Advertisement

Heterogeneous Preferences for Colorectal Cancer Screening in Germany: Results of a Discrete Choice Experiment

Open AccessPublished:August 28, 2022DOI:https://doi.org/10.1016/j.jval.2022.07.012

      Highlights

      • This discrete choice experiment on characteristics of colorectal cancer screening tests revealed significant heterogeneous preferences among German insurees. Three groups were identified: the largest group valued—unexpectedly—a process parameter the most, that is, comprehensive bowel preparation, whereas the other 2 groups focused on potential benefits, that is, either reduction in cancer-specific mortality or incidence.
      • The experiment included a within-set dominated pair for validity testing. A total of 42% chose the dominated alternative (type B respondents) and revealed significantly different preferences than type A respondents. The underlying reasons and individuals’ considerations remain unknown. Internal validity issues of choice data warrant adequate attention in future discrete choice experiments.
      • Observed preference heterogeneity suggests different informational needs that are to be addressed by public health and healthcare services to support everyone regardless of their health literacy level in making individual informed decisions in the context of colorectal cancer screening by weighing tests’ benefits, harms, and processes.

      Abstract

      Objectives

      Colorectal cancer (CRC) screening tests differ in benefits, harms, and processes, making individual informed decisions preference based. The objective was to analyze the preferences of insurees in Germany for characteristics of CRC screening modalities.

      Methods

      A generic discrete choice experiment with 2-alternative choice sets and 6 attributes (CRC mortality, CRC incidence, complications, preparation, need for transportation, and follow-up; 3 levels each) depicting characteristics of fecal testing, sigmoidoscopy, and colonoscopy was generated. Participants completed 8 choice tasks. Internal validity was tested using a within-set dominated pair. Between June and October 2020, written questionnaires were sent to a stratified random sample (n = 5000) of 50-, 55-, and 60-year-old insurees of the AOK (Allgemeine Ortskrankenkasse) Lower Saxony, who had previously received an invitation to participate in the organized screening program including evidence-based information. Preferences were analyzed using conditional logit, mixed logit, and latent-class model.

      Results

      From 1282 questionnaires received (26% [1282 of 4945]), 1142 were included in the analysis. Approximately 42% of the respondents chose the dominated alternative in the internal validity test. Three heterogeneous preference classes were identified. Most important attributes were preparation (class 1; n = 505, 44%), CRC mortality (class 2; n = 347, 30%), and CRC incidence (class 3; n = 290, 25%). Contrary to a priori expectations, a higher effort was preferred for bowel cleansing (class 1) and accompaniment home (classes 1 and 2).

      Conclusion

      Internal validity issues of choice data need further research and warrant attention in future discrete choice experiment surveys. The observed preference heterogeneity suggests different informational needs, although the underlying reasons remained unclear.

      Keywords

      Introduction

      Globally, colorectal cancer (CRC) ranks third and second in the number of newly diagnosed cancer cases and cancer-related deaths.
      • Ferlay J.
      • Ervik M.
      • Lam F.
      • et al.
      Global cancer observatory: cancer today. IARC.
      Guideline recommendations for screening differ among countries. Fecal occult blood testing, guaiac-based fecal occult blood test or fecal immunochemical test, sigmoidoscopy, and colonoscopy are the modalities most commonly used.
      • Helsingen L.M.
      • Vandvik P.O.
      • Jodal H.C.
      • et al.
      Colorectal cancer screening with faecal immunochemical testing, sigmoidoscopy or colonoscopy: a clinical practice guideline.
      • Bénard F.
      • Barkun A.N.
      • Martel M.
      • von Renteln D.
      Systematic review of colorectal cancer screening guidelines for average-risk adults: summarizing the current global recommendations.
      • Navarro M.
      • Nicolas A.
      • Ferrandez A.
      • Lanas A.
      Colorectal cancer population screening programs worldwide in 2016: an update.
      In Germany, the general population aged 50 years and older can choose between fecal immunochemical test and colonoscopy.
      Directive on organized cancer screening programs. Federal Joint Committee.
      Sigmoidoscopy is recommended for individuals who reject colonoscopy, but is not part of the organized screening program and is therefore not covered by statutory health insurance.
      Directive on organized cancer screening programs. Federal Joint Committee.
      ,
      S3-guideline colorectal cancer, long version 2.1, 2019. German Cancer society, German Cancer Aid, AWMF.
      Screening modalities differ in terms of their effectiveness, risk of harms, and procedural issues. Although stool-based tests primarily aim at detecting early-stage cancer to reduce cancer-specific mortality, evidence has consistently shown that endoscopic screening options can also reduce incidence by removing precancerous lesions.
      • Bretthauer M.
      • Kalager M.
      Principles, effectiveness and caveats in screening for cancer.
      • Brenner H.
      • Stock C.
      • Hoffmeister M.
      Effect of screening sigmoidoscopy and screening colonoscopy on colorectal cancer incidence and mortality: systematic review and meta-analysis of randomised controlled trials and observational studies.
      • Jodal H.C.
      • Helsingen L.M.
      • Anderson J.C.
      • Lytvyn L.
      • Vandvik P.O.
      • Emilsson L.
      Colorectal cancer screening with faecal testing, sigmoidoscopy or colonoscopy: a systematic review and network meta-analysis.
      • Lin J.S.
      • Perdue L.A.
      • Henrikson N.B.
      • Bean S.I.
      • Blasi P.R.
      Screening for colorectal cancer: updated evidence report and systematic review for the US Preventive Services Task Force [published correction appears in JAMA. 2021;326(3):279].
      Serious adverse events caused by the procedure itself can occur with both sigmoidoscopy and colonoscopy: an estimated 0.2 perforations (95% confidence interval [CI] 0.1-0.4) and 0.5 major bleeds (95% CI 0-1.3) per 10 000 sigmoidoscopies and 3.1 perforations (95% CI 2.3-4.0) and 14.6 major bleeds (95% CI 9.4-19.9) per 10 000 colonoscopies.
      • Lin J.S.
      • Perdue L.A.
      • Henrikson N.B.
      • Bean S.I.
      • Blasi P.R.
      Screening for colorectal cancer: updated evidence report and systematic review for the US Preventive Services Task Force [published correction appears in JAMA. 2021;326(3):279].
      Although fecal testing itself does not cause any side effects (except psychological distress from false-positive results
      • Eckstrom E.
      • Feeny D.H.
      • Walter L.C.
      • Perdue L.A.
      • Whitlock E.P.
      Individualizing cancer screening in older adults: a narrative review and framework for future research.
      ), harms can occur during colonoscopy after an abnormal stool test result.
      • Jodal H.C.
      • Helsingen L.M.
      • Anderson J.C.
      • Lytvyn L.
      • Vandvik P.O.
      • Emilsson L.
      Colorectal cancer screening with faecal testing, sigmoidoscopy or colonoscopy: a systematic review and network meta-analysis.
      ,
      • Lin J.S.
      • Perdue L.A.
      • Henrikson N.B.
      • Bean S.I.
      • Blasi P.R.
      Screening for colorectal cancer: updated evidence report and systematic review for the US Preventive Services Task Force [published correction appears in JAMA. 2021;326(3):279].
      Differences in practical issues relate, for example, to test frequency, the procedure itself, or its impact on work ability due to bowel cleansing and sedation.
      • Helsingen L.M.
      • Vandvik P.O.
      • Jodal H.C.
      • et al.
      Colorectal cancer screening with faecal immunochemical testing, sigmoidoscopy or colonoscopy: a clinical practice guideline.
      Recognizing that screening, which is associated with benefits but also with harms, is offered to nonsymptomatic citizens, the Council of the European Union
      Council recommendation of 2 December 2003 on cancer screening (2003/878/EC). The Council of the European Union.
      recommended in 2003 that participation should be based on individual informed decision making. In situations where no procedure is clearly superior to the others, such decisions can be preference based.
      Patient preference information – voluntary submission, review in premarket approval applications, humanitarian device exemption applications, and de novo requests, and inclusion in decision summaries and device labeling. Guidance for industry, Food and Drug Administration staff, and other stakeholders. US Food and Drug Administration.
      The SIGMO (Sigmoidoscopy as an evidence-based CRC screening test—a possible option?) study
      • Brinkmann M.
      • Diedrich L.
      • Krauth C.
      • Robra B.-P.
      • Stahmeyer J.T.
      • Dreier M.
      General populations’ preferences for colorectal cancer screening: rationale and protocol for the discrete choice experiment in the SIGMO study.
      evaluates whether sigmoidoscopy should be offered as an additional screening alternative in Germany. To assess and consider the acceptability of different available modalities among potential beneficiaries in the context of healthcare regulatory decisions, the US Food and Drug Administration
      Patient preference information – voluntary submission, review in premarket approval applications, humanitarian device exemption applications, and de novo requests, and inclusion in decision summaries and device labeling. Guidance for industry, Food and Drug Administration staff, and other stakeholders. US Food and Drug Administration.
      and the German independent Institute for Quality and Efficiency in Health Care
      General methods, version 5.0. Institute for Quality and Efficiency in Health Care.
      recommend, among others, stated preference methods such as choice-based conjoint analysis, also known as discrete choice experiment (DCE). Therefore, the objective of this study was to analyze the preferences of insurees in Germany for CRC screening using a DCE.

