment was administered using an online survey platform. A mixed logit and a latent class analysis was conducted.

      Results

      Participants (n = 294) preferred screening to be conducted by a cancer nurse, face-to-face, and at regular intervals (monthly or every 3 months). Participants also preferred follow-up care to be delivered by mental health professionals embedded within the cancer care team. Factors that influenced preferences were the low cost and short waiting times for access to care.

      Conclusions

      Cancer patients prefer cancer services with integrated mental healthcare services. To maximize patient uptake, anxiety and depression screening programs should be routinely offered, delivered by oncology healthcare staff in a face-to-face format, and, postscreening, to be care for by mental health professionals embedded within the cancer service.

      Keywords

      Introduction

      People living with a diagnosis of cancer often experience high levels of anxiety and depression (A&D), and this has been observed across different demographic backgrounds, cancer types, and stages.
      • Linden W.
      • Vodermaier A.
      • Mackenzie R.
      • Greig D.
      Anxiety and depression after cancer diagnosis: prevalence rates by cancer type, gender, and age.
      ,
      • Zabora J.
      • BrintzenhofeSzoc K.
      • Curbow B.
      • Hooker C.
      • Piantadosi S.
      The prevalence of psychological distress by cancer site.
      Comorbid A&D experienced by patients with cancer are strongly associated with poorer quality of life,
      • Annunziata M.A.
      • Muzzatti B.
      • Giovannini L.
      • et al.
      Is long-term cancer survivors’ quality of life comparable to that of the general population? An Italian study.
      poorer survival,
      • Chan C.M.
      • Wan Ahmad W.A.
      • Yusof M.M.
      • Ho G.F.
      • Krupat E.
      Effects of depression and anxiety on mortality in a mixed cancer group: a longitudinal approach using standardised diagnostic interviews.
      and an increased risk of suicide.
      • Buzgova R.
      • Jarosova D.
      • Hajnova E.
      Assessing anxiety and depression with respect to the quality of life in cancer inpatients receiving palliative care.
      In busy cancer services, A&D are often undetected or underestimated,
      • Fallowfield L.
      • Ratcliffe D.
      • Jenkins V.
      • Saul J.
      Psychiatric morbidity and its recognition by doctors in patients with cancer.
      and many patients report unmet needs for psychosocial care.
      • Rob S.-F.
      • Afaf G.
      • Allison B.
      • Billie B.
      • Louise B.
      • Peter C.
      The unmet supportive care needs of patients with cancer.
      A&D are treatable, and a strong evidence base for interventions exists.
      • Osborn R.L.
      • Demoncada A.C.
      • Feuerstein M.
      Psychosocial interventions for depression, anxiety, and quality of life in cancer survivors: meta-analyses.
      ,
      • Chien C.-H.
      • Liu K.-L.
      • Chien H.-T.
      • Liu H.-E.
      The effects of psychosocial strategies on anxiety and depression of patients diagnosed with prostate cancer: a systematic review.
      Early identification and treatment of A&D not only result in improved psychological outcomes but also greater cancer treatment adherence, improved doctor-patient communication, and fewer clinic calls and visits.
      • Howell D.
      • Hack T.F.
      • et al.
      Cancer Journey Survivorship Expert Panel
      Survivorship services for adult cancer populations: a pan-Canadian guideline.
      Several international and national guidelines exist for the assessment and management of A&D in cancer and general populations.
      • Howell D.
      • Keshavarz H.
      • Esplen M.
      • et al.
      A Pan-Canadian Practice Guideline: Screening, Assessment and Management of Psychosocial Distress, Depression and Anxiety in Adults with Cancer.
      • Ellis P.
      Australian and New Zealand clinical practice guidelines for the treatment of depression.
      • Holland J.C.
      • Andersen B.
      • Breitbart W.S.
      • et al.
      Distress management: clinical practice guidelines in Oncology.
      • Andersen B.L.
      • DeRubeis R.J.
      • Berman B.S.
      • et al.
      Screening, assessment, and care of anxiety and depressive symptoms in adults with cancer: an American Society of Clinical Oncology guideline adaptation.
      Despite this, few Australian cancer services routinely screen patients for A&D, and if screening does occur, patterns of referral, treatment, and follow-up are highly variable.
      PoCoG
      Clinical Pathway for the Screening, Assessment and Management of Anxiety and Depression in Adult Cancer Patients.
      The Psycho-oncology Co-operative Research Group (PoCoG) an Australian Cancer Clinical Trial Group, has developed the first Australian model of psycho-oncology care—the Australian Clinical Pathway for the Screening, Assessment, and Management of A&D in Adult Cancer Patients (ADAPT-CP).
      PoCoG
      Clinical Pathway for the Screening, Assessment and Management of Anxiety and Depression in Adult Cancer Patients.
      The ADAPT-CP was implemented across 12 clinical sites in New South Wales, Australia, as a part of the ADAPT randomized controlled trial.
      • Butow P.
      • Shaw J.
      • Shepherd H.L.
      • et al.
      Comparison of implementation strategies to influence adherence to the clinical pathway for screening, assessment and management of anxiety and depression in adult cancer patients (ADAPT CP): study protocol of a cluster randomised controlled trial.
      The ADAPT-CP standardizes the identification and management of A&D in cancer care. It recommends that all patients who attend a cancer service are routinely screened for A&D and on the basis of symptom severity allocated into a stepped-care model for treatment and follow-up. This stepped-care model incorporates 5 steps—from universal care and self-management for those with minimal or mild levels of A&D to specialist mental healthcare for those with severe levels A&D, with review and change in steps where necessary. Evidence-based recommendations on staff responsibilities and content and timing of interventions are provided for each step and tailored to available resources.
      Formalized routine screening for A&D using a validated screening tool is the first step of the pathway. If screening does not occur, subsequent phases (treatment and follow-up) of the ADAPT-CP may fail to be initiated.
      One of the underlying determinants of screening uptake is patient preferences; health services need to be acceptable and convenient for patients.
      • Carlson L.E.
      • Waller A.
      • Mitchell A.J.
      Screening for distress and unmet needs in patients with cancer: review and recommendations.
      Patient preferences for A&D treatments have been explored in the literature, and studies suggest that matching patient preferences has the potential to improve the uptake of A&D treatments. Dwight-Johnson et al
      • Dwight-Johnson M.
      • Lagomasino I.T.
      • Aisenberg E.
      • Hay J.
      Using conjoint analysis to assess depression treatment preferences among low-income Latinos.
      reported in primary care settings that depression treatment was more acceptable when assistance with the logistics of treatment (transportation and telephone consultations) was provided. Lokkerbol et al
      • Lokkerbol J.
      • van Voorthuijsen J.M.
      • Geomini A.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with anxiety disorder.
      ,
      • Lokkerbol J.
      • Geomini A.
      • van Voorthuijsen J.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with depression.
      found that patients with A&D preferred face-to-face treatment and shorter waiting times. Even after accounting for age and education, significant preference variations exist for the treatment of A&D. Lokkerbol et al
      • Lokkerbol J.
      • van Voorthuijsen J.M.
      • Geomini A.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with anxiety disorder.
      ,
      • Lokkerbol J.
      • Geomini A.
      • van Voorthuijsen J.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with depression.
      highlighted the importance of matching treatments to individual patients to enhance treatment uptake where possible. Hobden et al
      • Hobden B.
      • Turon H.
      • Bryant J.
      • Wall L.
      • Brown S.
      • Sanson-Fisher R.
      Oncology patient preferences for depression care: a discrete choice experiment.
      found that the level of concern about depression of patients with cancer could influence the type of care (clinician-directed or self-directed) they would like to receive and have noted that this will have implications for depression screening in clinical practice. In the same way, we believe that the uptake of A&D screening could be enhanced in the cancer population by understanding patient preferences. To the best of our knowledge, patient preferences for A&D screening in the cancer population have not yet been explored.
      Patient preferences drive uptake of a health intervention, and, therefore, it is essential to identify features of an intervention that improve uptake.
      • Viney R.
      • Lancsar E.
      • Louviere J.
      Discrete choice experiments to measure consumer preferences for health and healthcare.
      This article uses a discrete choice experiment (DCE) to identify and measure patient preferences concerning screening for A&D in cancer care. DCEs have been previously used to predict the uptake of health interventions in a range of health settings, for example, primary care,
      • Cheraghi-Sohi S.
      • Bower P.
      • Mead N.
      • McDonald R.
      • Whalley D.
      • Roland M.
      Making sense of patient priorities: applying discrete choice methods in primary care using ‘think aloud’ technique.
      palliative care,
      • Gomes B.
      • de Brito M.
      • Sarmento V.P.
      • et al.
      Valuing attributes of home palliative care with service users: a pilot discrete choice experiment.
      and cancer screening.
      • Kohler R.E.
      • Lee C.N.
      • Gopal S.
      • Reeve B.B.
      • Weiner B.J.
      • Wheeler S.B.
      Developing a discrete choice experiment in Malawi: eliciting preferences for breast cancer early detection services.
      This DCE aims to determine the features of an A&D screening program cancer patients value most, which can then inform decision makers about ways to improve the provision of patient-centered care.

