Skip to content

Measuring the Preferences of Patients With Type II Diabetes Using Best-worst Scaling and Discrete Choice Experiment

Advancing Stated-preference Methods for Measuring the Preferences of Patients With Type II Diabetes

Status
Completed
Phases
Unknown
Study type
Observational
Source
ClinicalTrials.gov
Registry ID
NCT02637622
Enrollment
1103
Registered
2015-12-22
Start date
2013-07-31
Completion date
2017-02-28
Last updated
2024-11-29

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Type II Diabetes

Keywords

conjoint analysis, stated-preference methods, patient centered outcome research (PCOR), community-based participatory research (CBPR), patient preference, diabetes

Brief summary

In 2012, the FDA Center for Devices and Radiological Health (CDRH) issued guidance to clarify the principal benefit-risk factors FDA considers during the reviews for premarket approval applications and de novo classification requests. In addition to a detailed description of benefits and risks, CDRH listed patient tolerance for risk and perspective on benefit as a factor that CDRH may consider in regulatory reviews. It underlined the need for developing methods to measure patient preference and incorporate it into regulatory decision-making. The purpose of this study is to advance methods for patient and community engagement in patient-centered outcome research (PCOR) and has three objectives. First, demonstrate good practices for patient and community involvement in PCOR projects by applying principles of community-based participatory research (CBPR). Second, address methodological gaps pertaining to the use of stated-preference methods in studying preferences in PCOR. These include identifying the best methods for designing a preference study and strategies for analyzing variation in preferences. The investigators also seek to assess the relevance of stated-preference methods to patients and stakeholders using both qualitative and quantitative methods. Third, demonstrate good practices for applying stated-preference methods by studying the preferences of patients with type II diabetes. While type II diabetes provides an important case study, this research will advance approaches and methods that will be broadly generalizable to other diseases, and to diverse patient and stakeholder groups. Clinical Significance: This project will illustrate and advance methods for assessing the values of patients and stakeholders. It will demonstrate how CBPR methods apply to PCOR studies and the value of stated-preference methods in measuring the preferences of patients and stakeholders and directing health care.

Detailed description

This study will apply the principles of CBPR to involve patients and stakeholders associated with a local community board and a national diabetes advisory board in key decisions in the project. During year 1, the investigators will utilize mixed methods to develop, pretest, and pilot the survey instruments to assess the preferences of patients with type II diabetes. In year 2, the investigators plan further engagement to finalize the survey instruments, and will implement a nationally representative, racially/ethnically diverse sample of patients with type II diabetes. Based on further consultation, the investigators will conduct statistical analysis, including stratified analyses and segmentation of patients with similar preferences. In year 3, mixed methods will be applied to assess beliefs of patients and other consumers about the relevance of this work and its generalizability to other PCOR topics. Finally, lay language reports will be developed to highlight patient and stakeholder engagement and the application of stated-preference methods to the study of the preferences of patients with type II diabetes.

Interventions

OTHERBest-Worst Scaling (Case 2)

Respondents receive questions asking them to choose the best and worst features of a hypothetical medication.

Respondents receive questions asking them to choose the medication they prefer between a pair of hypothetical medications.

Sponsors

Patient-Centered Outcomes Research Institute
CollaboratorOTHER
Johns Hopkins Bloomberg School of Public Health
Lead SponsorOTHER

Study design

Observational model
CASE_ONLY
Time perspective
CROSS_SECTIONAL

Eligibility

Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No

Inclusion criteria

* Participate in the GfK's KnowledgePanel * Self-reported Type II diabetes diagnosis

Exclusion criteria

* Does not have Type II diabetes diagnosis * Unable to communicate in English or Spanish

Design outcomes

Primary

MeasureTime frameDescription
Patient Medication Preference as Estimated by a Choice ModelOne-time surveyPreference estimate for each medicine attribute level from a conditional logit regression analysis. For each arm, there were six attributes of the medication, with 3 levels each. Within each attribute the parameter estimates for each of the levels sum to 0. If a parameter estimate for a level is above (below) 0 then the parameter is higher (lower) than average for that medication attribute.
Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelOne-time surveyRelative attribute importance (RAI) for each attribute in each arm measures the overall importance of that attribute. It is estimated by subtracting the lowest parameter estimate from the highest parameter estimate within each attribute. The RAI was then re-scaled on a 0 to 10 scale with 0 demonstrating no importance and 10 reflecting the most important attribute in each arm.

Secondary

MeasureTime frameDescription
Self-reported Difficulty in Understanding and Answering the Survey QuestionsOne-time surveyQuestions that asked the respondents to evaluate if the Best-Worst Scaling (BWS) or Discrete Choice Experiment (DCE tasks were easy to understand and answer.

