TABLE 3.

Quantitative results

Strategy	Study	Comparators	Results
Numerical formats
Percentages	Powers et al. (2011)	Risk factor education	Agreed information presented more clearly than risk factor education only (57% vs. 29% p = 0.008*1). More helpful in making decisions (47% vs. 31% p = 0.1*1). Less decisional conflict over risk reduction methods (p = 0.003)*2. No differences in health behaviours, blood pressure, medication adherence or smoking. Perceived risk declined at 3 months (p = 0.053)*1. *1 conditional logistic regression (Fisher exact test when data sparse) *2 t test
Risk ratios	Fair et al. (2008)	Percentages	Increased risk perceptions (p < 0.001)*1. Increased intentions to make lifestyle changes (p = 0.047)*2. Increased levels of worry (p = 0.0004)*2 and disturbance (p = 0.001)*2. *1 logistic regression *2 ANOVA
Heart age	Bonner et al. (2015)	Percentages	Viewed results as less credible (p < 0.001*) and had less of a positive emotional response (p < 0.001*). No difference in intentions to change lifestyle (reduce smoking p = 0.67, improve diet p = 0.47, improve physical activity p = 0.72, improve diet p = 0.72 or see a GP for further assessment p = 0.35). At 2 weeks, 32% of participants could recall heart age versus 16% for risk percentage. Heart age recall decreased at 2 weeks (32%) compared with immediately postintervention (65%). Participants with a younger heart age are more likely to recall risk (80% heart age vs. 63% percentage p = 0.009*) than those with an older heart age (both 61%; p > 0.999*). No difference in format and risk perceptions (p = 0.071*). *Mann‐Whitney test
	Lopez‐Gonzalez et al. (2015)	Percentages and control (no risk score)	Reduction in smoking (1.8% heart age vs. 0.4% percentage) and weight (−0.8 kg heart age vs. −0.2 kg percentages) at 12 weeks. At 12 months Framingham risk scores increased in the control group (+0.24%) and decreased in the risk percentage group (−0.2%) and the heart age group (−0.4%).
	Damman et al. (2018)	Percentages and risk ratios	Heart age increased intentions to be more physically active (F = 6.29; p = 0.13*) and to visit a GP for further screening (F = 5.23; p = 0.023*). Improved recall of verbal labels (F = 7.1; p = 0.008*). *ANOVA
Graphical displays
Icon arrays	Ruiz et al. (2013)	Percentages	Risk recall lower in the icon array group (p < 0.001*1). No difference in long‐term risk recall (p = 0.10*1). No difference in risk understanding (p = 0.31*1). No differences in perceptions of seriousness (p = 0.85*2), intention to change lifestyle (p = 0.15*2), intentions to follow medical treatment (p = 0.65*2) or overall satisfaction (p = 0.09*2) No differences in clarity (p = 0.13*2) or helpfulness of information (p = 0.43*2) *1 Chi‐squared test. *2 ANOVA
	Zikmund‐Fisher et al.  (2014)	Comparison of icons	Risk recall highest in restroom icons and photographs (both 81%). Risk recall lowest in blocks and faces (both 71%).
	Witteman et al. (2014)	Random sequencing	Animated randomness associated with better alignment between risk estimates and risk perceptions (F1,3576 = 6.12, p = 0.01*) but reduced lifestyle intention scores (F1,3572 = 11.1, p = 0.01*). Improved risk recall in low‐risk participants (F1,3544 = 7.06, p = 0.01*). *ANOVA
Icon arrays and bar graphs	French et al. (2004)	Numbers	Participants who received bar graph or icon array had lower levels of worry (F1,313–8.74; p < 0.01*) but were not more reassured. *ANOVA
	Adarkwah et al. (2019)	N/A each other	No difference in recall of interventions agreed upon with general practitioners at 3 months between the icon array group (1.04 ± 0.44) and the bar graph group (1.05 ± 0.39). Risk perception highest in bar graph group at 3 months (p = 0.032)*. Between baseline and 3 months, risk perceptions decreased in the bar graph group (p = 0.02)*. There was no change in the icon array group. *Student’s t test
	Navar et al. (2018)	N/A each other	22% of participants shown icon array reported a 10‐year risk of 15% to be high compared with 36% shown no icon and 35% shown a bar graph (p < 0.001)*. 5%–6% more participants were willing to take preventive treatment when shown bar graph compared with icon array *Two‐tailed test
Avatars	Witteman et al. (2014)	Icons and frequencies	Improved risk perceptions overall (F1,13,576 = 4.61, p = 0.03). Improved alignment between risk estimates and intentions to see a doctor (F1,356 = 6.38, p = 0.01) *nested factorial ANOVA
	Ruiz et al. (2016)	Voice and text	Improved intentions to change lifestyle (p = 0.3). No differences in risk recall or understanding (χ2 = 1.1. p = 0.57*). *ANOVA
Qualitative information	Damman et al. (2018)	Infographics	More correct answers for the recall of risk causes when the text was used.
