Table 1. Characteristics of studies included in the review.

Reference	Study design	Setting	Sample	Intervention	Health outcome	Results	Effect on SDH
Stanczyk et al. (2013) [14]	RCT	Netherlands	Smokers >16 years of age who were motivated to quit within the following six months - N = 139 - Mean age: 47,39 years - Educational level n (%) -Low (n = 68, 48.9%) -High (n = 71, 51.1%)	Web-based tailored intervention: (n = 74 to text condition and n = 65 to video condition) completed the first session of the web-based tailored intervention and were subsequently asked to fill out a questionnaire assessing message-processing mechanisms and future adherence.	- Processing mechanism - Future adherence	- No interaction effects between delivery strategies (video vs. text) and educational level on message-processing mechanisms and future adherence. - Lower educated participants showed higher attention (F1,138 = 3.97; P = .05) and processing levels (F1,138 = 4.58; P = .04). - Lower educated participants were more inclined to visit the computer-tailored intervention website again (F1,138 = 4.43; P = .04).	Moderate
Cantrell et al. (2013) [19]	Community intervention	USA	U.S. adult smokers from diverse racial/ethnic and SES groups recruited from two online research panels - N = 3,371(n = 3,371) - Mean age: 44,17 years	Web-based experimental study to view either pictorial health warning labels (n = 1,706) or text-only warnings (n = 1,665) found on tobacco packaging. Participants viewed the labels and reported their reactions.	- Intention to quit smoking. - Salience - Perceived impact - Credibility	- Pictorial health-warning message gave significant stronger reactions for each outcome salience. - No significant results were found for interactions between condition and education. - Intention to quit: mean % for low education 35.5% for text and 39% for text+pictorial vs. 47.4% and 49.8% for high education.	Weak - Risk exacerbating inequalities in health between educational groups.
Wendel-Vos et al. (2009) [20]	Community intervention	Netherlands	Men and women aged 31–70 years living in and around certain areas in the Netherland. - N = 3114	Community-based health intervention (Hartslag Limburg) aimed at preventing cardiovascular disease by comparing 5-year mean changes in lifestyle factors (1998–2003), between subjects from the intervention area (n = 2,356) and the control area (n = 758) and between men and women and for those with a low, moderate and high education. 50% of intervention took part in deprived areas	- Energy intake in terms of fat intake - Time spent on leisure-time - Walking - Bicycling - Sports - Smoking behavior	Among those with a low education level, a significant difference (p≤0.05) between the intervention region and the control region were found for: - Fat intake: −3.0 (g/d) for low educational level vs. −1.7 (g/d) for high educational level. - Time spent walking +2.3 (hrs/wk) for low educational level vs. +0.9 for high educational level. - Time spent bicycling +0.6 (hrs/wk) vs. −0.3 for high education (p≤0,05). - Smoking: among intervention 6,2% with low educational level quit smoking vs. 6.1% among high educational level	Weak - Too small differences to draw any conclusions.
Øverby et al. (2012) [21]	Community level intervention	Norway	Sixth- and seventh-grade pupils from 27 Norwegian elementary schools completed a questionnaire in 2001 (n = 1,488) together with sixth- and seventh-grade pupils from the same schools that completed the same questionnaire in 2008 (n = 1,339).	The project “Fruits and Vegetables Make the Marks” a national free school fruit program 2001–2008 with focus on SES	- Consumption of unhealthy snacks (soda, candy and potato chips)	The frequency of unhealthy snack consumption decreased and was more pronounced in schools that had been included in the national free school fruit program (−2.8 times/week). Effect was significant (p = 0,004) in reducing the frequency of unhealthy snack consumption in children of parents without higher education (−3.8 times/week).	Moderate
Cameron et al. (2014) [15]	Cluster-randomized controlled trial	Australia	542 mother/infant pairs from existing 62 first-time parent groups. - N = 389 first-time mothers and infants tested for maternal education - Mean 32.3 years for mothers	The Melbourne Infant, Feeding, Activity and Nutrition Trial (InFANT) 2008–2010 involving 6 × 2-hour dietarian-delivered sessions, DVD and written resources from infant age 4–15 months was assessed by comparing an intervention group (n = 191) with a control group (n = 198)	- Infant diet (3 × 24h diet recalls) - Physical activity (accelerometry - Television viewing - Body mass index, BMI	- BMI change −0.07 (p = 0.49) for low educational level vs. 0.04 for high educational level - Vegetable intake (g/d) 14.79 (p = 0.23) for low educational level vs. 8.65 for high educational level - Water intake: 65,35 g (p = 0,02) for low educational level vs. −6.16 for high educational level - Non-core drinks intake (g/d) 5.3 (p = 0.49) for low educational level vs. −7.67 for high educational level - Sweet snacks intake (g/d) −1.55 (p = 0.43) for low educational level vs. −5.22 for high educational level - Savory snacks intake (g/d) −1.17 (p = 0.35) for low educational level vs. −0.83 for high educational level - Television viewing (min/d) −19.43 (p = 0.01) for low educational level vs. −13.39 for high educational level - Physical activity (min/d) 1.78 (p = 0.77) for low educational level vs. −5.87 for high educational level	Weak - Risk of exacerbating inequalities in health between educational groups.