Countries

United States

Participant flow

Recruitment details

The survey was conducted through GfK KnowledgePanel. Eligible members were invited by email to participate in the survey. All participants were required to be 18 years or older with self-reported physician-diagnosed type 2 diabetes and able to read English or Spanish. African Americans and Latinos were oversampled.

Pre-assignment details

638 respondents (34%) did not respond to the survey and 139 people among those who responded (11%) were excluded based on the screening questions.

Participants by arm

ArmCount
Best-Worst Scaling (BWS) (Case 2)
Preference measured by best-worst scaling (case 2) Best-Worst Scaling (Case 2): Respondents receive questions asking them to choose the best and worst features of a hypothetical medication.
551
Discrete Choice Experiment (DCE)
Preference measured by discrete choice experiment Discrete Choice Experiment: Respondents receive questions asking them to choose the medication they prefer between a pair of hypothetical medications.
552
Total1,103

Baseline characteristics

CharacteristicBest-Worst Scaling (BWS) (Case 2)TotalDiscrete Choice Experiment (DCE)
Age, Continuous62.55 years61.93 years61.30 years
Complications
No
314 Participants640 Participants326 Participants
Complications
Yes
237 Participants463 Participants226 Participants
Education
Bachelor's degree or higher
156 Participants321 Participants165 Participants
Education
High school
174 Participants362 Participants188 Participants
Education
Less than high school
39 Participants82 Participants43 Participants
Education
Some college
182 Participants338 Participants156 Participants
Income
$25,000 - $49,999
160 Participants308 Participants148 Participants
Income
$50,000 - $74,999
111 Participants222 Participants111 Participants
Income
$75,000 and above
139 Participants300 Participants161 Participants
Income
Less than $25,000
141 Participants273 Participants132 Participants
Most recent HbA1c level
Don't know
92 Participants183 Participants91 Participants
Most recent HbA1c level
Over 7.0% but under 8.0%
144 Participants297 Participants153 Participants
Most recent HbA1c level
Over 8.0% but under 9.0%
46 Participants105 Participants59 Participants
Most recent HbA1c level
Over 9.0%
37 Participants58 Participants21 Participants
Most recent HbA1c level
Under 7.0%
232 Participants460 Participants228 Participants
No. of HbA1c level measured in past 6 months
1 time
253 Participants492 Participants239 Participants
No. of HbA1c level measured in past 6 months
Don't know
21 Participants32 Participants11 Participants
No. of HbA1c level measured in past 6 months
More than 2 times
246 Participants507 Participants261 Participants
No. of HbA1c level measured in past 6 months
None
31 Participants72 Participants41 Participants
No. of hypoglycemic events in the past 6 months
1 time
80 Participants160 Participants80 Participants
No. of hypoglycemic events in the past 6 months
2-5 times
144 Participants278 Participants134 Participants
No. of hypoglycemic events in the past 6 months
More than 5 times
49 Participants94 Participants45 Participants
No. of hypoglycemic events in the past 6 months
None
278 Participants571 Participants293 Participants
Other chronic conditions
No
90 Participants182 Participants92 Participants
Other chronic conditions
Yes
461 Participants921 Participants460 Participants
Race/Ethnicity, Customized
Black, Non-Hispanic
128 Participants254 Participants126 Participants
Race/Ethnicity, Customized
Hispanic
117 Participants236 Participants119 Participants
Race/Ethnicity, Customized
Other, Non-Hispanic
20 Participants38 Participants18 Participants
Race/Ethnicity, Customized
White, Non-Hispanic
286 Participants575 Participants289 Participants
Region of Enrollment
United States
551 Participants1103 Participants552 Participants
Self-reported health
Excellent
27 Participants60 Participants33 Participants
Self-reported health
Fair
103 Participants209 Participants106 Participants
Self-reported health
Good
233 Participants465 Participants232 Participants
Self-reported health
Poor
30 Participants53 Participants23 Participants
Self-reported health
Very good
158 Participants316 Participants158 Participants
Sex: Female, Male
Female
274 Participants553 Participants279 Participants
Sex: Female, Male
Male
277 Participants550 Participants273 Participants
Type of diabetes medicine used
No prescription medicine
62 Participants99 Participants37 Participants
Type of diabetes medicine used
Only injections/shots
48 Participants90 Participants42 Participants
Type of diabetes medicine used
Only pills
321 Participants666 Participants345 Participants
Type of diabetes medicine used
Pills and injections/shots
119 Participants246 Participants127 Participants
Type of diabetes medicine used
Refused
1 Participants2 Participants1 Participants
Years of diagnosis11.47 years
STANDARD_DEVIATION 8.26
11.35 years
STANDARD_DEVIATION 8.03
11.24 years
STANDARD_DEVIATION 7.81

Adverse events

Event typeEG000
affected / at risk
EG001
affected / at risk
deaths
Total, all-cause mortality
0 / 00 / 0
other
Total, other adverse events
0 / 00 / 0
serious
Total, serious adverse events
0 / 00 / 0

Outcome results

Primary

Patient Medication Preference as Estimated by a Choice Model

Preference estimate for each medicine attribute level from a conditional logit regression analysis. For each arm, there were six attributes of the medication, with 3 levels each. Within each attribute the parameter estimates for each of the levels sum to 0. If a parameter estimate for a level is above (below) 0 then the parameter is higher (lower) than average for that medication attribute.

