Contributions to the literature
-
Summaries are often used to communicate evidence synthesis findings; however, there is no consensus on the most effective way to communicate or what works for different audiences.
-
This review explored the effectiveness and acceptability of different summary formats for different audiences.
-
We identified recommendations to help evidence synthesis producers better communicate to different audiences. These include guidance on formatting, tailoring content for end users, instilling trust in the work, establishing and helping knowledge requirements, detailing the quality of included studies, and properly contextualising findings.
-
Results can guide the creation of summary formats better tailored to end user’s needs).
Background
Methods
Study designs and eligibility criteria
Search strategy and study selection
Data extraction and appraisal of studies
Analysis and synthesis of findings
Results
Search results
Characteristics of included studies
Author (year, country) | Collection method | Participants | Format(s) evaluated | Topics |
---|---|---|---|---|
Babatunde (2018, England) [40] | Semi-structured interviews (and questionnaires)b | N = 21. Clinicians (11), researchers (5), epidemiologists (3), health service/trial managers (2) | Evidence flowers and summary table | Musculoskeletal conditions |
Buljan (2020, Croatia) [41] | Focus groups | N = 20. Patient advocates (9), doctors (4), medical students (7)a | Plain language summary, infographic, scientific abstract | Breech presentation |
Busert (2018, International) [42] | Semi-structured interviews | N = 18. Public-health decision-makers | 4-page summary with Summary of findings (SoF) table and Grading of Recommendations, Assessment, Development and Evaluations (GRADE) ratings | Food, alcohol, and tobacco portion/packaging |
Dobbins (2004, Canada) [43] | Focus groups | N = 46. Medical officers (7), programme managers/coordinators (25), decision-makers (14) | Summary statement | Tobacco control |
Semi-structured interviews | N = 8. Guideline developers (3), healthcare providers (3), research funders (1), health insurers (1) | ‘Rapid products’ (evidence inventory, rapid response, rapid review) | Venous thromboembolism | |
Hartling (2018, Canada) [46] | Semi-structured interviews | N = 6. Decision-makers | 3-page summary | Youth mental health |
Marquez (2018, Canada) [47] | Semi-structured interviews (and survey) b Semi-structured interviews | N = 11. Healthcare managers (5), policymakers (6) N = 12. Healthcare managers (5), policymakers (7) | Summary prototype | Healthcare management/services |
Mustafa (2015, International) [48] | Semi-structured interviews, workshop discussions | N = 20. Researchers, health professionals, guideline developers | 3 formatsc of GRADE evidence tables | Diagnostic test accuracy reviews |
Newbery (2013, USA) [49] | Focus groups (with questionnaires) Individual feedback (and questionnaires) | N = 15. Health insurer (2), insurance/former policymaker (2), clinicians (3), researchers (2), governmental research directors (2), research consultant (1) N = 3. Community physicians | 7 differently formatted executive summaries | Acute otis media |
Opiyo (2013, Kenya) [50] | Semi-structured interviews | N = 16. Multidisciplinary guideline development group members | SoF tables, graded-entry summary, normal systematic review | Newborn care, hand hygiene |
Perrier (2014, Canada) [51] | Focus groups | N = 10. Family physicians | Case-based and evidence-based prototypes | Rosacea |
Perrier (2014, Canada) [52] | Focus groups | N = 32. Primary care physicians | Two summary prototypes | Rosacea |
Rosenbaum (2011, International) [53] | Semi-structured interviews | N = 18. Policymakers and managers | Short summaries | Healthcare management/services |
Rosenbaum (2010, International) [54] | Semi-structured interviews (and workshops) | N = 21. Health professionals, researchers | SoF tables | Deep vein thrombosis |
Semi-structured interviews | N = 6. Department director (1), health system experts (4), guideline developers (2) | MAGICapp, Tableau | Chronic pain | |
Steele (2021, England) [57] | Semi-structured interviews | N = 7. Mental health clinicians | One-page summary, full systematic review | Mental health |
Yepes-Nunez (2019, International) [58] | Semi-structured interviews | N = 32. Methodologists (21), meta-analysis users (5), clinicians (6) | SoF tables | Network meta analyses |
