Table G-5. Key Question 3, First Outcome

Table G-5. Key Question 3, first outcome

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Akl et al., 20121 / G1: Strong and weak wording for or against guideline-supported behavior (ACCP):
• “we recommend”
• “we suggest”
• “we suggest...not” “we recommend …not”
G2: Strong and weak wording for or against guideline-supported behavior (NICE):
• “clinicians should”
• “clinicians might”
• “clinicians might not”
• “clinicians should not”
G3: Strong and weak recommendations for or against guideline-supported behavior (GRADE):
• “we recommend”
• “we conditionally recommend”
• “we conditionally recommend…not”
• “we recommend … not” / Health-related decisions or behavior (applicable for general public/patients)
Appropriate or inappropriate course of action based on choice of one of eight response options that correlate with the 4 strengths of recommendation and one of two decisionmaking styles (paternalism and shared decisionmaking). To be considered appropriate, the chosen response option had to correspond to the strength and direction of recommendation presented in a table in the document (regardless of decisionmaking style). / Immediate posttest
Self-report / G1: 114
G2: 111
G3: 118 / Appropriate Choices, by language category(%):
“Strong for”
G1: 7%
G2: 9%
G3: 7%
“Weak for”
G1: 77%
G2: 46%
G3: 61%
“Weak against”
G1: 32%
G2: 55%
G3: 64%
“Strong against”
G1: 49%
G2: 42%
G3: 51% / Difference: (p value)
“Strong for”
G1 vs. G2: -2%*
G1 vs. G3: 0%*
G2 vs. G3: 2%*
p=0.91
“Weak for”
G1 vs. G2: 31%*
G1 vs. G3: 16%*
G2 vs. G3: -15%*
p=0.003
“Weak against”
G1 vs. G2: -23%*
G1 vs. G3: -32%*
G2 vs. G3: -9%*
p=0.002
“Strong against”
G1 vs. G2: 7%*
G1 vs. G3: -2%*
G2 vs. G3: -9%*
p=0.60 / Chi-square and regression analysis (results of regression not reported other than to say that the analysis “confirmed the findings of the bivariate analyses regarding the association b/w wording and appropriate course of action.)
For regression analysis (to determine what predicts appropriate actions): disease scenario, demographic characteristics, educational characteristics, and familiarity with the grading approaches
Note: results in columns m and n are unadjusted.

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Han et al., 20113
(Experiment 1) / G1: Point estimate in text
G2: Point estimate as graph
G3: Range in text
G4: Range as graph
NOTE: Each format tested before and after delivery of information about the populations average risk of colon cancer (6% versus 2 to 10%). / Knowledge about the evidence
Perceived risk of getting colon cancer in your lifetime. Measured using 2 items that were averaged: 1) Based on these results, how would you describe your chances of developing colon cancer in your lifetime? very low to very high on 5-point likert; 2) If I received these results, I would feel that I’m going to get colon cancer. strongly disagree to strongly agree on 5 point likert. / Immediate posttest
Self-report / N=240
G1: 60
G2: 60
G3: 60
G4: 60 / Before comparative risk information
(Not reported in text, values estimated from Figure 3)
Text
G1: 1.7*
Graph
G2: 2.1*
Text
G3: 2.0*
Graph
G4: 1.6*
*estimated from Figure
After comparative risk information
(perceived risk change, ranging from 0-1, estimated from Figure 5)
Text
G1: 0.70*
Graph
G2: 0.15*
Text
G3: 0.12*
Graph
G4: 0.35*
*estimated from Figure / Before comparative risk information
No significant main effect of ambiguity or representational format on perceived risk.
G3-G1: +0.3, p=NS
G4-G2: -0.5, p=NS
Ambiguity × Representational Format was
significant (F (1, 231) = 9.08, p=0.003)
After comparative risk information
Significant main effect of ambiguity and representational format on perceived risk (F (1, 229) = 4.86,
p=0.03).
Significant 3-way interaction of ambiguity x format x comparative risk information (Wilks’s
l=:92, Fð2, 227) = 9.41, p<0.001).
G3-G1: -0.58, p=0.03
G4-G2: +0.20, p=0.03 / MANOVA and ANOVA
Dispositional optimism and numeracy

