Online Trial Examining Validity of the Shared Decision Making Process Survey With Video Vignettes

This study will recruit subjects online and randomly assigned them to one of four arms. The arms vary by clinical decision (colorectal cancer screening or treatment of high cholesterol) by video order (poor shared decision making followed by good or good shared decision making followed by poor). Participants will view two videos and complete the Shared Decision Making process survey along with a few other measures after each video. Our main hypothesis is that respondents watching the good shared decision making videos will score higher on the Shared Decision Making Process survey compared to those watching the poor videos.

Study staff are working with a national sampling firm to recruit subjects and obtain 400 responses. Subjects were randomly assigned to one of four arms. (1) Colorectal cancer screening good shared decision making video then poor video second (2) Colorectal cancer screening poor shared decision making video then good video (3) Treatment of high cholesterol good video then poor video and (4) Treatment of high cholesterol poor shared decision making video first then good video. Participants completed measures of Shared Decision Making after each video. The sample size was determined to ensure sufficient power to detect differences between the good and the poor shared decision making videos in this repeated measure design and analyses were planned to be separate for each arm (i.e. one analysis for the colorectal cancer screening videos and a separate parallel analysis of the statins for high cholesterol video). To detect an eta2 effect size of .04 with an alpha of 0.05 with 80% power would require 190 observations per clinical condition. Study staff rounded this to 200 observations per clinical condition, for a total required sample size of 400 patients. The interventions were short Shared Decision Making Videos that were developed as part of two training courses on shared decision making by investigators at Massachusetts General Hospital. For the analyses, study staff will examine the descriptives of the Shared Decision Making Process items for the two clinical conditions and orders. Study staff will examine rates of missing data to determine acceptability, and will examine descriptive results to see whether the scores span the range of total possible scores, are normally distributed, and whether there is evidence of floor or ceiling effects. Then, study staff examine discriminant validity of the measure by examining whether scores for the good videos are higher then for the poor videos. Further, study staff will examine concurrent validity with the alternative shared decision making measure.

Subjects were randomly assigned to one of the two clinical conditions (colorectal cancer screening or high cholesterol) then randomly assigned to view the two videos in one of two orders. Participants either viewed the good shared decision making video first (and the poor shared decision making video second) or the poor shared decision making video first (and the good shared decision making video second). Thus we used a 2 (clinical condition; colorectal cancer screening v. high cholesterol) x 2 (video; good v. poor) x 2 (order; good first v. poor first) factorial design.

Participants see videos of a conversation around taking medications (statins) for high cholesterol. Patients see good video first and poor video second.

Participants see videos of a conversation around screening for colorectal cancer. Patients see good video first and poor video second.

Participants see videos of a conversation around taking medications (statins) for high cholesterol. Patients see poor video first and good video second.

Participants see videos of a conversation around screening for colorectal cancer. Patients see poor video first and good video second.

The short video illustrated key components of a high quality shared decision making conversation between a doctor and patient actor around treatment of high cholesterol.

The short video illustrated a typical conversation between a doctor and patient actor that did not cover key aspects of share decision making around treatment of high cholesterol.

The short video illustrated key components of a high quality shared decision making conversation between a doctor and patient actor around screening for colorectal cancer.

The short video illustrated a typical conversation between a doctor and patient actor that did not cover key aspects of share decision making around screening for colorectal cancer.

The Shared Decision Making Process is a short patient-reported survey that measures the amount of shared decision making that occurs in an interaction. Scores range from 0-4 where higher values indicate a better shared decision making process occurred.

The SDM-Q9 is a 9-item patient reported measure of the amount of shared decision making that occurs in an interaction. Scores range from 0-45 where higher values indicate a better shared decision making process occurred.

Single item asking the participant who made the ultimate decision in the video. The categorical response options are 1) the patient made the decision, 2)the provider made the decision, or 3) both patient and provider made the decision together.

Inclusion Criteria: Speak English No prior diagnosis of colorectal cancer No history of heart attack No history of stroke Exclusion Criteria: None

The study team will create a complete, cleaned, de-identified copy of the final data set for each online field test. After the main analyses have been published, information for accessing the data will be made available on the Health Decision Sciences Center website and in publications of the data. Dr. Sepucha will share a de-identified data set with outside investigators at no cost, according to approved Massachusetts General Hospital/Partners policies for data sharing. Investigators from other sites will be able to request the data and will be required to complete a data use agreement that ensures that all local Institutional Review Board requirements are met before using the data, that they will not attempt to identify any data in the dataset, and that they will not share the data set with anyone outside their project team.