      Methods

      The reporting of this DCE was based on the checklist for conjoint analysis applications in health developed by the ISPOR Good Research Practices for Conjoint Analysis Task Force.
      • Bridges J.F.P.
      • Hauber A.B.
      • Marshall D.
      • et al.
      Conjoint analysis applications in health-a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force.
      Ethical approval was obtained from the Ethics Committee of Hannover Medical School (reference number 8671_BO_K_2019). The SIGMO study is registered at the German Clinical Trials Register (DRKS00019010); a study protocol was published.
      • Brinkmann M.
      • Diedrich L.
      • Krauth C.
      • Robra B.-P.
      • Stahmeyer J.T.
      • Dreier M.
      General populations’ preferences for colorectal cancer screening: rationale and protocol for the discrete choice experiment in the SIGMO study.

      Identification and Selection of Attributes and Levels

      The identification and selection of attributes were done in a stepwise process.
      • Brinkmann M.
      • Diedrich L.
      • Krauth C.
      • Robra B.-P.
      • Stahmeyer J.T.
      • Dreier M.
      General populations’ preferences for colorectal cancer screening: rationale and protocol for the discrete choice experiment in the SIGMO study.
      First, an initial overview of possible attributes and their relative importance for decision making was obtained through a systematic review of studies eliciting general populations’ stated preferences for CRC screening tests (Appendix 1 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012). Second, the relevance of the attributes found and other possible characteristics were discussed in 4 focus groups with German-speaking 50- to 60-year-olds in November 2019 (Appendix 2 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).
      • Brinkmann M.
      • Von Holt I.
      • Diedrich L.
      • Krauth C.
      • Seidel G.
      • Dreier M.
      Attributes Characterizing Colorectal Cancer Screening Tests That Influence Preferences of Individuals Eligible for Screening in Germany: A Qualitative Study.
      Third, the multidisciplinary research team agreed on the final list of 6 attributes (Table 1) including benefits, harms, and procedural issues to enable informed decision making, as well as the attribute labels.
      Table 1Overview of attributes, attribute levels, and descriptions used in the DCE on CRC screening preferences in Germany.
      AttributesAttribute levelsDescriptions
      Develop bowel cancer3 of 10008 of 100010 of 1000Some procedures can even prevent bowel cancer. As a result, fewer people develop bowel cancer.

      There are figures for this: How many people in 1000 are expected to develop bowel cancer in the next 10 years?
      • Without participation, 10 people will develop bowel cancer; depending on the procedure, 3 to 8 of 1000 people will develop the disease.
      Die from bowel cancer1 of 10002 of 10003 of 1000Bowel cancer can be detected at an early stage and then treated. As a result, fewer people die of bowel cancer.

      There are figures for this: How many people in 1000 are expected to die of bowel cancer in the next 10 years?
      • Without participation, 3 people die; depending on the procedure, 1 to 2 of 1000 people die.
      Complications0 of 10001 of 10002 of 1000With some procedures, complications that often require hospitalization can occur: bleeding, perforation, or cardiovascular events.

      There are figures for this: How many people in 1000 have complications?
      • Depending on the procedure, 0 to 2 of 1000 people.
      PreparationNoneEnema before examinationDrinking 2 to 4 liters, no foodSome procedures require preparation to cleanse the bowel.

      There are 3 different options:
      • You have to drink 2 to 4 liters of a laxative the day before the examination and avoid solid food.
      • You will have an enema just before the examination.
      • No preparation is required.
      Need for transportationNeverOccasionallyAlmost alwaysFor some procedures, a sedative may be given. In this case, you will need an adult to accompany you home.
      Follow-up0 of 1000340 of 1000440 of 1000If the result is positive, a follow-up is required. For this, the bowel must be clean: you have to drink 2 to 4 liters of a laxative the day before the examination and avoid solid food. Rarely, complications can occur during the examination. A sedative is almost always given, and you will need an adult to accompany you home.

      There are figures for this: in how many of 1000 people is a follow-up required?
      • Depending on the procedure, between 0 and 440 of 1000 people.
      CRC indicates colorectal cancer; DCE, discrete choice experiment.
      Cancer-specific incidence and mortality represent important benefit outcomes
      • Rimer B.K.
      • Briss P.A.
      • Zeller P.K.
      • Chan E.C.Y.
      • Woolf S.H.
      Informed decision making: what is its role in cancer screening?.
      ; the ability of endoscopic screening methods to prevent cancer by removing precancerous lesions is a key difference compared with stool-based tests.
      • Bretthauer M.
      • Kalager M.
      Principles, effectiveness and caveats in screening for cancer.
      The levels used to define both attributes were based on a scientific report on benefits and harms of CRC screening tests by the German independent Institute for Quality and Efficiency in Health Care,
      Invitation and decision aid for bowel cancer screening. Institute for Quality and Efficiency in Health Care.
      which formed the evidence basis for the development of the invitation and decision aid sent out as part of the organized screening program for CRC in Germany. Complications requiring hospitalization, such as bleeding or perforation, are potential harms to the screening participant caused by the procedure itself.
      • Eckstrom E.
      • Feeny D.H.
      • Walter L.C.
      • Perdue L.A.
      • Whitlock E.P.
      Individualizing cancer screening in older adults: a narrative review and framework for future research.
      ,
      • Rimer B.K.
      • Briss P.A.
      • Zeller P.K.
      • Chan E.C.Y.
      • Woolf S.H.
      Informed decision making: what is its role in cancer screening?.
      The levels were calculated based on results of a network meta-analysis evaluating the effectiveness, harms, and burdens of CRC screening tests.
      • Jodal H.C.
      • Helsingen L.M.
      • Anderson J.C.
      • Lytvyn L.
      • Vandvik P.O.
      • Emilsson L.
      Colorectal cancer screening with faecal testing, sigmoidoscopy or colonoscopy: a systematic review and network meta-analysis.
      Preparation and the need for transportation as practical issues having impact on everyday life are attributes in which colonoscopy, sigmoidoscopy, and stool-based tests differ significantly. The qualitative levels describing the need for transportation were derived from results of an evaluation of the English Bowel Cancer Screening Programme
      • Ball A.J.
      • Rees C.J.
      • Corfe B.M.
      • Riley S.A.
      Sedation practice and comfort during colonoscopy: lessons learnt from a national screening programme.
      and routine flexible sigmoidoscopy practice in Scotland
      • Maliampurakal A.J.
      • McMillan D.C.
      • Anderson J.H.
      • Horgan P.G.
      • Mansouri D.
      Factors associated with the efficacy of polyp detection during routine flexible sigmoidoscopy.
      regarding the use of sedation. Finally, the need for follow-up in case of an initial positive screening test result described the additional perceived burden. Its levels represent the cumulative need for one follow-up within the next 10 years assuming a screening attendance of 100% calculated based on test positivity rates of a single round of stool-based tests and sigmoidoscopy.
      • Lin J.S.
      • Piper M.A.
      • Perdue L.A.
      • et al.
      Screening for colorectal cancer: an updated systematic review for the U.S. Preventive Services Task Force: evidence syntheses no. 135. Kaiser Permanente Center for Health Research.