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Han et al., 20113
(Experiment 2) / G1: Range in text (precision)
G2: Range in text + solid bar graph (precision)
G3: Range in text + blurred bar graph (precision) / Knowledge about the evidence
Perceived risk of getting colon cancer in your lifetime. Measured using 2 items that were averaged: 1) Based on these results, how would you describe your chances of developing colon cancer in your lifetime? very low to very high on 5-point likert; 2) If I received these results, I would feel that I’m going to get colon cancer. Strongly disagree to strongly agree on 5 point likert. / Immediate posttest
Self-report / N=135
G1: 45
G2: 45
G3: 45 / NR / Effect size NR, NS. / MANOVA and ANOVA
Dispositional optimism and numeracy

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Longman 20124 / G1: Risk estimate as a point (precision)
G2: Risk estimate as a small range (precision)
G3: Risk estimate as a large range (precision) / Accuracy of risk perception
Assessed using 3 items. Point risk estimate/large range: (1) ‘‘If 100 people take Drug A, how many people will/what is the maximum number of people that will develop temporary facial skin discoloration?’’ (2)“‘If 100 people take Drug A, how many people will not/what is the maximum number of people that will not develop temporary facial skin discoloration?’’ (3)‘‘Another available medication for the treatment of severe acne is known as Drug B. In taking Drug B, 32 people out of 100/20 to 44 people out of 100 will develop temporary facial skin discoloration, compared to Drug A / Measured immediately after intervention
Self-report / N = 120 / # correctly responding to all 3 questions:
G1: 93.3%*
G2: 33.3%*
G3: 35%*
Proportion of participants who answered all 3 questions correctly, by risk information source.
Doctor as risk information source:
G1: 0.931 (0.859, 0.967)
G2: 0.324 (0.223, 0.445)
G3: 0.340 (0.236, 0.463)
Pharmaceutical company as risk information source:
G1: 0.936 (0.866, 0.970)
G2: 0.342 (0.237, 0.465)
G3: 0.359 (0.252, 0.482) / Accuracy of risk perception:
G2-G1:
% difference: -60*
p<0.001
OR: 0.036
95% CI: 0.016, 0.077
G3-G1:
% difference:
-58.3*
p<0.001
OR: 0.038
95% CI: 0.018, 0.083
G3-G2:
% difference: +1.7*
p=0.62
OR: 1.08
95% CI: 0.80, 1.44
No difference by source / Within subject correlation of responses
Chi-squared; logistic regression

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Longman 20124 (continued) / where (20 people out of 100/8 to 32 people out of 100) will develop temporary facial skin
discoloration. What is the difference/
maximum difference in the number of people who will develop temporary facial skin discoloration between Drug A and Drug B?’’

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
McCormack et al., 20115 / G1: control (no treatment control)
G2: Prostate-Only (Net benefit)
G3: Men’s Health (Net benefit in context of other more beneficial services) / Knowledge about the evidence
10 “demonstrated-knowledge” questions were used to calculate participant’s knowledge of the
contents of the interventions. A knowledge index score was computed and ranged from 0 to 10. (10 = all correct) / Baseline, 6 months, 12 months
Self-report / Overall N=376
G1: 122
G2: 89
G3: 165 / Mean knowledge scores at 6 months (range 0-10):
G1: 3.6
G2: 5.1
G3: 4.9
Mean knowledge scores at 12 months (range 0-10):
G1: 3.7
G2: 4.5
G3: 4.5
/ Mean knowledge scores at 6 months, absolute difference:
G3-G1: 1.3*, p=NR
G2-G1:1.5*, p=NR
G3-G2: 0.2*, p=NR
Mean knowledge score increase at 12 months from baseline, absolute difference:
G3-G1: +1.5* , p<0.001
G2-G1: +0.9* , p<0.05
/ GEE modeling
Education, marital status, prior PSA testing, health status, health literacy, race
Higher education, being married,
ever had a PSA test, excellent/very good self-reported
health (versus fair/poor), and greater health literacy were associated with higher knowledge scores. Lower knowledge scores were associated with being Black versus White.