Author (year, country) | Participants | Intervention and comparators | Primary (secondary) outcomes and operationalization (number of questions, type, scales) | Focus |
---|---|---|---|---|
Buljan (2018, Croatia) [59] | N = 163 (eligible across trials) 99 patient representatives, 64 doctors (171 students)a | Infographic, plain language summary, scientific abstract (doctors only) | Understanding/knowledge (10, open ended) Reading experience (5, summative, 10-point scale) User-friendliness (5, summative, 10-point scale) | Breech presentation |
Carrasco-Labra (2016, International) [60] | N = 284 Health professionals (122), guideline developers (42), researchers (120) | 2 versions (1 existing, 1 alternate) of Grading of Recommendations, Assessment, Development and Evaluations (GRADE), Summary of Findings (SoF) tables | Understanding (7, multiple choice, 5-point scale) Accessibility of information (3, 7-point scale; 1, 5-point scale) Satisfaction (6, yes/no) Preference (1, 7-point scale) | Paediatric probiotics |
Opiyo (2013, Kenya) [50] | N = 70 Paediatricians (32), medical/nursing officers (18), researchers (5), healthcare trainers (5), governmental/clinical officers (7), pharmacists (2), administrator (1) | 3 different topic ‘evidence packs’ 1. Normal systematic review (SR) 2. SR plus SoF tables 3. Graded-entry SR | Understanding (2 per format, 3-point scale) Composite endpoint (1, 5-point scale) Clarity (1 per format, 3-point scale) Accessibility (2 per format, 5-point scale) | Hand hygiene, newborn care, newborn feeding regimens |
Rosenbaum (2010, International) [61] | N = 72 (RCT1) Healthcare professionals N = 33 (RCT2) Staff from Cochrane entities | Normal Cochrane review (CR) with no SoF table CR with SoF table (limited formatting) CR with SoF table (full formatting) Normal Cochrane review (CR) with no SoF table CR with SoF table (revised) | User satisfaction (unclear, multiple choice) Perceived understanding and ease of use (7, 8-point scale) Understanding (4, unclear) Time spent finding key results (1, continuous) | Deep vein thrombosis |
Author (year) | Brief description of intervention | Intervention location, mode of delivery, time limit | Materials and components |
---|---|---|---|
Buljan (2018, Croatia) [59] | Infographic, plain language summary, scientific abstract | Online, electronic, none | Participants read one of the summary formats, followed by the survey (first a numeracy test with sufficient delay for the knowledge test). Patient representatives were presented with the infographic or PLS, and doctors were presented all three formats |
Carrasco-Labra (2016, International) [60] | Summary of findings table | Online, electronic, 25 mi | Participants were exposed to one table containing either the new or current format, and the outcomes understanding, accessibility of information, satisfaction, and preference were assessed. Participants were then shown the table to which they were not initially allocated, and their preference was assessed |
Opiyo (2013, Kenya) [50] | Evidence summaries in 3 formats (A, B, C) | In-person workshop, paper, 45 min | Summaries were delivered to participants as prereading materials 1 month before the workshop. Participants completed questionnaires on the first day of the guideline development workshop before the panel discussions about guidance recommendations |
Rosenbaum (2010, International) [61] | RCT1: Normal Cochrane review (CR) with no SoF table CR with SoF table (limited formatting) CR with SoF table (full formatting) RCT2: Normal Cochrane review (CR) with no SoF table CR with SoF table (revised) | RCT1: In person workshop, unclear, unclear RCT2: In-person workshop, unclear, unclear | Participants first answered a questionnaire based on the version of the review they had received. Then, all participants were shown both formatting versions of the SoF tables and were instructed to answer a final set of questions measuring their preferences and attitudes about the inclusion summary of findings table in reviews |
Quality appraisal
Quantitative analysis
Knowledge or understanding
Author (year) interventions | Primary (secondary) outcome measures | Results | |||
---|---|---|---|---|---|
Understanding/knowledge | Satisfaction/reading experience | Accessibility/ease of use | Preference | ||