      Choice Tasks and Experimental Design

      Each choice set contrasted 2 generic alternatives (Fig. 1), requesting a forced choice without an opt-out or status quo alternative. The experimental design was created using SAS (SAS Institute Inc, Cary, USA; version 9.4).
      • Kuhfeld W.F.
      Marketing research methods in SAS: experimental design, choice, conjoint, and graphical techniques. SAS Institute Inc.
      Six attributes with 3 levels each allow 729 (36) possible level combinations and (729 × 728) / 2 = 265 356 choice tasks.
      • Lancsar E.
      • Louviere J.
      Conducting discrete choice experiments to inform healthcare decision making: a user’s guide.
      ,
      • Ryan M.
      • Gerard K.
      • Amaya-Amaya M.
      Using Discrete Choice Experiments to Value Health and Health Care.
      To increase practicability and reduce cognitive complexity, a fractional factorial, D-efficient design (relative D-efficiency 74.6764; D-error 0.0418) with zero priors and 32 choice sets was generated and blocked into 4 versions of 8 choice tasks each. To control for potential bias due to attribute order, 2 different sequences were created for each version. The resulting 8 versions were randomly assigned to participants.
      Figure thumbnail gr1
      Figure 1Example choice task of the DCE on CRC screening preferences in Germany as presented to participants.
      CRC indicates colorectal cancer; DCE, discrete choice experiment.

      Questionnaire Design

      Data were collected using a written, self-administered postal questionnaire (English version in Appendix 3 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012, German version from authors on request) that consisted of 4 parts: (1) previous experience with CRC screening tests, (2) the DCE, (3) intention to participate in CRC screening tests, and (4) sociodemographic characteristics and health-related information.
      • Chew L.D.
      • Griffin J.M.
      • Partin M.R.
      • et al.
      Validation of screening questions for limited health literacy in a large VA outpatient population.
      ,
      • Brauns H.
      • Scherer S.
      • Steinmann S.
      The CASMIN educational classification in international comparative research.
      To ensure equal understanding among participants and to prevent consideration of other unobservable aspects in decision making, the attributes and levels included were described in detail in the introductory section of the DCE (Table 1) (Appendix 3 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012). For each choice task, participants were then asked which of the 2 devised procedures they would choose (Fig. 1). To check responses on rationality and consistency, a within-set dominated pair
      • Johnson F.R.
      • Yang J.-C.
      • Reed S.D.
      The internal validity of discrete choice experiment data: a testing tool for quantitative assessments.
      was included as first choice set (Fig. 1), so that each participant had to answer 9 choice tasks in total.
      Understanding and the amount of complexity of the DCE were piloted in 10 cognitive interviews in February 2020 (Appendix 4 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).

      Sampling and Data Collection

      The study was conducted among a stratified random sample of 50-, 55-, and 60-year-old insurees of the statutory health insurance company AOK (Allgemeine Ortskrankenkasse) Lower Saxony. Two to 3 weeks in advance to the study questionnaire, they had received a written invitation to participate in the national CRC screening program including comprehensive evidence-based information on the available tests, their benefits, and risks from their health insurance company.
      Directive on organized cancer screening programs. Federal Joint Committee.
      Excluded were insurees (1) with CRC or a chronic inflammatory bowel disease who do not belong to the asymptomatic target population and whose preferences may differ from those of the general population due to regular surveillance colonoscopies, (2) whose affairs were managed by a guardian, (3) who had not been continuously insured with the AOK Lower Saxony since 2015, or (4) who were employees of the health insurance company (Appendix 5 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).
      Sample size calculation was based on the equation ntac1000(or500) from Johnson and Orme
      • Orme B.
      Getting Started With Conjoint Analysis: Strategies for Product Design and Pricing Research.
      allowing stratification. With 8 choice sets (t), 2 alternatives within each choice set (a), and 3 levels across all attributes (c), the minimum sample size was n = 188 (or n = 94). Insurees were stratified based on age, sex, and previous experience with CRC screening, giving 8 different subgroups (Appendices 5 and 6 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012). We assigned a minimum sample size of n = 188 each for the 50-year-old men and 55-year-old women and of n = 94 each for the other strata. The resulting total sample size was n = 940. Assuming a minimum response rate of 20%, 5000 insurees were invited to participate. The initial mailing of the questionnaires took place between June 22, 2020, and June 25, 2020. To increase the response, questionnaires were sent out once again 2 and a half weeks later to insurees who had not yet completed them. Participation was voluntary and based on informed consent including the right to refuse or withdraw at any time without any disadvantages. In compliance with data protection requirements, the AOK Lower Saxony knew the response status of their insurees but not the answers in the questionnaire itself, whereas the research team at Hannover Medical School did not receive any information allowing personal identification of participants. Questionnaires received up to October 31, 2020, were considered for statistical analysis.

      Statistical Analysis

      Descriptive analysis and, based on routine data of the insurees, a nonresponder analysis by age, sex, and previous experience with CRC screening were conducted using SPSS (versions 25.0, 26.0; IBM Corp, Armonk, NY). The preference data were analyzed 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.
      • Ryan M.
      • Kolstad J.R.
      • Rockers P.C.
      • Dolea C.
      How to conduct a discrete choice experiment for health workforce recruitment and retention in remote and rural areas: a user guide with case studies. WHO.
      • Ryan M.
      • Gerard K.
      Using discrete choice experiments to value health care programmes.
      To account for preference heterogeneity across respondents, a random-parameters logit and a latent-class model (LCM) were estimated.
      • Bridges J.F.P.
      • Hauber A.B.
      • Marshall D.
      • et al.
      Conjoint analysis applications in health-a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force.
      ,
      • 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.
      ,
      • Ryan M.
      • Gerard K.
      Using discrete choice experiments to value health care programmes.
      ,
      • Zhou M.
      • Thayer W.M.
      • Bridges J.F.P.
      Using latent class analysis to model preference heterogeneity in health: a systematic review.
      In deciding on the appropriate number of classes, goodness of fit and interpretability were considered.
      • 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.
      ,
      • Zhou M.
      • Thayer W.M.
      • Bridges J.F.P.
      Using latent class analysis to model preference heterogeneity in health: a systematic review.
      Differences in class-specific characteristics were examined using chi-squared tests
      • Zhou M.
      • Thayer W.M.
      • Bridges J.F.P.
      Using latent class analysis to model preference heterogeneity in health: a systematic review.
      with a significance level of P ≤ .05. Relative attribute importance was calculated as the range between the most and least preferred level of an attribute
      • Orme B.
      Getting Started With Conjoint Analysis: Strategies for Product Design and Pricing Research.
      ,
      • 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.
      ,
      • Gonzalez J.M.
      A guide to measuring and interpreting attribute importance.
      and then normalized by setting one attribute to a value of 10 as the reference attribute for all classes.
      • Gonzalez J.M.
      A guide to measuring and interpreting attribute importance.
      All model estimations were conducted using STATA (version 15.1; StataCorp LLC, TX). Attributes were included as categorical variables. Levels were effects coded, with the last level being determined as reference. Numerical attributes were not specified as continuous variables because model fit was reduced, and the results from the categorical model indicated different relative marginal utilities associated with a 1-unit change over the range of levels within attributes.
      • 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.
      Each variable corresponded to a main effect; interaction terms between attributes were not included because model fit was only slightly improved, and choices were explained equally well compared with the main effects model.
      Participants who chose the dominated alternative in the internal validity test (procedure A in Fig. 1) were not excluded, but analysis was stratified by type of response (“rationality”).
      • Lancsar E.
      • Louviere J.
      Conducting discrete choice experiments to inform healthcare decision making: a user’s guide.

      Results

      Of 1282 questionnaires received (response rate: 25.9% [1282 of 4945]), 1142 were considered in the analysis of the preference data (Appendices 5 and 6 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).