Vander Ploeg et al. (2014) [22]	Quasi-experimental trial	Canada	Grade five school students (n = 412) from 10 invited school in 2009, with follow-up in 2011 (n = 339) compared to 20 control schools (n = 845 students) with follow up in 2011 (n = 680)	Comprehensive School Health (CSH) program implemented in schools located in socioeconomically disadvantaged neighborhoods. The Alberta Project Promoting active Learning and healthy Eating in Schools (APPLE Schools). SES was determined from parent self-report. Low-active, active, and high-active children were defined according to step-count tertiles. Grade five students included in analysis in 2009 (n = 198) with follow up in 2011 (n = 196) were compared with controls in 2009 (n = 450) and at follow-up in 2011 (n = 300)	- Physical activity	From 2009 to 2011, children within the low-education groups from intervention schools experienced increases in physical activity (+23,8%) and 23.6% greater than children within these groups from comparison schools, respectively	Moderate
Van der Waerden et al. (2013) [16]	RCT	Netherlands	Low SES women (20–55 years) with elevated stress or depressive symptoms levels - N = 161 - Mean age: 43.9 years	Study subjects randomly assigned to the combined exercise/psycho-education (EP, n = 55), exercise only (E, n = 46), or a waiting list control condition (WLC, n = 48) with postponed intervention four months later. Follow up at 2, 6 and 12 months	- Depressive symptoms (CES-D) - Perceived stress symptoms (PSS)	Both interventions (E + EP) among women with the lowest educational level had lower PSS at post-test than women from control condition: - E = -4,13 (p = 0.02, effect size = 0.35) - EP = -4,12 (p = 0.024, effect size = 0.53)	Moderate - Important that the intervention target those with low educational level
DeSocio et al. (2013) [17]	RCT	USA	Unmarried adolescent (n = 429) mothers <19 years of age were randomly assigned to an intervention and a control group. The participants met at least two of three criteria for social disadvantage: (a) Unmarried (b) <12 years of education (c) Unemployed - Mean age: 18 years (range 12–33)	The Memphis New Mothers’ Home Visitation Program 1990–1991. The intervention group (n = 132) received free transportation and developmental screening plus intensive nurse home visitation through their infants’ second birthdays, in total 57 nurse visits.The control group (n = 297) received both free transportation and developmental screening/referral for their children at 6, 12, and 24 months of age.	- Self-agency	Adolescent mothers with lower cognitive ability who were behind their age-appropriate grade level in school made the greatest self-agency gains (p <0,003)	Weak - Could however be a way to reach those with low educational level
Kreuter et al. (2010) [18].	RCT	USA	African American women (n = 489) age ≥40 and never diagnosed with breast cancer were recruited from low-income neighborhoods. - Mean age: 61.1 years	Recruited women were randomly assigned to narrative video (n = 244) or informational video (n = 245) Telephone follow-up interviews at 3 and 6 months post-baseline	- Use of mammography	Use of mammography at 6-month follow-up did not differ for the narrative vs. informational groups, but one exception was among women with low educational level (65% vs. 32%, p < .01). For women with high educational level this was not seen (42,1% vs. 43,1%, p = 0,91)	Weak - Could however be a way to reach those with low educational level