      Methods

      Identification and Development of Attributes and Levels

      Attributes and levels were identified through a review of the literature and on the basis of recommendations outlined in the ADAPT-CP.
      PoCoG
      Clinical Pathway for the Screening, Assessment and Management of Anxiety and Depression in Adult Cancer Patients.
      MEDLINE, Cochrane, EconLit, PsycINFO, and PsycARTICLES were searched using medical subject headings and keyword terms: (((discrete choice or DCE or conjoint analysis) or discrete choice experiment∗ or (valuation and (DCE or conjoint analysis)) and (anxiety or depression)). Only full-text DCE articles reporting results for adults with cancer and published in English between 1999 and 2019 were included. A total of 142 articles were identified, and 20 studies met the inclusion criteria. A total of 32 attributes were extracted and categorized (see Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.05.014). Only one
      • Dwight Johnson M.
      • Apesoa-Varano C.
      • Hay J.
      • Unutzer J.
      • Hinton L.
      Depression treatment preferences of older white and Mexican origin men.
      included attributes concerning A&D screening. We decided to split the delivery of screening attribute used by Dwight Johnson et al
      • Dwight Johnson M.
      • Apesoa-Varano C.
      • Hay J.
      • Unutzer J.
      • Hinton L.
      Depression treatment preferences of older white and Mexican origin men.
      into 2 attributes to be able to elicit preferences for (1) the method of screening (online form, pen and paper form, face-to-face interview, phone call) and (2) the health professional involved (social worker, psychologist, cancer nurse, cancer doctor, general practitioner).
      To reflect preferences toward a screening program like the ADAPT-CP, the selection of the attributes was informed by the 3 components of the ADAPT-CP: (1) screening, (2) intervention, and (3) review and follow-up. Attribute selection was also informed by clinical knowledge and experience from psycho-oncology experts at PoCoG.