      Characteristics of Respondents and Nonresponder Analysis

      A description of the sample (n = 1142) is presented in Table 2. Differences between responders (n = 1282) and nonresponders (n = 3663) are presented in Appendix 7 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012.
      Table 2Self-reported characteristics of respondents (n = 1142).
      CharacteristicsTotalType A
      Type A chose the dominant alternative in the internal validity choice set.
      respondents
      Type B
      Type B chose the dominated alternative in the internal validity choice set.
      respondents
      P value
      Chi-squared test with a significance level of P ≤ .05.
      n = 1142(%)n = 666(%)n = 476(%)
      Demographics
      Sex(n = 1133)(n = 660)(n = 473).363
       Female581(51.3)346(52.4)235(49.7)
       Male552(48.7)314(47.6)238(50.3)
      Age, years(n = 1135)(n = 663)(n = 472).121
       50189(16.7)121(18.3)68(14.4)
       55228(20.1)138(20.8)90(19.1)
       60718(63.3)404(60.9)314(66.5)
      General education level, CASMIN(n = 1086)(n = 638)(n = 448)< .001
       No completed/low493(45.4)238(37.3)255(56.9)
       Mediate417(38.4)285(44.7)132(29.5)
       High176(16.2)115(18.0)61(13.6)
      Current employment status(n = 1112)(n = 649)(n = 463).023
       Employed full-time595(53.5)347(53.5)248(53.6)
       Employed part-time301(27.1)191(29.4)110(23.8)
       Nonemployed216(19.4)111(17.1)105(22.7)
      Migration background(n = 1055)(n = 623)(n = 432).005
       Yes179(17.0)89(14.3)90(20.8)
       No876(83.0)534(85.7)342(79.2)
      Level of certainty in choosing a procedure
      Certainty in choosing a procedure(n = 1106)(n = 644)(n = 462)< .001
       Not certain/rather not certain484(43.8)248(38.5)236(51.1)
       Rather certain/certain622(56.2)396(61.5)226(48.9)
      Screening history and intention to participate in screening
      Previous FOBT experience(n = 1110)(n = 648)(n = 462).748
       Yes879(79.2)511(78.9)368(79.7)
       No231(20.8)137(21.1)94(20.3)
      Previous colonoscopy experience(n = 1111)(n = 646)(n = 465)< .001
       Yes671(60.4)362(56.0)309(66.5)
       No440(39.6)284(44.0)156(33.5)
      Intention to participate in CRC screening(n = 1104)(n = 644)(n = 460).034
       Yes, FIT252(22.8)149(23.1)103(22.4)
       Yes, colonoscopy430(38.9)249(38.7)181(39.3)
       Yes, but not yet decided on a procedure144(13.0)98(15.2)46(10.0)
       No278(25.2)148(23.0)130(28.3)
      BMI and family history of CRC
      BMI
      Calculated from self-reported height and weight.
      categories, WHO
      (n = 1116)(n = 656)(n = 460).111
       Underweight/normal weight348(31.2)220(33.5)128(27.8)
       Preobesity415(37.2)232(35.4)183(39.8)
       Obesity class I-III353(31.6)204(31.1)149(32.4)
      CRC history of first degree relatives, parents(n = 1121)(n = 654)(n = 467).445
       Yes126(11.2)70(10.7)56(12.0)
       No930(83.0)550(84.1)380(81.4)
       I don’t know65(5.8)34(5.2)31(6.6)
      Health literacy
      How often do you have someone help you read hospital materials?(n = 1131)(n = 660)(n = 471).002
       Never/seldom807(71.4)497(75.3)310(65.8)
       Sometimes206(18.2)103(15.6)103(21.9)
       Often/always118(10.4)60(9.1)58(12.3)
      How often do you have problems learning about your medical condition because of difficulty reading hospital materials?(n = 1131)(n = 661)(n = 470).204
       Never/seldom704(62.2)419(63.4)285(60.6)
       Sometimes289(25.6)171(25.9)118(25.1)
       Often/always138(12.2)71(10.7)67(14.3)
      How confident are you filling out forms by yourself?(n = 1133)(n = 661)(n = 472).002
       Extremely/quite a bit579(51.1)367(55.5)212(44.9)
       Somewhat410(36.2)220(33.3)190(40.3)
       A little bit/not at all144(12.7)74(11.2)70(14.8)
      Note. Not all questions were completed by all participants. n is the number who answered the question if different from 1142.
      BMI indicates body mass index; CASMIN, comparative analysis of social mobility in industrial nations; CRC, colorectal cancer; FIT, fecal immunochemical test; FOBT, fecal occult blood test; WHO, World Health Organization.
      Type A chose the dominant alternative in the internal validity choice set.
      Type B chose the dominated alternative in the internal validity choice set.
      Chi-squared test with a significance level of P ≤ .05.
      § Calculated from self-reported height and weight.

      Internal Validity of the Preference Data

      Table 2 demonstrates that 476 (41.7%) of the 1142 respondents chose the dominated alternative in the internal validity test (type B respondents). Type A (n = 666) compared with type B respondents (n = 476) were on average higher educated, were less frequently unemployed or migrants, had less frequently colonoscopy experience and had more frequently not yet decided which procedure to use next for screening, declined participation in screening less frequently, had higher health literacy capacities, and had a higher level of certainty in the choices made.

      Preferences for Characteristics of CRC Screening Tests

      Statistically significant preference weights for all but one level (complications in 1 of 1000) in the conditional logit model for the total sample (Appendix 8 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012) indicate that all attributes had an impact on choice. Nevertheless, preference weights for preparation and need for transportation had not the a priori expected signs. Taking into account that 41.7% of the respondents chose the dominated alternative in the internal validity test (Table 2), a conditional logit model stratified by “rationality” was calculated. Goodness of fit for each of the 2 subgroups was better than for the overall model, so only these results are presented in the following (Fig. 2) (
      • Ferlay J.
      • Ervik M.
      • Lam F.
      • et al.
      Global cancer observatory: cancer today. IARC.
      in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).
      Figure thumbnail gr2
      Figure 2Conditional logit model (effects coded) with 95% confidence intervals for type A (n = 666) and type B (n = 476) respondents from the DCE on CRC screening preferences.
      CRC indicates colorectal cancer; DCE, discrete choice experiment.
      For type A respondents, all but 2 levels (complications in 1 of 1000; no transportation required) were statistically significant at P ≤ .05, indicating that all attributes had an impact on choice. Most preference weights had the signs expected a priori: Participants preferred fewer new cases and deaths caused by CRC, no complications, and no follow-up to more cases, deaths, complications, or follow-up tests. Occasionally needing transportation was preferred to almost always needing transportation. The preference weights of the attribute preparation were unexpected: No preparation was preferred to bowel cleansing by taking a laxative and avoiding solid food, which in turn was preferred to an enema (injection of liquid into the rectum) immediately before the examination. Most important attribute was cancer-specific mortality, closely followed by cancer-specific incidence. Complications and need for transportation were least important.
      For type B respondents, all but 4 levels (developing CRC in 3 of 1000; developing CRC in 8 of 1000; complications in 1 of 1000; follow-up required in 440 of 1000) were statistically significant at P ≤ .05. The preference weights had the a priori expected signs for the 2 attributes dying from CRC and complications: Fewer deaths caused by CRC and no complications were each preferred to more cancer-related deaths and complications. Developing CRC in 10 of 1000 was rejected. Both no preparation and preparation using enema had a negative influence on choice compared with preparation with laxatives, with no preparation required being least preferred. Comparably, almost always requiring transportation home was preferred to occasionally and never needing transportation home. More examinations after a positive, initial test result were preferred to no follow-up needed. The most important attribute was preparation, followed by need for transportation, and cancer-specific mortality, whereas follow-up testing, complications, and cancer-specific incidence were least important.

      Preference Heterogeneity: Random-Parameters Logit Model

      For type A respondents, standard deviations were large and statistically significant (
      • Ferlay J.
      • Ervik M.
      • Lam F.
      • et al.
      Global cancer observatory: cancer today. IARC.
      in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012). In type B respondents, fewer standard deviations were statistically significant. Nevertheless, large and statistically significant standard deviations were present for preparation and need for transportation.