Time frame: One-time survey

Population: Respondents who completed the survey.

ArmMeasureGroupValue (MEAN)
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 1%0.983 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 0.5%0.284 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 0%-1.267 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c stable 6 days/week0.859 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c stable 4 days/week-0.030 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHbA1c stable 2 days/week-0.828 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelNo hypoglycemia0.640 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHypoglycemia during the day-0.145 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelHypoglycemia during day and/or night-0.495 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelNo nausea1.158 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelNausea for 30 min/day-0.262 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice ModelNausea for 90 min/day-0.896 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model1 pill/day0.561 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model2 pills/day0.128 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model1 pill and 1 injection/day-0.689 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model$10 out-of-pocket cost0.900 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model$30 out-of-pocket cost-0.051 utility of
Best-Worst Scaling (Case 2)Patient Medication Preference as Estimated by a Choice Model$50 out-of-pocket cost-0.850 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model2 pills/day0.348 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 1%0.362 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelNo nausea0.739 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 0.5%0.261 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model$50 out-of-pocket cost-0.628 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c decrease by 0%-0.623 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelNausea for 30 min/day0.236 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c stable 6 days/week0.395 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model1 pill and 1 injection/day-0.805 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c stable 4 days/week0.073 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelNausea for 90 min/day-0.975 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHbA1c stable 2 days/week-0.469 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model$30 out-of-pocket cost0.077 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelNo hypoglycemia0.325 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model1 pill/day0.456 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHypoglycemia during the day0.104 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice Model$10 out-of-pocket cost0.550 utility of
Discrete Choice ExperimentPatient Medication Preference as Estimated by a Choice ModelHypoglycemia during day and/or night-0.429 utility of
Primary

Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice Model

Relative attribute importance (RAI) for each attribute in each arm measures the overall importance of that attribute. It is estimated by subtracting the lowest parameter estimate from the highest parameter estimate within each attribute. The RAI was then re-scaled on a 0 to 10 scale with 0 demonstrating no importance and 10 reflecting the most important attribute in each arm.

Time frame: One-time survey

Population: Respondents who completed the survey.

ArmMeasureGroupValue (MEAN)
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelHbA1c decrease10 units on a scale
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelDuration of stable HbA1c7.496 units on a scale
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelFrequency of hypoglycemia5.043 units on a scale
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelDuration of nausea9.125 units on a scale
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelTreatment burden5.553 units on a scale
Best-Worst Scaling (Case 2)Weight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelOut-of-pocket cost7.777 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelTreatment burden7.358 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelHbA1c decrease5.745 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelDuration of nausea10 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelDuration of stable HbA1c5.040 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelOut-of-pocket cost6.873 units on a scale
Discrete Choice ExperimentWeight Respondents Assign to Medication Attribute (Relative Attribute Importance) Assessed by a Choice ModelFrequency of hypoglycemia4.395 units on a scale
Secondary

Self-reported Difficulty in Understanding and Answering the Survey Questions

Questions that asked the respondents to evaluate if the Best-Worst Scaling (BWS) or Discrete Choice Experiment (DCE tasks were easy to understand and answer.

Time frame: One-time survey

Population: Respondents who completed the survey.

ArmMeasureGroupCategoryValue (COUNT_OF_PARTICIPANTS)
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsStrongly disagree20 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsDisagree49 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsNeither agree nor disagree123 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsAgree248 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsStrongly agree105 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsRefused6 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsStrongly disagree18 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsDisagree73 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsNeither agree nor disagree133 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsAgree225 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsStrongly agree94 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsRefused8 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesStrongly disagree8 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesDisagree9 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesNeither agree nor disagree82 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesAgree314 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesStrongly agree132 Participants
Best-Worst Scaling (Case 2)Self-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesRefused6 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesDisagree13 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsStrongly disagree13 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsAgree262 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsDisagree36 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesRefused0 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsNeither agree nor disagree111 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsStrongly agree89 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsAgree292 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesNeither agree nor disagree83 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsStrongly agree99 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsRefused2 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to understand the questionsRefused1 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesStrongly agree126 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsStrongly disagree15 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesStrongly disagree12 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsDisagree61 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsMy answers are consistent with my preferencesAgree318 Participants
Discrete Choice ExperimentSelf-reported Difficulty in Understanding and Answering the Survey QuestionsI found it easy to answer all the questionsNeither agree nor disagree123 Participants

Source: ClinicalTrials.gov · Data processed: Feb 4, 2026