      Development

      An initial list of attributes and levels was presented to and reviewed by health economics experts at the Centre for Health Economics Research and Evaluation, University of Technology, and psycho-oncology experts at PoCoG, University of Sydney.
      A choice experiment was then developed in which cancer respondents were asked to consider screening for A&D (the vignette). Participants were then presented with a series of choice sets each describing 2 alternative screening services and, in each choice set, asked to state their preference.
      Cognitive interviews were used to test respondent comprehension and interpretation of DCE attributes.
      • Katz D.A.
      • Stewart K.R.
      • Paez M.
      • et al.
      Development of a discrete choice experiment (DCE) questionnaire to understand veterans’ preferences for tobacco treatment in primary care.
      ,
      • Schildmann K.E.
      • Groeneveld I.E.
      • Denzel J.
      • et al.
      Discovering the hidden benefits of cognitive interviewing in two languages: the first phase of a validation study of the integrated palliative care outcome scale.
      Cognitive interviews are commonly used when developing DCEs in health.
      • Cheraghi-Sohi S.
      • Bower P.
      • Mead N.
      • McDonald R.
      • Whalley D.
      • Roland M.
      Making sense of patient priorities: applying discrete choice methods in primary care using ‘think aloud’ technique.
      • Gomes B.
      • de Brito M.
      • Sarmento V.P.
      • et al.
      Valuing attributes of home palliative care with service users: a pilot discrete choice experiment.
      • Kohler R.E.
      • Lee C.N.
      • Gopal S.
      • Reeve B.B.
      • Weiner B.J.
      • Wheeler S.B.
      Developing a discrete choice experiment in Malawi: eliciting preferences for breast cancer early detection services.
      Participants for the cognitive interviews were recruited from 2 sources: an online panel provider, Stable Research (SR), and a survey and questionnaires group within a breast cancer support and advocacy group, Breast Cancer Network Australia (BCNA).
      An interview schedule guided the interviews, and participant comprehension was assessed across 4 components: (1) comprehension of the attribute item, (2) retrieval of relevant information, (3) use of that information to make a judgment, and (4) providing a response.
      • Tourangeau R.
      • Rips L.J.
      • Rasinski K.A.
      The Psychology of Survey Response.
      After completing all choice sets, participants were also asked whether there were any other attributes that would affect their choice.
      Seven participants, aged 18 years or older who had a cancer diagnosis, were interviewed.
      Difficulties in responding to 2 of the attributes were due to insufficient context in the vignette and unclear terminology. As a result, changes were made iteratively throughout the interview process; more context was provided in the background and terminology was refined to aid respondent comprehension. All cognitive interview participants reported that they did not consider other attributes to be important in their choices and that the presented attributes were sufficiently comprehensive. The final list of 8 attributes and their levels is presented in Table 1.
      Table 1Finalized attributes and levels.
      AttributeLevelsConcept
      Screening is
      • 1.
        Available if you ask
      • 2.
        Routinely offered to everyone
      Do you prefer to ask or be approached for screening?
      You will be screened by
      • 1.
        A social worker
      • 2.
        A psychologist
      • 3.
        A cancer nurse
      • 4.
        Your cancer doctor
      • 5.
        Your general practitioner
      Who would you prefer to be screened by?
      Screening method
      • 1.
        Online form
      • 2.
        Pen and paper form
      • 3.
        Face-to-face interview
      • 4.
        Phone call
      What screening method would you prefer?
      You will be screened
      • 1.
        Once only
      • 2.
        Monthly
      • 3.
        Every 3 months
      • 4.
        Once a year
      How often would you like to be screened?
      The screening process will take
      • 1.
        15 mins
      • 2.
        30 mins
      • 3.
        1 hour
      • 4.
        3 hours
      On top of your other medical appointments, you will spend an extra X for the screening process.
      Waiting time to be screened
      • 1.
        Immediately available
      • 2.
        Week
      • 3.
        Month
      Perceived urgency to be screened by a health professional
      Follow-up and care will be provided by
      • 1.
        A peer support group
      • 2.
        Your general practitioner
      • 3.
        A psychologist based in the community
      • 4.
        The psycho-oncology service within the cancer care service
      If screening indicates you need help with anxiety and/or depression, I would prefer to be referred to X.
      The maximum cost of follow-up and care to you is
      • 1.
        $0
      • 2.
        $50
      • 3.
        $150
      • 4.
        $300
      The maximum cost associated with follow-up and care

      Design and Simulation

      A main effects design was constructed using 2 generators and an orthogonal array with 32 options.
      • Street D.J.
      • Burgess L.
      The Construction of Optimal Stated Choice Experiments Theory and Methods.
      This gives rise to 64 choice sets. The design had a D-efficiency of 84.8% for an assumed prior with all entries equal to 0. A simulation study for various initial nonzero priors yielded both small standard errors and ranges for each estimated parameter, demonstrating the design was robust to different previous specifications.
      The 64 choice sets were grouped into 4 blocks of 16 choice sets, and respondents were randomly assigned to one of the blocks. Each choice set comprised a forced choice between 2 alternatives: screening program 1 and screening program 2. An opt-out option was provided after the forced choice between the 2 alternatives to identify those who would rather not participate in a screening program all together.
      The vignette and an example choice set are presented in Appendices 2 and 3 in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.05.014, respectively.
      A pilot study was undertaken with a general population sample (n = 77) to ensure that the experiment performed as intended. Statistical analysis was conducted using 2 models: (1) conditional logit model and (2) mixed logit (MIXL) model. Because results were consistent with expectations (ordered and in the expected direction), no changes were made to the DCE.

      Recruitment and Data Collection

      This study was administered online through the survey platform of Survey Engine, a survey provider company with experience in administration of DCEs. People with an experience of cancer were recruited using the same sample frames as for the cognitive interviews. In particular, participants were recruited from 2 sources (1) members of the BCNA surveys and questionnaires group and (2) participants in the SR online panel. Participants were eligible if they were aged 18 years older who resided in Australia and had a previous diagnosis of cancer. In the case of the BCNA sample, the participants were sent an invitation link by BCNA to complete the survey online. In the case of SR, participants who were eligible were sent a link by SR to complete the survey online.
      This study was approved by the University of Technology Sydney Human Research Ethics Committee (reference no. ETH18-2507) and the BCNA research committee (dated August 2, 2019).

      Analysis

      Descriptive statistics were used to describe the demographic characteristics of the sample. MIXL and latent class (lclogit) models were used to analyze the choice experiment responses. Willingness-to-accept measures were also calculated. The analysis was conducted using Stata version 15 (StataCorp LLC, College Station, TX).

      Mixed Logit

      The MIXL model relates the probability of choosing an alternative within a choice set to the attribute levels used to describe each option and also allows for the specification of one or more parameters as randomly distributed.
      • Hole A.R.
      Mixed logit modelling in Stata - An overview.
      In addition, the MIXL includes a random error term that adjusts for variations in an individual’s preferences.
      • Hauber A.B.
      • González J.M.
      • Groothuis-Oudshoorn C.G.M.
      • et al.
      Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
      The MIXL model is outlined in Eq. 1, where the utility for individual i associated with choice j in scenario s is as follows:
      Uijs=βXijs+(ηiXijs+eijs)
      (1)


      where βi is a vector of coefficients, Xijs is a vector of explanatory variables, and ηi is a vector of person specific deviations from the mean. The MIXL model estimates a set of mean preference weights and a set of standard deviations of effects across the sample. Interpretation of mean preference weights is made in relation to a base level. The standard deviations indicate variability in the mean preference weights; larger (smaller) values indicate greater (smaller) variability.
      • Hauber A.B.
      • González J.M.
      • Groothuis-Oudshoorn C.G.M.
      • et al.
      Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
      In both the MIXL and latent class (lclogit) models, all attributes except cost and waiting time were dummy coded (categorical), with one level (the base level) being omitted. Cost and waiting time were specified as continuous variables to facilitate estimation of willingness to accept and willingness to wait (WTW), as discussed later. Linear and higher order specifications of the cost and wait time variables were tested, and these were not statistically significant.