      Preference Heterogeneity: LCM

      Three latent classes of respondents with similar preferences were identified (Fig. 3) (Appendix 11 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012). Class 1 (n = 505, 44% of respondents; bowel preparation) preferred bowel preparation using laxatives and avoiding solid food, almost always needing transportation home (preference weights of both attributes had not the expected signs), and low cancer-related incidence and mortality rates. Preparation was by far the most important attribute (Fig. 4) (Appendix 11 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012), whereas complications and follow-up testing had no impact on choice. Choices of class 2 (n = 347, 30%; cancer death prevention) were predominantly guided by low mortality rates. They also preferred fewer follow-up tests, occasionally needing transportation home (preference weights had not the signs expected a priori), low rates of new cancer cases, and not needing bowel preparation. Relative importance of attributes was (in descending order) cancer-related mortality, follow-up, need for transportation, cancer-specific incidence, and preparation. Complications did not influence decision making. Members of class 3 (n = 290, 25%; cancer prevention) preferred low rates of new cancer-related cases. Their choice behavior was also determined by not needing bowel preparation, low complication rates, and never requiring transportation home. Preventing CRC was most important, followed by preparation and complications, whereas requiring a follow-up and dying from CRC had little or no impact on choice.
      Figure thumbnail gr3
      Figure 3LCM (effects coded) with 95% confidence intervals from the DCE on CRC screening preferences.
      CRC indicates colorectal cancer; DCE, discrete choice experiment; LCM, latent-class model.
      Figure thumbnail gr4
      Figure 4Relative importance of attributes for the 3 latent classes, defined as the range between the highest and lowest preference weight within an attribute and normalized, with preparation as the most important attribute of the largest class (class 1; n = 505) set to a value of 10 as the reference attribute for all classes.
      Compared with class 1, members of classes 2 and 3 were on average higher educated, had higher health literacy capacities (confidence filling out forms), and were less likely to be migrants, to be obese, to be uncertain about the choices made, or to choose the dominated alternative in the internal validity test (Appendix 12 in Supplemental Materials found at https://dx.doi.org/10.1016/j.jval.2022.07.012).