Infographics	Damman et al. (2018)	Qualitative information	Infographics negatively influenced recall of risk causes (F = 7.73; p = 0.006*). Information evaluated more negatively with infographics (F = 8.83; p = 0.003*). Infographics negatively influenced subjective risk comprehension (F = 10.14; p = 0.002*). 67% of participants with adequate health literacy considered infographic information useable versus 54% with inadequate health literacy. Figures rose to 73% and 76% respectively when no graphics were used. *ANOVA
Timeframes
Lifetime risk	Fair et al. (2008)	10 year risk	Lifetime risk led to higher incidences of worry (p = 0.23)*, disturbance (p = 0.17)* and less reassurance (p = 0.16)*. *ANOVA
5, 10, 15 & 20 year risk	Frileux et al. (2004)	N/A each other	Shorter timeframes led to higher intentions to adopt preventive behaviours (F3,414 = 229.33: p < 0.00001). *ANOVA with repeated measures
Genetic Risk Scores	Domenech et al. (2016)	No risk score	18 (58.1%) participants in the genetic risk score group had hypertension control compared with 14 (38.9%) in the control group (p = 0.008)* at 16 weeks *Pearson chi‐square test
	Knowles et al. (2017)	Framingham risk score	No difference in low‐density lipoprotein cholesterol at 3 months (p = 0.59)* or at 6 months (p = 0.75)*. The genetic risk score group reported moderate weight loss in high‐risk participants (−2.3 kg ± 3 vs. 0.0 kg ± 3, p = 0.002*). *Hodges‐Lehmann statistic
Cardiovascular imaging
Coronary artery calcium scores	Johnson et al. (2015)	No comparison	68% of participants could accurately identify their risk score based on their coronary artery calcium score. 24% of high‐risk participants identified that they were in the high‐risk group. There were improvements in health‐promoting behaviour (p < 0.001*). *ANOVA
	Orakzai et al. (2008)	No comparison	Initiating aspirin therapy, increasing exercise and modifying diet increased with increasing coronary artery calcium scores (all p < 0.001* for trends). 56% high‐risk participants modified their diet and 67% increased exercise. *t test and Mann‐Whitney rank‐sum test
	Kalia et al. (2006)	No comparison	Statin compliance at 3 (±2) years was highest in the group with the highest coronary artery calcium scores (91%) and lowest in the low‐risk group (44%). Dietary modifications increased from 41% to 64%. 71% stopped smoking. 65% increased exercise.
Carotid ultrasounds	Näslund et al. (2019)	Percentages	Largest decrease in Framingham risk scores in the carotid visualization group (−0∙58 [95% CI –0∙86 to −0∙30] vs. 0∙35 [0∙08–0∙63]). Larger reduction in low‐density lipoprotein cholesterol in the carotid visualization group (0.3 mmol/L vs. 0.12 mmol/L). Larger decrease in smoking in carotid visualization group (1.25% vs. 1.01%).
	Korcarz et al. (2008)	No comparison	Higher levels of plaque led to increased intentions to take cholesterol‐lowering medication (p = 0.02*) and an increased likelihood of having heart disease (p = 0.004*) and developing heart disease (p < 0.001*). Normal scans also lead to increased motivation to exercise (p = 0.003*). *multiple linear regression model