      Cost Equivalence Measures

      On the basis of the MIXL estimates, 2 cost equivalence measures using the wait time and cost attributes (and associated confidence intervals) were calculated using the “nlcom” command: (1) WTW using the waiting time attribute and (2) willingness to pay (WTP) using the cost attribute. WTW and WTP are each estimated as ratios—either as the ratio of the value of the coefficient of interest (x) to the negative of the cost or wait time attribute (y). Waiting time and cost were entered in the model as continuous, and the linear specification was tested. Interpretation of estimates for each attribute level is made in comparison with the base case. It is important to note that the cost attribute refers to the cost of treatment if required, and, therefore, it does not represent a WTP for screening as such, but a measure of the relative value of different screening programs. Caution should be applied in interpreting these results because they will be somewhat confounded by the respondent’s previous beliefs about the likelihood of needing follow-up treatment.

      lclogit Model

      In addition to the MIXL model, lclogit models can further explore preference heterogeneity. The model assumes that classes (groups) of respondents exist within the sample, and the preference weights within each class are identical but are also systematically different from preference weights in other classes.
      • Hauber A.B.
      • González J.M.
      • Groothuis-Oudshoorn C.G.M.
      • et al.
      Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
      Within each class, preference weights are estimated using a conditional logit model.
      • Hauber A.B.
      • González J.M.
      • Groothuis-Oudshoorn C.G.M.
      • et al.
      Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
      A user-written Stata module,
      • Pacifico D.
      • Yoo H.I.
      lclogit: a Stata command for fitting latent-class conditional logit models via the expectation-maximization algorithm.
      lclogit, was used to conduct this analysis. Demographics of each class were tabulated using by class: tabulate in Stata after model estimation.
      The optimal number of lclogit models were chosen by examining the goodness-of-fit statistics; Bayesian information criterion and the conditional Akaike information criterion.
      • Pacifico D.
      • Yoo H.I.
      lclogit: a Stata command for fitting latent-class conditional logit models via the expectation-maximization algorithm.

      Results

      Description of the Study Population

      A total of 660 participants started the survey. Of these, 2 were screened out owing to age and sex quotas already having been filled, 12 did not meet inclusion criteria, 6 duplicate entries, and 346 were timed out of the survey either by the respondent or the survey system after 15 minutes of inactivity. This resulted in a sample of 294 completed surveys: BCNA (n = 130) and SR (n = 164). Only respondents (n = 294) who completed all 16 choice sets were included in the analysis.
      Demographic characteristics of the sample are presented in Table 2. Compared with the general Australian population, our participants were generally older (as expected of a cancer population) and more educated. We also had a high proportion of females, which is to be expected with recruitment from BCNA. Most of the respondents received a diagnosis of breast cancer, and their first diagnosis of cancer was more than 5 years ago. Most respondents have normal levels of anxiety (57.1%) and depression (71.8%).
      Table 2Demographic characteristics.
      Frequency (n = 294)%
      Sex (M/F)69/22523.5/76.5
      Age (y)
       0-2400
       25-3420.5
       35-44123.5
       45-545318
       55-6411138
       65-749533
       75-84217
       85+00
      Education
       Primary school20.9
       High school or equivalent5719.3
       Diploma or equivalent8227.8
       Tertiary education15051
       Prefer not to say31
      Work Status
       Full time work5217.6
       Part-time work5619
       Retired12040.8
       Not working3612.5
       Other3010.2
      Household income (before tax and deductions per annum)
       <Median Australian weekly income12843.6
       ≥Median Australian weekly income10936.9
       Prefer not to say5719.4
      Type of cancer (first diagnosis)
       Skin cancer (include melanoma, BCC, and SCC)4816.3
       Colorectal82.7
       Breast15552.7
       Prostate289.5
       Lung cancer (include trachea, pleura and bronchus)51.7
       Cervical93.1
      Cancer of other female reproductive organs(include uterus and ovary)82.7
       Bladder/kidney51.7
       Stomach00
       Leukemia31
       Non-Hodgkin lymphoma31
       Other types of lymphoma51.7
       Cancer of unknown primary site05.8
       Other1716.3
      Years since diagnosis (first diagnosis)
       <2 y289.52
       2-5 y7124.15
       > 5 y19566.33
      Number of comorbidities (other than cancer)
       0196.5
       118562.9
       25819.7
       3-43210.9
       ≥500
      Clinical level of anxiety (HADS)
       Normal (0-7)16857.1
       Mild (8-10)6321.4
       Moderate (11-14)4515.3
       Severe (15-21)186.1
      Clinical level of depression (HADS)
       Normal (0-7)21171.8
       Mild (8-10)4916.7
       Moderate (11-14)268.8
       Severe (15-21)82.7
      BCC indicates basal cell carcinoma; HADS, Hospital Anxiety and Depression Scale; SCC, squamous cell carcinoma.
      Notably, 43% of the sample never opted out of their chosen screening program. Most respondents (63%) chose the opt-out in 4 or fewer of the 16 choice sets they saw in the DCE. Only 8 respondents (3%) always chose the opt-out.