      Discussion

      Although studies eliciting stated preferences of the general or average-risk population for CRC screening using a DCE have already been conducted in particular in the United States, The Netherlands, and Australia,
      • Ghanouni A.
      • Smith S.G.
      • Halligan S.
      • et al.
      Public preferences for colorectal cancer screening tests: a review of conjoint analysis studies.
      ,
      • Wortley S.
      • Wong G.
      • Kieu A.
      • Howard K.
      Assessing stated preferences for colorectal cancer screening: a critical systematic review of discrete choice experiments.
      to the best of our knowledge, this is the first one performed in Germany. Including 1142 respondents, the DCE revealed heterogeneous preferences among the insurees and a high number of respondents who did not answer the dominant choice set as expected a priori (type B respondents). Type A respondents (n = 666, 58.3%) are characterized by higher education, fewer unemployed and migrants, less likely rejecting screening, higher health literacy, and a higher level of certainty in the choices made. They preferred screening modalities that result in both fewer new cancer cases and fewer cancer-related deaths. Type B respondents (n = 476, 41.7%) preferred tests that require bowel cleansing with laxatives and fluid diet and transportation home due to sedation. Contrary to a priori expectations, both favored a higher (organizational) effort. The LCM identified 3 groups with similar preferences. Choice behavior of the largest class (class 1; n = 505, 44% of respondents) was based, above all, on procedural issues (comprehensive bowel cleansing, occasionally requiring transportation home), whereas the other 2 classes primarily valued a benefit outcome most: either fewer cancer-related deaths (class 2; n = 347, 30%) or fewer new cancer cases (class 3; n = 290, 25%). Class 1 was characterized by lower education, more migrants, more frequently choosing the dominated alternative in the internal validity test, more uncertainty about the choices made, more obesity, and lower health literacy.
      An unexpectedly high proportion of 42% of the respondents chose the procedure considered being clearly inferior in the internal validity choice set. Given that “irrational” appearing decision making can arise from several factors, excluding respondents not having passed the test on internal validity might delete valid choices and bias the results.
      • Lancsar E.
      • Louviere J.
      Deleting ‘irrational’ responses from discrete choice experiments: a case of investigating or imposing preferences?.
      Apparently, some of the respondents found it reassuring to be picked up and accompanied home after the examination and did not value this as a (negative) expense but rather as desirable. This is in line with a study that explored why respondents’ preferences are not consistent with the assumptions of utility theory by combining completing a DCE with the qualitative think aloud method.
      • Ryan M.
      • Watson V.
      • Entwistle V.
      Rationalising the ‘irrational’: a think aloud study of discrete choice experiment responses.
      Participants who chose the alternative with a higher number of days of preprocedure dietary restrictions associated more extensive bowel cleansing with greater accuracy of the examination.
      • Ryan M.
      • Watson V.
      • Entwistle V.
      Rationalising the ‘irrational’: a think aloud study of discrete choice experiment responses.
      It is possible that the observed choices of the initially “irrational” appearing respondents were not arbitrary, but reflected sound (unobservable or omitted) reasons and “rational” considerations, even if these differ from a priori expectations of the research team. Nevertheless, the large number of type B respondents may also be due to shortcomings in the descriptions of attributes and levels or an increased cognitive complexity, despite qualitative pretesting.
      • Lancsar E.
      • Louviere J.
      Deleting ‘irrational’ responses from discrete choice experiments: a case of investigating or imposing preferences?.
      To enable reliable assumptions about whether or not “irrational” appearing responses are preference based, questions that test the understanding of the attributes used, open-ended questions about motivations or reasons for choosing the dominated alternative in the internal validity test, or questions about the level of confidence in each of the choices made could be implemented in future DCE surveys.
      • Bridges J.F.P.
      • Hauber A.B.
      • Marshall D.
      • et al.
      Conjoint analysis applications in health-a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force.
      The complementary use of qualitative methods during or after completing a DCE could also contribute to a better understanding of individual responses.
      • Soekhai V.
      • Bekker-Grob EW de
      • Ellis A.R.
      • Vass C.M.
      Discrete choice experiments in health economics: past, present and future.
      The LCM revealed 3 classes with an important heterogeneity in preferences for characteristics of CRC screening. The largest class (class 1) valued procedural issues the most, which contrasts to previous studies showing comparable attributes to be of secondary importance.
      • Marshall D.A.
      • Johnson F.R.
      • Kulin N.A.
      • et al.
      How do physician assessments of patient preferences for colorectal cancer screening tests differ from actual preferences? A comparison in Canada and the United States using a stated-choice survey.
      • Marshall D.A.
      • Johnson F.R.
      • Phillips K.A.
      • Marshall J.K.
      • Thabane L.
      • Kulin N.A.
      Measuring patient preferences for colorectal cancer screening using a choice-format survey.
      • Groothuis-Oudshoorn C.G.M.
      • Fermont J.M.
      • van Til J.A.
      • Ijzerman M.J.
      Public stated preferences and predicted uptake for genome-based colorectal cancer screening.
      • Mansfield C.
      • Ekwueme D.U.
      • Tangka F.K.L.
      • et al.
      Colorectal cancer screening: preferences, past behavior, and future intentions.
      Members of this class were on average more likely to choose the dominated alternative in the internal validity test. Consequently, the same assumptions as above hold true, at least for the unexpected directions of preference weights of bowel cleansing and accompaniment home. At the same time, members of class 1 did (almost) not value the potential benefits of screening, which are presented as numbers suggesting that class 1 may have inferior numeracy skills as a reason for not having considered these attributes.
      • Rothman R.L.
      • Montori V.M.
      • Cherrington A.
      • Pignone M.P.
      Perspective: the role of numeracy in health care.
      A lower educational level and lower health literacy of class 1 being associated with lower numeracy may support this hypothesis. Lower health literacy itself, especially regarding the ability to understand, to evaluate, and to use information for decision making, could further explain the high(er) value of procedural issues.
      • Bitzer E.M.
      • Sørensen K.
      [Health literacy].
      • Sørensen K.
      • van den Broucke S.
      • Fullam J.
      • et al.
      Health literacy and public health: a systematic review and integration of definitions and models.
      • Okan O.
      • Bauer U.
      • Levin-Zamir D.
      • Pinheiro P.
      • Sørensen K.
      International Handbook of Health Literacy: Research, Practice and Policy Across the Lifespan.
      • Sørensen K.
      • Pelikan J.M.
      • Röthlin F.
      • et al.
      Health literacy in Europe: comparative results of the European health literacy survey (HLS-EU).
      Considering the certainty of the decisions made as an indication of cognitive burden, it can also be assumed that ignoring numerical attributes in the choice tasks led to heuristics simplifying decision making.
      • Heidenreich S.
      • Watson V.
      • Ryan M.
      • Phimister E.
      Decision heuristic or preference? Attribute non-attendance in discrete choice problems.
      Another reason to disregard benefit outcomes may be that subjects took the benefits for granted due to a long-standing cancer screening communication that emphasized benefits and downplayed harms to maximize participation rates.
      • Hersch J.K.
      • Nickel B.L.
      • Ghanouni A.
      • Jansen J.
      • McCaffery K.J.
      Improving communication about cancer screening: moving towards informed decision making.
      • Gummersbach E.
      • Piccoliori G.
      • Zerbe C.O.
      • et al.
      Are women getting relevant information about mammography screening for an informed consent: a critical appraisal of information brochures used for screening invitation in Germany, Italy, Spain and France.
      • Dreier M.
      • Borutta B.
      • Seidel G.
      • et al.
      Communicating the benefits and harms of colorectal cancer screening needed for an informed choice: a systematic evaluation of leaflets and booklets.
      • Caverly T.J.
      • Hayward R.A.
      • Reamer E.
      • et al.
      Presentation of benefits and harms in US cancer screening and prevention guidelines: systematic review.
      The implicit assumption of benefit may be even higher for procedures offered and reimbursed by the statutory health insurance.
      For members of classes 2 and 3, the most important attribute was either to detect CRC at an early stage (class 2) or to prevent CRC by removing precancerous lesions (class 3). This relevance of benefit outcomes for decision making is in line with the overall screening logic and previous studies eliciting stated preferences for CRC screening.
      • Salkeld G.
      • Solomon M.
      • Short L.
      • Ryan M.
      • Ward J.E.
      Evidence-based consumer choice: a case study in colorectal cancer screening.
      • van Dam L.
      • Hol L.
      • Bekker-Grob EW de
      • et al.
      What determines individuals’ preferences for colorectal cancer screening programmes? A discrete choice experiment.
      • Pignone M.P.
      • Brenner A.T.
      • Hawley S.
      • et al.
      Conjoint analysis versus rating and ranking for values elicitation and clarification in colorectal cancer screening.
      • Brenner A.
      • Howard K.
      • Lewis C.
      • et al.
      Comparing 3 values clarification methods for colorectal cancer screening decision-making: a randomized trial in the US and Australia.
      Nevertheless, all of these studies included a screening modality’s ability to reduce cancer incidence only in combination with the reduction of cancer mortality.
      • Pignone M.P.
      • Brenner A.T.
      • Hawley S.
      • et al.
      Conjoint analysis versus rating and ranking for values elicitation and clarification in colorectal cancer screening.
      ,
      • Brenner A.
      • Howard K.
      • Lewis C.
      • et al.
      Comparing 3 values clarification methods for colorectal cancer screening decision-making: a randomized trial in the US and Australia.
      Future DCE surveys should include a separate attribute on incidence, given that to prevent cancer is clearly more beneficial than to prevent death from cancer and, as indicated by the results of the present study, is highly relevant for decision making for at least 25% of insurees.
      Although a comprehensive and unbiased presentation of benefits and harms is essential for enabling individual informed decision making,
      • Rimer B.K.
      • Briss P.A.
      • Zeller P.K.
      • Chan E.C.Y.
      • Woolf S.H.
      Informed decision making: what is its role in cancer screening?.
      complications had no impact on choices for both class 1 and 2 members. Nevertheless, this confirms previous results of comparable studies
      • Marshall D.A.
      • Johnson F.R.
      • Phillips K.A.
      • Marshall J.K.
      • Thabane L.
      • Kulin N.A.
      Measuring patient preferences for colorectal cancer screening using a choice-format survey.
      • Groothuis-Oudshoorn C.G.M.
      • Fermont J.M.
      • van Til J.A.
      • Ijzerman M.J.
      Public stated preferences and predicted uptake for genome-based colorectal cancer screening.
      • Mansfield C.
      • Ekwueme D.U.
      • Tangka F.K.L.
      • et al.
      Colorectal cancer screening: preferences, past behavior, and future intentions.
      ,
      • van Dam L.
      • Hol L.
      • Bekker-Grob EW de
      • et al.
      What determines individuals’ preferences for colorectal cancer screening programmes? A discrete choice experiment.
      in which attributes such as pain or discomfort or the risk of complications were predominantly secondary. However, when asked, insurees want at least to be informed about potential harms with high priority.
      • Dreier M.
      • Krueger K.
      • Walter U.
      Patient-rated importance of key information on screening colonoscopy in Germany: a survey of statutory health insurance members.
      The partially low importance of both attributes on benefits, that is, reducing disease incidence, and harms discovered huge deficits in understanding the rationale and key facts of screening procedures. Nevertheless, like in subjects who preferred process parameters for unknown reasons and as discussed earlier for type A and type B respondents, we cannot exclude simplified heuristic decision making for lower valuing key attributes on benefits and harms. In addition to further research efforts, public health and healthcare services may address these heterogeneous preferences by adapting information services to the needs of the target groups
      • Austoker J.
      • Giordano L.
      • Hewitson P.
      • Villain P.
      International Agency for Research on Cancer
      European guidelines for quality assurance in colorectal cancer screening and diagnosis. First edition--communication.
      and prioritizing and intensifying interventions to increase health literacy.
      This study has limitations. First, using an experimental design without an opt-out alternative resulted in a forced choice treating every respondent as a “demander” because voting against screening participation was, unlike in reality, not possible.
      • Ryan M.
      • Skåtun D.
      Modelling non-demanders in choice experiments.
      This may have led to missing values or even irrational appearing (protest) answers among responding “nondemanders” and must be considered when deriving (policy) recommendations.
      • Ryan M.
      • Skåtun D.
      Modelling non-demanders in choice experiments.
      Second, as recommended by ISPOR,
      • Bridges J.F.P.
      • Hauber A.B.
      • Marshall D.
      • et al.
      Conjoint analysis applications in health-a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force.
      attribute identification and selection were based on evidence from a systematic review and focus groups.
      • Brinkmann M.
      • Von Holt I.
      • Diedrich L.
      • Krauth C.
      • Seidel G.
      • Dreier M.
      Attributes Characterizing Colorectal Cancer Screening Tests That Influence Preferences of Individuals Eligible for Screening in Germany: A Qualitative Study.
      Nevertheless, given that the number of attributes used to define the choice tasks has an impact on the cognitive complexity of the experiment, it is possible that attributes considered relevant for decision making, for example, overdiagnosis,
      • Kalager M.
      • Wieszczy P.
      • Lansdorp-Vogelaar I.
      • Corley D.A.
      • Bretthauer M.
      • Kaminski M.F.
      Overdiagnosis in colorectal cancer screening: time to acknowledge a blind spot.
      ,
      • Robra B.-P.
      Harms and benefits of cancer screening.
      were excluded to ensure feasibility. Third, 4 of 6 attributes were quantitative, with numbers (rates) presented as natural frequencies. Although the same denominators were used to facilitate comparability
      • Bunge M.
      • Mühlhauser I.
      • Steckelberg A.
      What constitutes evidence-based patient information? Overview of discussed criteria.
      and quantitative attributes are associated with a lower risk of being interpreted differently than qualitative ones,
      • Ryan M.
      • Gerard K.
      • Amaya-Amaya M.
      Using Discrete Choice Experiments to Value Health and Health Care.
      it is possible that respondents with lower numeracy skills had difficulties in understanding and interpreting the choice tasks correctly.
      • Rothman R.L.
      • Montori V.M.
      • Cherrington A.
      • Pignone M.P.
      Perspective: the role of numeracy in health care.
      ,
      • Okan O.
      • Bauer U.
      • Levin-Zamir D.
      • Pinheiro P.
      • Sørensen K.
      International Handbook of Health Literacy: Research, Practice and Policy Across the Lifespan.
      Finally, this survey was conducted during the COVID-19 pandemic. Although studies from Germany show that the use of screening colonoscopy in the summer of 2020 reached the level of the same period in the previous year or even higher, the extent to which the pandemic influenced preferences remains unknown.
      Veränderung der vertragsärztlichen Leistungsinanspruchnahme während der COVID-Krise
      Tabellarischer Trendreport für das Jahr 2020. Central Research Institute of Ambulatory Health Care in Germany.

      Conclusion

      This DCE revealed heterogeneous preferences. A majority of respondents valued the potential benefits of screening tests most whereas the others focused on the tests’ process parameters or utility. This warrants further research on the underlying reasons and individuals’ motivations including internal validity issues of DCE methods. Different preferences also suggest different informational needs that public health and healthcare services should address to support everyone regardless of their health literacy level in making individual informed decisions by weighing benefits, harms, and processes.

