A 4 item Breast Cancer Treatment Deliberation Scale asked how deliberative patients were in making their treatment decision. The items asked how often they performed deliberative activities with answers of 'not at all'/'a little'/'somewhat'/'quite a bit'/'a lot'. Each response was assigned a value of 1('not at all') to 5('a lot), with a high value representing a greater amount of deliberation. The values of all 4 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (not at all) to 5 (a lot). The deliberation scale was then modeled using linear mixed models to determine the effect of the intervention on patient deliberation.

A 5 item subjective decision quality scale measured how satisfied patients were with their treatment decision. Patients were asked how well they agreed with 5 statements, with responses of 'not at all'/'a little bit'/'somewhat'/'quite a bit'/'very much'. All responses were assigned values from 1 to 5, with higher values reflecting greater decision satisfaction. Two of the statements reflected satisfaction with the decision and were coded as 1 for 'not at all' through 5 for 'very much'. The other three statements reflected dissatisfaction with the decision and were coded as 5 for 'not at all' through 1 for 'very much' The values of all 5 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (not at all) to 5 (very much). The subjective decision quality scale was modeled by linear mixed model to determine the effect of intervention on patient response.

Self-reported knowledge about systemic treatment using a 5 item Breast Cancer Knowledge Measure (adapted) consisting of 5 questions to test patients' knowledge about systemic treatment options for breast cancer. We compared the number of patients who gave correct answers to at least 80% of the questions between two groups. We used both unadjusted and adjusted generalized linear mixed models with logit as the link function.

A 10-item decision preparedness scale asked whether the web intervention helped patients prepare for their treatment decision. Each item asked about a different aspect of decision preparation, with responses of 'not at all'/'a little'/'somewhat'/'quite a bit'/'a great deal'. Each response was assigned a value of 1('not at all') to 5('a great deal'), with a high value representing a greater amount of preparedness. The values of all 10 items were combined using the arithmetic mean, to create a standardized scale ranging from 1(not at all) to 5 (a great deal). The decision making scale was modeled by linear mixed model to determine the effect intervention on patient preparation for decision making.

A 4 item subjective decision quality scale measured how satisfied patients were with their chemotherapy treatment decision. Patients were asked to rate the amounts of information, involvement, time, and overall satisfaction associated with their chemotherapy decisions. Possible responses ranged from 'not enough' to 'just right' to 'too much'. All responses were assigned values from 1 to 5, with a response of 'just right' coded as 5 points, and both 'not enough' and 'too much' coded as 1 point. The values of all 4 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (low decision quality) to 5 (high decision quality). The subjective decision quality scale was used as an outcome in linear mixed models to determine the effect of intervention on patient decision quality.

Total of all reporting groups

Self reported values were evaluated using a 5 item question set adapted from Decision Quality Instrument. The questions determined patient desire for outcomes such as keeping their natural breast and avoiding radiation on a scale from 0 to 10. Patients were classified as values-concordant if their actual treatment aligned with their values score predicted treatment and otherwise were classified as non-concordant. The binary values-concordance variable was modeled as a function of intervention effect using both unadjusted and adjusted logistic mixed model regression.

A 12 item decision preparedness scale asked whether the web intervention helped patients prepare for their treatment decision. Each item asked about a diefferent aspect of decision preparation, with responses of 'not at all'/'a little'/'somewhat'/'quite a bit'/'a great deal'. Each response was assigned a value of 1('not at all') to 5('a great deal'), with a high value representing a greater amount of preparedness. The values of all 12 items were combined using the arithmetic mean, to create a standardized scale ranging from 1(not at all) to 5 (a great deal). The decision making scale was modeled by linear mixed model to determine the effect intervention on patient preparation for decision making.

A 4 item Breast Cancer Treatment Deliberation Scale asked how deliberative patients were in making their treatment decision. The items asked how often they performed deliberative activities with answers of 'not at all'/'a little'/'somewhat'/'quite a bit'/'a lot'. Each response was assigned a value of 1('not at all') to 5('a lot), with a high value representing a greater amount of deliberation. The values of all 4 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (not at all) to 5 (a lot). The deliberation scale was then modeled using linear mixed models to determine the effect of the intervention on patient deliberation.

participants

A 5 item subjective decision quality scale measured how satisfied patients were with their treatment decision. Patients were asked how well they agreed with 5 statements, with responses of 'not at all'/'a little bit'/'somewhat'/'quite a bit'/'very much'. All responses were assigned values from 1 to 5, with higher values reflecting greater decision satisfaction. Two of the statements reflected satisfaction with the decision and were coded as 1 for 'not at all' through 5 for 'very much'. The other three statements reflected dissatisfaction with the decision and were coded as 5 for 'not at all' through 1 for 'very much' The values of all 5 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (not at all) to 5 (very much). The subjective decision quality scale was modeled by linear mixed model to determine the effect of intervention on patient response.

Self-reported knowledge about systemic treatment using a 5 item Breast Cancer Knowledge Measure (adapted) consisting of 5 questions to test patients' knowledge about systemic treatment options for breast cancer. We compared the number of patients who gave correct answers to at least 80% of the questions between two groups. We used both unadjusted and adjusted generalized linear mixed models with logit as the link function.

The numbers analyzed for these secondary outcomes were taken from the study's systemic portion of the study which took place 9 months after enrollment at which point response rate dropped and therefore numbers are less than those for primary outcomes which were done 4-5 weeks after enrollment.

A 10-item decision preparedness scale asked whether the web intervention helped patients prepare for their treatment decision. Each item asked about a different aspect of decision preparation, with responses of 'not at all'/'a little'/'somewhat'/'quite a bit'/'a great deal'. Each response was assigned a value of 1('not at all') to 5('a great deal'), with a high value representing a greater amount of preparedness. The values of all 10 items were combined using the arithmetic mean, to create a standardized scale ranging from 1(not at all) to 5 (a great deal). The decision making scale was modeled by linear mixed model to determine the effect intervention on patient preparation for decision making.

A 4 item subjective decision quality scale measured how satisfied patients were with their chemotherapy treatment decision. Patients were asked to rate the amounts of information, involvement, time, and overall satisfaction associated with their chemotherapy decisions. Possible responses ranged from 'not enough' to 'just right' to 'too much'. All responses were assigned values from 1 to 5, with a response of 'just right' coded as 5 points, and both 'not enough' and 'too much' coded as 1 point. The values of all 4 items were combined using the arithmetic mean, to create a standardized scale ranging from 1 (low decision quality) to 5 (high decision quality). The subjective decision quality scale was used as an outcome in linear mixed models to determine the effect of intervention on patient decision quality.

Static version (non-interactive) version of CanSORT decision tool

Static version of CanSORT tool