      MIXL Model

      Table 3 presents the results of the MIXL. The results indicate a preference for a cancer nurse to conduct face-to-face screening at regular intervals (monthly and every 3 months) compared with a social worker. Postscreening care and follow-up by the psycho-oncology team is also most preferred compared with a peer support group. Low cost and shorter waiting time are also preferred. The magnitude of each standard deviation (for 11 of 19 attribute levels) is almost equivalent to its corresponding coefficient and is also statistically significant, indicating that variability in individual patient preferences exists.
      • Hauber A.B.
      • González J.M.
      • Groothuis-Oudshoorn C.G.M.
      • et al.
      Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
      All parameters were specified as random, except for cost and waiting time.
      Table 3Mixed logit results.
      AttributesLevelsCoefficientSD
      Screening is (Base = available if you ask)Routinely offered to everyone0.384
      P<.001.
      0.521
      P<.001.
      You will be screened by (Base = a social worker)A psychologist0.643
      P<.001.
      0.812
      P<.001.
      A cancer nurse0.761
      P<.001.
      0.646
      P<.001.
      Your cancer doctor0.621
      P<.001.
      0.437
      P<.01.
      Your general practitioner0.370
      P<.001.
      0.578
      P<.001.
      Screening method (Base = pen and paper questionnaire)Online questionnaire0.0280.052
      Face-to-face interview0.471
      P<.001.
      −0.300
      Phone call0.0000.783
      P<.001.
      You will be screened (Base = once only)Monthly0.778
      P<.001.
      0.885
      P<.001.
      Every 3 mo0.754
      P<.001.
      0.584
      P<.001.
      Once a year0.269
      P<.001.
      0.294
      The screening process will take (Base = 15 min)30 min0.021−0.004
      1 h−0.0040.009
      3 h−0.356
      P<.001.
      0.253
      Follow-up and care will be provided by (Base = a peer support group)Your general practitioner0.387
      P<.001.
      0.652
      P<.001.
      A psychologist based in the community0.650
      P<.001.
      −0.260
      The psycho-oncology service within the cancer care service1.044
      P<.001.
      0.610
      P<.001.
      Wait time continuous (fixed)-−0.017
      P<.001.
      -
      Cost continuous (fixed)-−0.006
      P<.001.
      -
      Observations9408
      LL−2738
      AIC5548
      BIC5806
      AIC indicates Akaike information criterion; BIC, Bayesian information criterion; LL, log likelihood.
      P<.01.
      P<.001.

      WTW

      The attribute for waiting time was used to estimate a WTW. Because this was the wait time to be screened and for follow-up and care, it can be more readily interpreted as a measure of strength of preference for the screening service. Figure 1 presents the WTW estimates for (1) health professionals involved in screening and (2) postscreening follow-up and care.
      Figure thumbnail gr1
      Figure 1WTW estimates graph.
      CI indicates confidence intervals; GP, general practitioner; WTW, willingness to wait.
      Respondents are most willing to wait to participate in a screening service if a cancer nurse conducts the screening (44.49 days) and least willing to wait for a general practitioner (21.62 days) compared with a social worker. After screening, respondents are willing to wait 61.08 days for a screening service that has a psycho-oncology team in the cancer service compared with a peer support group.

      WTP

      As a secondary measure of willingness to accept, the cost attribute was used to calculate a WTP measure. Because the cost attribute was related to cost of follow-up rather than screening, it should be interpreted as providing an indication of the value of different screening services rather than a WTP for screening directly. The results that are presented in Figure 2 show that people living with cancer place a high value on 3 aspects of a screening program: (1) the screening health professional, (2) screening regularity, and (3) care and follow-up postscreening.
      Figure thumbnail gr2
      Figure 2WTP estimates graph.
      CI indicates confidence intervals; GP, general practitioner; WTP, willingness to pay.
      The measure of WTP is highest ($156.19) for a screening program that involved screening by a nurse compared with a social worker (base level). WTP for a screening program that involves monthly screening was $159.77 and for 3 monthly screening was $154.77 compared with once-off screening. Follow-up and care by the psycho-oncology team were valued the most at $214.43 compared with a peer support group.

      lclogit Analysis

      Table 4 presents the results of the lclogit model.
      Table 4Latent class logit results.
      AttributesLevelsClass 1Class 2
      CoefficientSECoefficientSE
      Screening is (Base = available if you ask)Routinely offered to everyone0.316
      P<.001.
      (0.061)0.447
      P<.01.
      (0.141)
      You will be screened by (Base = a social worker)A psychologist0.458
      P<.001.
      (0.079)0.810
      P<.001.
      (0.205)
      A cancer nurse0.741
      P<.001.
      (0.083)−0.104(0.209)
      Your cancer doctor0.531
      P<.001.
      (0.081)0.752
      P<.001.
      (0.226)
      Your general practitioner0.388
      P<.001.
      (0.088)0.248(0.185)
      Screening method (Base = pen and paper questionnaire)Online questionnaire−0.010(0.074)0.013(0.169)
      Face-to-face interview0.422
      P<.001.
      (0.066)0.165(0.154)
      Phone call0.104(0.074)−0.298(0.160)
      You will be screened (Base = once only)Monthly1.014
      P<.001.
      (0.089)−1.385
      P<.001.
      (0.346)
      Every 3 mo0.813
      P<.001.
      (0.074)−0.482
      P<.05.
      (0.215)
      Once a year0.303
      P<.001.
      (0.077)−0.792
      P<.001.
      (0.221)
      The screening process will take (Base = 15 min)30 mins0.090(0.075)−0.447
      P<.05.
      (0.205)
      1 h0.126(0.066)−0.584
      P<.001.
      (0.164)
      3 h−0.191
      P<.05.
      (0.075)−1.078
      P<.001.
      (0.271)
      Waiting time to be screened (Base = immediately available)1 wk−0.128
      P<.05.
      (0.056)−0.038(0.119)
      1 mo−0.353
      P<.001.
      (0.061)−0.298
      P<.05.
      (0.135)
      Follow-up and care will be provided by (Base = a peer support group)Your general practitioner0.348
      P<.001.
      (0.076)0.519
      P<.01.
      (0.176)
      A psychologist based in the community0.504
      P<.001.
      (0.071)1.075
      P<.001.
      (0.215)
      The psycho-oncology service within the cancer care service0.771
      P<.001.
      (0.079)1.474
      P<.001.
      (0.261)
      Maximum cost of follow-up and care to you is (Base = $0)$50−0.071(0.097)−1.092
      P<.001.
      (0.229)
      $150−0.432
      P<.001.
      (0.115)−2.221
      P<.001.
      (0.357)
      $300−0.666
      P<.001.
      (0.114)−4.127
      P<.001.
      (0.482)
      Average class share73%27%
      Class membership model parameters
      Gender (female)0.3020
      Cancer (breast)−0.0170
      Panel (BCNA)0.4100
      Income (≥median Australian income)0.6750
      Income (prefer not to say)−0.3460
      Constant0.4620
      Observations9408
      LL−2734
      AIC5570.44
      BIC5754.619
      CAIC5804.619
      AIC indicates Akaike information criterion; BIC, Bayesian information criterion; CAIC, conditional Akaike information criterion; LL, log likelihood; SE, standard error.
      P<.05.
      P<.01.
      P<.001.
      On the basis of the Bayesian information criterion and conditional Akaike information criterion, a 2-class model was selected to explore class membership further. Four class membership variables were selected: gender (male vs female), type of cancer (other vs breast), panel (SR vs BCNA), and income (<median Australian weekly income vs ≥median Australian weekly income and prefer not to say). These were dummy coded and included in the lclogit model.
      With a 2-class model, 73% fall in class 1 and 27% fall in class 2. Compared with class 2, class 1 respondents have higher proportions of those who are women and tertiary educated. Class 1 respondents prefer to be screened by a cancer nurse, whereas those in class 2 prefer a psychologist. Class 1 respondents prefer to be rescreened at regular intervals and even up to 1 year, while class 2 respondents prefer once-off screening. Both classes prefer low cost; nevertheless, class 2 is more cost-sensitive than class 1. No class membership variables significantly predicted class membership, indicating that there may be other variables that are not captured in this study contributing to class membership.