      Article and Author Information

      Author Contributions: Concept and design: Brinkmann, Krauth, Robra, Stahmeyer, Dreier
      Acquisition of data: Brinkmann, Diedrich, Hemmerling
      Analysis and interpretation of data: Brinkmann, Diedrich, Hemmerling, Krauth, Robra, Stahmeyer, Dreier
      Drafting of the manuscript: Brinkmann
      Critical revision of the paper for important intellectual content: Brinkmann, Diedrich, Hemmerling, Krauth, Robra, Stahmeyer, Dreier
      Statistical analysis: Brinkmann, Hemmerling, Stahmeyer
      Provision of study materials or patients: Brinkmann, Diedrich, Hemmerling
      Obtaining funding: Krauth, Stahmeyer, Dreier
      Supervision: Krauth, Robra, Stahmeyer, Dreier
      Conflict of Interest Disclosures: Mses Brinkmann, Diedrich, and Hemmerling and Drs Krauth, Stahmeyer, and Dreier reported receiving grants from the German Federal Joint Committee’s Innovation Fund (grant number 01VSF18007). Dr Krauth reported receiving project-specific financial support to the Institute for Epidemiology, Social Medicine and Health Systems Research (Hannover Medical School) from the German Federal Ministry of Education and Research and the German Federal Ministry of Health and personal payments for support in project applications and economic analyses from Inav GmbH (Institut für angewandte Versorgungsforschung), Berlin. Dr Robra reported serving as a member of the scientific advisory boards of Inav GmbH, Berlin, and of the WIdO (AOK Research Institute, Wissenschaftliches Institut der AOK) for the past 3 years and reported being honorary chairman of the Landesschiedsamt für die vertragsärztliche Versorgung Sachsen-Anhalt (Board of Arbitration, statutory health services, Federal State of Saxony-Anhalt) from 2013 to 2020. No other disclosures were reported.
      Funding/Support: This study was funded by the German Federal Joint Committee’s Innovation Fund (grant number 01VSF18007 ).
      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.