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Perneger et., al., 20106 and 20117 / G1: control = minimal risk info, minimal benefit info
G2: minimal risk info, moderate benefit info
G3: minimal risk info, a lot of benefit info
G4: moderate risk info, minimal benefit info
G5: moderate risk info, moderate benefit info
G6: moderate risk info, a lot of benefit info
G7: a lot of risk info, minimal benefit info
G8: a lot of risk info, moderate benefit info
G9: a lot of risk info, a lot of benefit info
Each participant received varying information about the benefits and harms of a screening test for an unnamed cancer. / Health-related decisions or behavior (applicable for general public/patients)
Decision Evaluation= Combined score ranging from 0 (lowest) to 100 (highest) based on 3 items from a modified decisional conflict scale and 3 items from the satisfaction with decision scale / Given in postal survey
Self-report / N=2333 / DECISION SATISFACTION=
G1:85.9 (17.4)
G2: 86.4 (14.4)
G3: 86.1 (16.0)
G4: 79.2 (19.5)
G5: 79.4 (18.1)
G6: 81.8 (17.0)
G7: 79.3 (17.6)
G8: 81.1 (18.8)
G9: 83.2 (17.2)
Mean decision satisfaction:
Minimal risk, aggregate benefit: 85.9
Mod risk, aggregate benefit: 80.4
Lot of risk, aggregate benefit: 81.2
Minimal benefit, aggregate risk: 81,4
Moderate benefit, aggregate risk: 82.5
Lot of benefit, aggregate risk: 83.6:
% Test Refusal:
Minimal risk, aggregate benefit: 8.8
Minimal benefit, aggregate risk: 16.6 / Adjusted absolute difference in Decision Satisfaction:
More than minimal vs. minimal risk: -5.1 (-6.6, -3.6)
More than minimal vs. minimal benefit: 1.1 (-0.4 to 3.6)
OR for test refusal (compared to minimal information):
Minimal risk info: 1.0
Moderate risk info (FP): 2.5 (1.8 to 3.4)
Lot of risk info (FP + FN): 3.0 (2.2 to 4.2)
Minimal benefit info: 1.0
Mod benefit info (survival): 1.0 (0.7 to 1.3)
Level of benefit info (survival and reassurance): 1.0 (0.7 to 1.3) / Two-way ANOVA
Age, screening in past 3 years, desire for information

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Schwartz et al., 20118 / G1: control: No explanation of heart drug or heartburn drug
G2: Nondirective explanation of heart drug or heartburn drug
G3: Directive explanation of heart drug or heartburn drug
Each participant sequentially randomized to 1 0f 3 groups for heart drug and then for heartburn drug / Health-related decisions or behavior (applicable for general public/patients)
Choice of the better drug (more distal outcomes) / Immediately following intervention
Self-report / Overall N=2944
G1: 981
G2: 981
G3: 982 / Heart drug:
G1: 59%
G2: 71%
G3: 71% / Heart drug:
G1-G2 Difference: 12 %
95% CI: 7-18
p=NR
G1-G3 Difference: 12 %
95% CI: 7-18
p=NR / Unclear- use the SVY series of commands- and postestimation commands for CI
None

Table G-5. Key question 3 first outcome (continued)

Author,
Year / Groups / Outcome #1, Exact Measure Used / Timing of Measurement,
Data Source / N Analyzed for this Outcome / Results by Group / Differences in Groups / Statistical Methods Used, Covariates Controlled for in Analysis
Sheridan 20129 / G1: Educational video on highway safety (control)
G2: Video-based decision aid and coaching session for patients (net benefit)
Combined analysis of two trials in which G2 includes prostate only information or prostate information framed in the context of other men’s health services. / Knowledge about the evidence:
True-false questions
highlighting the benign natural history of most prostate cancers and the high likelihood of side effects with treatments delivered for prostate cancer detected by PSA screening: 1) “Some men can live long lives with prostate cancer,” 2) “most men diagnosed with prostate cancer die of something else,” 3) “problems with sexual function is a common side effect of prostate cancer treatments,” and 4) “problems with urination is a common side effect of prostate cancer treatments.” / Measured immediately after intervention
Self-report / Total N=128
G1: 70
G2: 58 / % Men having key knowledge
G1: 13%
G2: 47% / G2-G1:
Absolute difference: +34%
95% CI: 19% to 50%
Fully adjusted RR: 4.28
95% CI: 2.30 to 6.45
p=NR / Combined data from two randomized controlled trials so adjusted for random effects of physician and practice (Fully adjusted RR)
Mixed effects logistic regression

Abbreviations: ACCP = American College of Clinical Pharmacy; ANOVA = ANalysis Of Variance; b/t = between; DX = diagnosis; FN=false negative; FN=false positive; G = group; GEE = generalized estimating equations method; GRADE = Grading of Recommendations Assessment, Development and Evaluation; M=Mean; MANOVA = Multivariate analysis of variance; NA = not applicable; NICE = National Institute for Health and Clinical Excellence; NR = not reported; NS=not significant; OR = odds ratio; PSA = prostate-specific antigen; SD = standard deviation; SVY = survey; vs. = versus