      Discussion

      This study reports cancer population preferences for A&D screening programs in cancer care. Whether the aim is to maximize uptake or achieve patient-centered care, our findings provide important new information for a successful implementation of routine A&D screening in cancer services. These findings will also give decision makers the evidence to make informed resource allocation decisions that will not only maximize uptake but are patient-centered as well.
      Our results demonstrate that people with cancer on average prefer a screening program that is routinely offered and done face-to-face with a cancer nurse with follow-up care provided by the specialist psycho-oncology team within the cancer service. The lclogit analysis revealed 2 classes of cancer patients. Most participants (class 1, 73%) prefer screening to be done with a cancer nurse at regular intervals. Although a smaller proportion of participants (class 2, 27%), characterized by higher proportions of tertiary educated females, prefer once-off screening with a psychologist. Both classes prefer low cost. Although most participants fit within class 1, there are still more than a quarter that fit within class 2, demonstrating that there are distinct classes with different preferences. On a clinical level, this means that health services should implement a screening program that matches the preferences of class 1 participants, and if possible, flexibility in clinical service delivery should be made available to suit those in class 2. Overall, this reflects a preference for a cancer service that offers integrated mental healthcare, whereby both cancer and mental healthcare can be accessed within 1 single health service. Patients with cancer have a higher WTW and WTP when the screening program involves screening undertaken by an oncology health professional—cancer nurse (WTW = 44.92 days and WTP = $156.29). Respondents are willing to wait up to 2 months (61.08 days) and have a WTP value of $214.43 to participate in a screening program that provides follow-up care delivered by a psycho-oncology team within the cancer service.
      The preference for integrated mental healthcare may reflect an individual’s desire to simplify their patient experience. Care that is easier to access has been shown to positively influence a patient’s decision to undertake treatment.
      • Dwight-Johnson M.
      • Lagomasino I.T.
      • Aisenberg E.
      • Hay J.
      Using conjoint analysis to assess depression treatment preferences among low-income Latinos.
      ,
      • Foster L.
      • Harris L.
      • Black B.
      Preference for mode of delivery of cognitive behavior therapy in social anxiety.
      Existing literature also demonstrates that integrated care is more cost-effective in the cancer context.
      • Yoo S.J.C.
      • Nyman J.A.
      • Cheville A.L.
      • Kroenke K.
      Cost effectiveness of telecare management for pain and depression in patients with cancer: results from a randomized trial.
      ,
      • Walker J.
      • Hansen C.H.
      • Martin P.
      • et al.
      Integrated collaborative care for major depression comorbid with a poor prognosis cancer (SMaRT Oncology-3): a multicentre randomised controlled trial in patients with lung cancer.
      Although our findings explore the preferences of people with cancer for A&D screening programs, findings are consistent with the preferences of the general population, for the treatment of A&D. The existing literature relating to A&D treatment interventions reports that people prefer interventions that are conducted face-to-face,
      • Lokkerbol J.
      • van Voorthuijsen J.M.
      • Geomini A.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with anxiety disorder.
      ,
      • Lokkerbol J.
      • Geomini A.
      • van Voorthuijsen J.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with depression.
      ,
      • Lau M.A.
      • Colley L.
      • Willett B.R.
      • Lynd L.D.
      Employee’s preferences for access to mindfulness-based cognitive therapy to reduce the risk of depressive relapse—A discrete choice experiment.
      with a nurse,
      • Groenewoud S.
      • Van Exel N.J.A.
      • Bobinac A.
      • Berg M.
      • Huijsman R.
      • Stolk E.A.
      What influences patients' decisions when choosing a health care provider? Measuring preferences of patients with knee arthrosis, chronic depression, or Alzheimer's disease, using discrete choice experiments.
      that are low cost,
      • Foster L.
      • Harris L.
      • Black B.
      Preference for mode of delivery of cognitive behavior therapy in social anxiety.
      ,
      • Ride J.
      • Lancsar E.
      Women's preferences for treatment of perinatal depression and anxiety: a discrete choice experiment.
      and where waiting time is minimal.
      • Lokkerbol J.
      • van Voorthuijsen J.M.
      • Geomini A.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with anxiety disorder.
      ,
      • Lokkerbol J.
      • Geomini A.
      • van Voorthuijsen J.
      • et al.
      A discrete-choice experiment to assess treatment modality preferences of patients with depression.
      To date, this is the first DCE exploring patient preferences toward A&D screening programs in cancer care. One of the strengths of this study is the extensive identification and development process conducted to ensure attribute and level content validity and its relevance to policy and the target population. Moreover, a pilot study and analysis were conducted to ensure the survey instrument performed as designed before launching with the cancer population. Upon completion of the DCE, cognitive burden was assessed by asking participants to rate from strongly disagree to strongly agree the following statements: (1) “I considered the whole description whilst completing the task” and (2) “I found it easy to imagine the scenarios.” For both statements, most respondents answered agree or strongly agree: 83.17% for statement (1) and 72.49% for statement (2). The results indicate that social workers are the least preferred health professional to conduct screening. Interestingly, only one participant in sample was being cared for by a social worker at the time of this study. This marked reluctance to be cared for by a social worker may not be an actual preference, but rather an evidence of status quo bias (endowment effect),
      • Dorsey D.
      Preferences, welfare, and the Status-Quo Bias.
      ,
      • Salkeld G.
      • Ryan M.
      • Short L.
      The veil of experience: do consumers prefer what they know best?.
      in that respondents prefer what they had previously experienced over a new alternative.
      Patient preferences for A&D screening in cancer care are complex and could include a long list of attributes and levels. The attributes and levels included were considered most important to people living with a diagnosis of cancer and was validated through cognitive interviews. Nevertheless, there are other attributes that were identified in the literature search but were not included, such as travel time, and other convenience factors and intervention effectiveness.
      Recruitment for nonbreast cancer participants proved difficult through a panel provider, and future studies could explore other recruitment sources such as a cancer service. Future studies should be aware that respondents recruited from cancer support and advocacy groups could potentially bias the recruited sample; in the case of BCNA, 66% of BCNA respondents have received tertiary education versus 22% in the general population.
      Furthermore, this study was conducted with an adult Australian cancer population. Therefore, without further validation, our results may not be generalizable to cancer populations in a different country owing to differences in how health systems are organized and operated.