      Supplemental Materials

      References

        • Ferlay J.
        • Ervik M.
        • Lam F.
        • et al.
        Global cancer observatory: cancer today. IARC.
        https://gco.iarc.fr/
        Date accessed: August 13, 2021
        • Helsingen L.M.
        • Vandvik P.O.
        • Jodal H.C.
        • et al.
        Colorectal cancer screening with faecal immunochemical testing, sigmoidoscopy or colonoscopy: a clinical practice guideline.
        BMJ. 2019; 367: l5515
        • Bénard F.
        • Barkun A.N.
        • Martel M.
        • von Renteln D.
        Systematic review of colorectal cancer screening guidelines for average-risk adults: summarizing the current global recommendations.
        World J Gastroenterol. 2018; 24: 124-138
        • Navarro M.
        • Nicolas A.
        • Ferrandez A.
        • Lanas A.
        Colorectal cancer population screening programs worldwide in 2016: an update.
        World J Gastroenterol. 2017; 23: 3632-3642
      1. Directive on organized cancer screening programs. Federal Joint Committee.
        https://www.g-ba.de/richtlinien/104/
        Date accessed: June 15, 2020
      2. S3-guideline colorectal cancer, long version 2.1, 2019. German Cancer society, German Cancer Aid, AWMF.
        • Bretthauer M.
        • Kalager M.
        Principles, effectiveness and caveats in screening for cancer.
        Br J Surg. 2013; 100: 55-65
        • Brenner H.
        • Stock C.
        • Hoffmeister M.
        Effect of screening sigmoidoscopy and screening colonoscopy on colorectal cancer incidence and mortality: systematic review and meta-analysis of randomised controlled trials and observational studies.
        BMJ. 2014; 348: g2467
        • Jodal H.C.
        • Helsingen L.M.
        • Anderson J.C.
        • Lytvyn L.
        • Vandvik P.O.
        • Emilsson L.
        Colorectal cancer screening with faecal testing, sigmoidoscopy or colonoscopy: a systematic review and network meta-analysis.
        BMJ Open. 2019; 9e032773
        • Lin J.S.
        • Perdue L.A.
        • Henrikson N.B.
        • Bean S.I.
        • Blasi P.R.
        Screening for colorectal cancer: updated evidence report and systematic review for the US Preventive Services Task Force [published correction appears in JAMA. 2021;326(3):279].
        JAMA. 2021; 325: 1978-1997
        • Eckstrom E.
        • Feeny D.H.
        • Walter L.C.
        • Perdue L.A.
        • Whitlock E.P.
        Individualizing cancer screening in older adults: a narrative review and framework for future research.
        J Gen Intern Med. 2013; 28: 292-298
      3. Council recommendation of 2 December 2003 on cancer screening (2003/878/EC). The Council of the European Union.
      4. Patient preference information – voluntary submission, review in premarket approval applications, humanitarian device exemption applications, and de novo requests, and inclusion in decision summaries and device labeling. Guidance for industry, Food and Drug Administration staff, and other stakeholders. US Food and Drug Administration.
        https://www.fda.gov/media/92593/download
        Date accessed: February 5, 2021
        • Brinkmann M.
        • Diedrich L.
        • Krauth C.
        • Robra B.-P.
        • Stahmeyer J.T.
        • Dreier M.
        General populations’ preferences for colorectal cancer screening: rationale and protocol for the discrete choice experiment in the SIGMO study.
        BMJ Open. 2021; 11e042399
      5. General methods, version 5.0. Institute for Quality and Efficiency in Health Care.
        • Bridges J.F.P.
        • Hauber A.B.
        • Marshall D.
        • et al.
        Conjoint analysis applications in health-a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force.
        Value Health. 2011; 14: 403-413
        • Brinkmann M.
        • Von Holt I.
        • Diedrich L.
        • Krauth C.
        • Seidel G.
        • Dreier M.
        Attributes Characterizing Colorectal Cancer Screening Tests That Influence Preferences of Individuals Eligible for Screening in Germany: A Qualitative Study.
        Patient Prefer Adherence. 2022; 16: 2051-2066
        • Rimer B.K.
        • Briss P.A.
        • Zeller P.K.
        • Chan E.C.Y.
        • Woolf S.H.
        Informed decision making: what is its role in cancer screening?.
        Cancer. 2004; 101: 1214-1228
      6. Invitation and decision aid for bowel cancer screening. Institute for Quality and Efficiency in Health Care.
        https://www.iqwig.de/en/projects/p15-01.html
        Date accessed: August 11, 2021
        • Ball A.J.
        • Rees C.J.
        • Corfe B.M.
        • Riley S.A.
        Sedation practice and comfort during colonoscopy: lessons learnt from a national screening programme.
        Eur J Gastroenterol Hepatol. 2015; 27: 741-746
        • Maliampurakal A.J.
        • McMillan D.C.
        • Anderson J.H.
        • Horgan P.G.
        • Mansouri D.
        Factors associated with the efficacy of polyp detection during routine flexible sigmoidoscopy.
        Frontline Gastroenterol. 2018; 9: 135-142
        • Lin J.S.
        • Piper M.A.
        • Perdue L.A.
        • et al.
        Screening for colorectal cancer: an updated systematic review for the U.S. Preventive Services Task Force: evidence syntheses no. 135. Kaiser Permanente Center for Health Research.
        • Kuhfeld W.F.
        Marketing research methods in SAS: experimental design, choice, conjoint, and graphical techniques. SAS Institute Inc.
        • Lancsar E.
        • Louviere J.
        Conducting discrete choice experiments to inform healthcare decision making: a user’s guide.
        Pharmacoeconomics. 2008; 26: 661-677
        • Ryan M.
        • Gerard K.
        • Amaya-Amaya M.
        Using Discrete Choice Experiments to Value Health and Health Care.
        Springer, Dordrecht, The Netherlands2008
        • Chew L.D.
        • Griffin J.M.
        • Partin M.R.
        • et al.
        Validation of screening questions for limited health literacy in a large VA outpatient population.
        J Gen Intern Med. 2008; 23: 561-566
        • Brauns H.
        • Scherer S.
        • Steinmann S.
        The CASMIN educational classification in international comparative research.
        in: Hoffmeyer-Zlotnik J.H.P. Wolf C. Advances in Cross-National Comparison. Springer, Boston, MA2003: 221-244
        • Johnson F.R.
        • Yang J.-C.
        • Reed S.D.
        The internal validity of discrete choice experiment data: a testing tool for quantitative assessments.
        Value Health. 2019; 22: 157-160
        • Orme B.
        Getting Started With Conjoint Analysis: Strategies for Product Design and Pricing Research.
        2nd ed. Research Publishers LLC, Madison, WI2010
        • 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
        • Ryan M.
        • Kolstad J.R.
        • Rockers P.C.
        • Dolea C.
        How to conduct a discrete choice experiment for health workforce recruitment and retention in remote and rural areas: a user guide with case studies. WHO.
        • Ryan M.
        • Gerard K.
        Using discrete choice experiments to value health care programmes.
        Appl Health Econ Health Policy. 2003; 2: 55-64
        • Zhou M.
        • Thayer W.M.
        • Bridges J.F.P.
        Using latent class analysis to model preference heterogeneity in health: a systematic review.
        Pharmacoeconomics. 2018; 36: 175-187
        • Gonzalez J.M.
        A guide to measuring and interpreting attribute importance.
        Patient. 2019; 12: 287-295
        • Ghanouni A.
        • Smith S.G.
        • Halligan S.
        • et al.
        Public preferences for colorectal cancer screening tests: a review of conjoint analysis studies.
        Expert Rev Med Devices. 2013; 10: 489-499
        • Wortley S.
        • Wong G.
        • Kieu A.
        • Howard K.
        Assessing stated preferences for colorectal cancer screening: a critical systematic review of discrete choice experiments.
        Patient. 2014; 7: 271-282
        • Lancsar E.
        • Louviere J.
        Deleting ‘irrational’ responses from discrete choice experiments: a case of investigating or imposing preferences?.
        Health Econ. 2006; 15: 797-811
        • Ryan M.
        • Watson V.
        • Entwistle V.
        Rationalising the ‘irrational’: a think aloud study of discrete choice experiment responses.
        Health Econ. 2009; 18: 321-336
        • Soekhai V.
        • Bekker-Grob EW de
        • Ellis A.R.
        • Vass C.M.
        Discrete choice experiments in health economics: past, present and future.
        Pharmacoeconomics. 2019; 37: 201-226
        • Marshall D.A.
        • Johnson F.R.
        • Kulin N.A.
        • et al.
        How do physician assessments of patient preferences for colorectal cancer screening tests differ from actual preferences? A comparison in Canada and the United States using a stated-choice survey.
        Health Econ. 2009; 18: 1420-1439
        • Marshall D.A.
        • Johnson F.R.
        • Phillips K.A.
        • Marshall J.K.
        • Thabane L.
        • Kulin N.A.
        Measuring patient preferences for colorectal cancer screening using a choice-format survey.
        Value Health. 2007; 10: 415-430
        • Groothuis-Oudshoorn C.G.M.
        • Fermont J.M.
        • van Til J.A.
        • Ijzerman M.J.
        Public stated preferences and predicted uptake for genome-based colorectal cancer screening.
        BMC Med Inform Decis Mak. 2014; 14: 18
        • Mansfield C.
        • Ekwueme D.U.
        • Tangka F.K.L.
        • et al.
        Colorectal cancer screening: preferences, past behavior, and future intentions.
        Patient. 2018; 11: 599-611
        • Rothman R.L.
        • Montori V.M.
        • Cherrington A.
        • Pignone M.P.
        Perspective: the role of numeracy in health care.
        J Health Commun. 2008; 13: 583-595
        • Bitzer E.M.
        • Sørensen K.
        [Health literacy].
        Gesundheitswesen. 2018; 80: 754-766
        • Sørensen K.
        • van den Broucke S.
        • Fullam J.
        • et al.
        Health literacy and public health: a systematic review and integration of definitions and models.
        BMC Public Health. 2012; 12: 80
        • Okan O.
        • Bauer U.
        • Levin-Zamir D.
        • Pinheiro P.
        • Sørensen K.
        International Handbook of Health Literacy: Research, Practice and Policy Across the Lifespan.
        Policy Press, Bristol, Chicago2019
        • Sørensen K.
        • Pelikan J.M.
        • Röthlin F.
        • et al.
        Health literacy in Europe: comparative results of the European health literacy survey (HLS-EU).
        Eur J Public Health. 2015; 25: 1053-1058
        • Heidenreich S.
        • Watson V.
        • Ryan M.
        • Phimister E.
        Decision heuristic or preference? Attribute non-attendance in discrete choice problems.
        Health Econ. 2018; 27: 157-171
        • Hersch J.K.
        • Nickel B.L.
        • Ghanouni A.
        • Jansen J.
        • McCaffery K.J.
        Improving communication about cancer screening: moving towards informed decision making.
        Public Health Res Pract. 2017; 272731728
        • Gummersbach E.
        • Piccoliori G.
        • Zerbe C.O.
        • et al.
        Are women getting relevant information about mammography screening for an informed consent: a critical appraisal of information brochures used for screening invitation in Germany, Italy, Spain and France.
        Eur J Public Health. 2010; 20: 409-414
        • Dreier M.
        • Borutta B.
        • Seidel G.
        • et al.
        Communicating the benefits and harms of colorectal cancer screening needed for an informed choice: a systematic evaluation of leaflets and booklets.
        PLoS One. 2014; 9e107575
        • Caverly T.J.
        • Hayward R.A.
        • Reamer E.
        • et al.
        Presentation of benefits and harms in US cancer screening and prevention guidelines: systematic review.
        J Natl Cancer Inst. 2016; 108: djv436
        • Salkeld G.
        • Solomon M.
        • Short L.
        • Ryan M.
        • Ward J.E.
        Evidence-based consumer choice: a case study in colorectal cancer screening.
        Aust N Z J Public Health. 2003; 27: 449-455
        • van Dam L.
        • Hol L.
        • Bekker-Grob EW de
        • et al.
        What determines individuals’ preferences for colorectal cancer screening programmes? A discrete choice experiment.
        Eur J Cancer. 2010; 46: 150-159
        • Pignone M.P.
        • Brenner A.T.
        • Hawley S.
        • et al.
        Conjoint analysis versus rating and ranking for values elicitation and clarification in colorectal cancer screening.
        J Gen Intern Med. 2012; 27: 45-50
        • Brenner A.
        • Howard K.
        • Lewis C.
        • et al.
        Comparing 3 values clarification methods for colorectal cancer screening decision-making: a randomized trial in the US and Australia.
        J Gen Intern Med. 2014; 29: 507-513
        • Dreier M.
        • Krueger K.
        • Walter U.
        Patient-rated importance of key information on screening colonoscopy in Germany: a survey of statutory health insurance members.
        BMJ Open. 2018; 8e019127
        • Austoker J.
        • Giordano L.
        • Hewitson P.
        • Villain P.
        • International Agency for Research on Cancer
        European guidelines for quality assurance in colorectal cancer screening and diagnosis. First edition--communication.
        Endoscopy. 2012; 44: SE164-SE185
        • Ryan M.
        • Skåtun D.
        Modelling non-demanders in choice experiments.
        Health Econ. 2004; 13: 397-402
        • Kalager M.
        • Wieszczy P.
        • Lansdorp-Vogelaar I.
        • Corley D.A.
        • Bretthauer M.
        • Kaminski M.F.
        Overdiagnosis in colorectal cancer screening: time to acknowledge a blind spot.
        Gastroenterology. 2018; 155: 592-595
        • Robra B.-P.
        Harms and benefits of cancer screening.
        in: Bauer A.W. Hofheinz R.-D. Utikal J.S. Ethical Challenges in Cancer Diagnosis and Therapy. Springer, Cham2021: 85-104
        • Bunge M.
        • Mühlhauser I.
        • Steckelberg A.
        What constitutes evidence-based patient information? Overview of discussed criteria.
        Patient Educ Couns. 2010; 78: 316-328
        • Veränderung der vertragsärztlichen Leistungsinanspruchnahme während der COVID-Krise
        Tabellarischer Trendreport für das Jahr 2020. Central Research Institute of Ambulatory Health Care in Germany.