Buljan (2018) [59] • Infographic • PLS • SA (doctors only) | Understanding/knowledge (max score = 10) Reading experience (max score = 50) User-friendliness (max score = 50) | Patients (n = 99), median score (95% CI) Infographic: 7.0 (6.0–7.0) PLS: 7.0 (6.0–7.0) P = 0.511 Doctors (n = 64), median score (95% CI) Infographic: 8.0 (6.0–8.0) PLS: 8.0 (7.0–9.0) SA: 8.0 (5.9–9.0) P = 0.611 Significant predictor of knowledge score • Patients only: awareness of Cochrane SRs (OR 5.3; 95% CI: 1.7–16.6), 13.4% of variance | Reading experience Patients (n = 99), median score (95% CI) Infographic: 33.0 (28.0–36.0) PLS: 22.5 (19.0–27.4) P < 0.001 Doctors (n = 64), median score (95% CI) Infographic: 37.0 (26.8–41.3) PLS: 32.0 (30.0–39.9) SA: 24.0 (21.3–27.2) P = 0.002 | User-friendliness Patients (n = 99), median score (95% CI) Infographic: 30.0 (25.5–34.5) PLS: 21.0 (19.0–25.0) P < 0.001 Doctors (n = 64), median score (95% CI) Infographic: 36.0 (30.9–40.0) PLS: 29.0 (26.8–36.2) SA: 25.0 (23.5–27.2) P = 0.003 | Not reported |
Carrasco-Labra (2016) [60] • Existing GRADE SoF table • Alternate GRADE SoF tables | Understanding (7 multiple-choice questions on 5-point scale, analysed at question level) Accessibility of information (5-point scale), satisfaction (6 yes/no questions analysed at question level), preference (7-point scale) | 1. 4/7 items risk difference (RD, 95% CI) in favour of alternate SoF tables 2. Understanding of quality of evidence and treatment effect RD: 62% (52–71), P < 0.001 3. Ability to determine risk difference RD: 63% (54.6–71), P < 0.001 4. Ability to quantify risk RD: 6% (0.1–13.3), P = 0.06 5. Understanding of quality of evidence RD: 7% (0.1–12.4), P = 0.06 6. 3/7 items similar results (RD 95% CI) between formats 7. Ability to interpret risk RD: 0% (−5.3–5.4), P = 0.99 8. Ability to relate N of participant/studies and outcomes RD: −3% (−7.5–1.7), P = 1.00 9. Ability to interpret footnotes RD: 7% (−2–15), P = 0.18 | Questions where largest proportion in favour of alternate SoF tables: 5/6 Questions where largest proportion in favour of existing SoF table: 1/6 | Overall accessibility mean difference (MD (SE)) in favour of alternate SoF: MD 0.3 (0.11), P = 0.001 | MD (SE) in favour of alternate SoF: 2.8 (1.6) |
Opiyo (2013) [50] • Normal systematic review (SR) • SR plus SoF tables • Graded-entry SR | Understanding (2 questions per format on 3-point scale) Composite endpoint (on 5-point scale) Clarity (1 question per format on 3-point scale) Accessibility (2 questions per format on 5-point scale) | Odds ratio (OR) (95% CI) SR plus SoF versus SR, OR 0.59 (0.32–1.07) Graded-entry SR versus SR, OR 0.66 (0.36–1.21) Sub-group analyses: policymakers understanding • SR plus SoF OR 1.5 (0.15–15.15) • Graded-entry OR 1.5 (0.64 t–3.54) | Not reported | Accessibility SR plus SoF versus SR • OR (95% CI) 0.91 • (0.57–1.46) • MD (95% CI) 0.11 (−0.71–0.48) Graded-entry SR versus SR • OR 1.06 (1.06 to 2.20) • - MD (95% CI) 0.52 (0.06–0.99) | Not reported |
Rosenbaum (2010) [61] RCT 1 • Cochrane review (CR) with no SoF table • CR with SoF table (limited formatting) • CR with SoF table (full formatting) | User satisfaction (multiple-choice questionnaire) Perceived understanding and ease of use (7 questions on 8-point scale) | Proportion who agree/strongly agree main findings were easy to understand % (95% CI) • No SoF table: 56 (37–75) • With SoF table (both formats): 60 (46–74) • P = 0.54 | Not reported | Proportion who agree/strongly agree very accessible % (95% CI) • No SoF table: 17 (2–32) • With SoF table (both formats): 41 (27–56) • P = 0.037 | 65% agreed CR should include SOF with the proposed format |
Rosenbaum (2010) RCT 2 • Cochrane review (CR) with no SoF table • CR with SoF table (limited formatting) • CR with SoF table (full formatting) | Understanding (4 questions) Time spent finding 5 key results | 2/4 difference in proportion of questions correctly answered (%, 95% CI) favouring with SOF 1. Risk in the control group: 44% (21–67) versus 93% (81–100), P = 0.003) 2. Risk in the intervention group: 11% (0–26) versus 87% (69–100), P < 0.001) 2/4 no difference in proportion of questions correctly answered (%, 95% CI) 1. Confidence of review authors: without SoF 67% (45–88) vs. with SoF 87% (69–100), P = 0.18 2. Identifying important outcomes: without SoF 33% (9–57) versus with SoF 53% (28–79), P = 0.27 | Not reported | Mean time (min) finding answers Risk in the control group • No SoF table: 4 • With SoF table: 1.5 • P = 0.02 Risk in the intervention group • No SoF table: 2.8 • With SoF table: 1.3 P = 0.118 Confidence of review authors • No SoF table: 1.5 • With SoF table: 2.1 • P = 0.47 Identifying important outcomes • No SoF table: 1.9 • With SoF table: 2.0 • P = 0.88 | 84% agreed CR should include SOF with the proposed format |