      Conclusion

      Patients with cancer prefer a cancer service that offers integrated mental healthcare services. To maximize patient uptake, cancer services should implement an A&D screening program that is routinely offered and delivered by oncology healthcare staff in a face-to-face format, with postscreening follow-up care provided by a psycho-oncology team that is embedded within the cancer service. Our findings contribute to an area of limited evidence and can be used to design evidence-based and patient-centered screening services for patients with cancer. Health services are resource constrained, and implementing or tailoring services that match patient preferences can be burdensome. Nevertheless, by doing so, there is potential to improve clinical service delivery and overall health service efficiency.

      Article and Author Information

      Author Contributions: Concept and design: Yim, Arora, Shaw, Street, Pearce, Viney
      Acquisition of data: Yim, Viney
      Analysis and interpretation of data: Yim, Arora, Shaw, Street, Viney
      Drafting of the manuscript: Yim, Arora, Shaw, Viney
      Critical revision of the paper for important intellectual content: Yim, Arora, Shaw, Street, Pearce, Viney
      Statistical analysis: Yim, Arora, Shaw, Street
      Obtaining funding: Yim
      Administrative, technical, or logistic support: Yim
      Supervision: Arora, Shaw, Pearce, Viney
      Conflict of Interest Disclosures: The authors reported no conflicts of interest.
      Funding/Support: This research is funded by a scholarship from the NSW Health PhD Scholarship program, and additional top-up funding from the ADAPT program (based at the University of Sydney). Funding for the Anxiety and Depression Pathway (ADAPT) Program was provided by a Cancer Institute NSW Translational Program Grant (ID 14/TPG/1-02). Data collection was supported by an NSW Health PhD Accelerator grant.
      Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

      Acknowledgment

      We acknowledge the participants who completed the questionnaire.

      Supplemental Material

      References

        • Linden W.
        • Vodermaier A.
        • Mackenzie R.
        • Greig D.
        Anxiety and depression after cancer diagnosis: prevalence rates by cancer type, gender, and age.
        J Affect Disord. 2012; 141: 343-351
        • Zabora J.
        • BrintzenhofeSzoc K.
        • Curbow B.
        • Hooker C.
        • Piantadosi S.
        The prevalence of psychological distress by cancer site.
        Psychooncology. 2001; 10: 19-28
        • Annunziata M.A.
        • Muzzatti B.
        • Giovannini L.
        • et al.
        Is long-term cancer survivors’ quality of life comparable to that of the general population? An Italian study.
        Support Care Cancer. 2015; 23: 2663-2668
        • Chan C.M.
        • Wan Ahmad W.A.
        • Yusof M.M.
        • Ho G.F.
        • Krupat E.
        Effects of depression and anxiety on mortality in a mixed cancer group: a longitudinal approach using standardised diagnostic interviews.
        Psychooncology. 2015; 24: 718-725
        • Buzgova R.
        • Jarosova D.
        • Hajnova E.
        Assessing anxiety and depression with respect to the quality of life in cancer inpatients receiving palliative care.
        Eur J Oncol Nurs. 2015; 19: 667-672
        • Fallowfield L.
        • Ratcliffe D.
        • Jenkins V.
        • Saul J.
        Psychiatric morbidity and its recognition by doctors in patients with cancer.
        Br J Cancer. 2001; 84: 1011-1015
        • Rob S.-F.
        • Afaf G.
        • Allison B.
        • Billie B.
        • Louise B.
        • Peter C.
        The unmet supportive care needs of patients with cancer.
        Cancer. 2000; 88: 226-237
        • Osborn R.L.
        • Demoncada A.C.
        • Feuerstein M.
        Psychosocial interventions for depression, anxiety, and quality of life in cancer survivors: meta-analyses.
        Int J Psychiatry Med. 2006; 36: 13-34
        • Chien C.-H.
        • Liu K.-L.
        • Chien H.-T.
        • Liu H.-E.
        The effects of psychosocial strategies on anxiety and depression of patients diagnosed with prostate cancer: a systematic review.
        Int J Nurs Stud. 2014; 51: 28-38
        • Howell D.
        • Hack T.F.
        • et al.
        • Cancer Journey Survivorship Expert Panel
        Survivorship services for adult cancer populations: a pan-Canadian guideline.
        Curr Oncol. 2011; 18: e265-e281
        • Howell D.
        • Keshavarz H.
        • Esplen M.
        • et al.
        A Pan-Canadian Practice Guideline: Screening, Assessment and Management of Psychosocial Distress, Depression and Anxiety in Adults with Cancer.
        Canadian Association of Psychosocial Oncology and Canadian Partnership Against Cancer, Toronto, ON2015
        • Ellis P.
        Australian and New Zealand clinical practice guidelines for the treatment of depression.
        Aust N Z J Psychiatry. 2004; 38: 389-407
        • Holland J.C.
        • Andersen B.
        • Breitbart W.S.
        • et al.
        Distress management: clinical practice guidelines in Oncology.
        J Natl Compr Canc Netw. 2007; 5: 66-98
        • Andersen B.L.
        • DeRubeis R.J.
        • Berman B.S.
        • et al.
        Screening, assessment, and care of anxiety and depressive symptoms in adults with cancer: an American Society of Clinical Oncology guideline adaptation.
        J Clin Oncol. 2014; 32: 1605-1619
        • PoCoG
        Clinical Pathway for the Screening, Assessment and Management of Anxiety and Depression in Adult Cancer Patients.
        Psycho-oncology Co-operative Research Group, Sydney, Australia2017
        • Butow P.
        • Shaw J.
        • Shepherd H.L.
        • et al.
        Comparison of implementation strategies to influence adherence to the clinical pathway for screening, assessment and management of anxiety and depression in adult cancer patients (ADAPT CP): study protocol of a cluster randomised controlled trial.
        BMC Cancer. 2018; 18: 1077
        • Carlson L.E.
        • Waller A.
        • Mitchell A.J.
        Screening for distress and unmet needs in patients with cancer: review and recommendations.
        J Clin Oncol. 2012; 30: 1160-1177
        • Dwight-Johnson M.
        • Lagomasino I.T.
        • Aisenberg E.
        • Hay J.
        Using conjoint analysis to assess depression treatment preferences among low-income Latinos.
        Psychiatr Serv. 2004; 55: 934-936
        • Lokkerbol J.
        • van Voorthuijsen J.M.
        • Geomini A.
        • et al.
        A discrete-choice experiment to assess treatment modality preferences of patients with anxiety disorder.
        J Med Econ. 2018; : 1-24
        • Lokkerbol J.
        • Geomini A.
        • van Voorthuijsen J.
        • et al.
        A discrete-choice experiment to assess treatment modality preferences of patients with depression.
        J Med Econ. 2018; : 1-23
        • Hobden B.
        • Turon H.
        • Bryant J.
        • Wall L.
        • Brown S.
        • Sanson-Fisher R.
        Oncology patient preferences for depression care: a discrete choice experiment.
        Psychooncology. 2019; 28: 807-814
        • Viney R.
        • Lancsar E.
        • Louviere J.
        Discrete choice experiments to measure consumer preferences for health and healthcare.
        Expert Rev Pharmacoecon Outcomes Res. 2002; 2: 319-326
        • Cheraghi-Sohi S.
        • Bower P.
        • Mead N.
        • McDonald R.
        • Whalley D.
        • Roland M.
        Making sense of patient priorities: applying discrete choice methods in primary care using ‘think aloud’ technique.
        Fam Pract. 2007; 24: 276-282
        • Gomes B.
        • de Brito M.
        • Sarmento V.P.
        • et al.
        Valuing attributes of home palliative care with service users: a pilot discrete choice experiment.
        J Pain Symptom Manage. 2017; 54: 973-985
        • Kohler R.E.
        • Lee C.N.
        • Gopal S.
        • Reeve B.B.
        • Weiner B.J.
        • Wheeler S.B.
        Developing a discrete choice experiment in Malawi: eliciting preferences for breast cancer early detection services.
        Patient Prefer Adherence. 2015; 9: 1459-1472
        • Dwight Johnson M.
        • Apesoa-Varano C.
        • Hay J.
        • Unutzer J.
        • Hinton L.
        Depression treatment preferences of older white and Mexican origin men.
        Gen Hosp Psychiatry. 2013; 35: 59-65
        • Katz D.A.
        • Stewart K.R.
        • Paez M.
        • et al.
        Development of a discrete choice experiment (DCE) questionnaire to understand veterans’ preferences for tobacco treatment in primary care.
        Patient. 2018; 11: 649-663
        • Schildmann K.E.
        • Groeneveld I.E.
        • Denzel J.
        • et al.
        Discovering the hidden benefits of cognitive interviewing in two languages: the first phase of a validation study of the integrated palliative care outcome scale.
        Palliat Med. 2015; 30: 599-610
        • Tourangeau R.
        • Rips L.J.
        • Rasinski K.A.
        The Psychology of Survey Response.
        Cambridge University Press, Cambridge2000
        • Street D.J.
        • Burgess L.
        The Construction of Optimal Stated Choice Experiments Theory and Methods.
        Wiley, Sydney, Australia2007
        • GmBH. S
        Survey Engine.
        • Hole A.R.
        Mixed logit modelling in Stata - An overview.
        • Hauber A.B.
        • González J.M.
        • Groothuis-Oudshoorn C.G.M.
        • et al.
        Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force.
        Value Health. 2016; 19: 300-315
        • Pacifico D.
        • Yoo H.I.
        lclogit: a Stata command for fitting latent-class conditional logit models via the expectation-maximization algorithm.
        Stata J. 2013; 13: 625-639
        • Foster L.
        • Harris L.
        • Black B.
        Preference for mode of delivery of cognitive behavior therapy in social anxiety.
        J CyberTher Rehab. 2011; 4: 351-361
        • Yoo S.J.C.
        • Nyman J.A.
        • Cheville A.L.
        • Kroenke K.
        Cost effectiveness of telecare management for pain and depression in patients with cancer: results from a randomized trial.
        Gen Hosp Psychiatry. 2014; 36: 599-606
        • Walker J.
        • Hansen C.H.
        • Martin P.
        • et al.
        Integrated collaborative care for major depression comorbid with a poor prognosis cancer (SMaRT Oncology-3): a multicentre randomised controlled trial in patients with lung cancer.
        Lancet Oncol. 2014; 15: 1168-1176
        • Lau M.A.
        • Colley L.
        • Willett B.R.
        • Lynd L.D.
        Employee’s preferences for access to mindfulness-based cognitive therapy to reduce the risk of depressive relapse—A discrete choice experiment.
        Mindfulness. 2012; 3: 318-326
        • Groenewoud S.
        • Van Exel N.J.A.
        • Bobinac A.
        • Berg M.
        • Huijsman R.
        • Stolk E.A.
        What influences patients' decisions when choosing a health care provider? Measuring preferences of patients with knee arthrosis, chronic depression, or Alzheimer's disease, using discrete choice experiments.
        Health Serv Res. 2015; 50: 1941-1972
        • Ride J.
        • Lancsar E.
        Women's preferences for treatment of perinatal depression and anxiety: a discrete choice experiment.
        PLoS One. 2016; 11e0156629
        • Dorsey D.
        Preferences, welfare, and the Status-Quo Bias.
        Australasian Journal of Philosophy. 2010; 88: 535-554
        • Salkeld G.
        • Ryan M.
        • Short L.
        The veil of experience: do consumers prefer what they know best?.
        Health Econ. 2000; 9: 267-270