Table 1

Summary of the effects of single strategies and multifaceted strategies on adherence to desired practice

StrategyBenchmark review
Author, year (reference)
OutcomeBenchmark review results—single strategy alone vs no strategyBenchmark review—detailsBenchmark review —overall conclusionBenchmark review—other comparisonsBenchmark reviews vs other (non-benchmark) reviews
Overall results consistent with other relevant reviews*?
Professional-level strategies
 A&FIvers et al, 201234Compliance with desired practiceD*†: median absolute RD‡§=3% (IQR 1.8–7.7%)26 RCTs (661 clusters/groups of health providers and 605 health professions); low-moderate risk of biasSmall (range: small to modest)A&F with or without other strategies vs no strategy:
D†: median RD‡§=4.3% (IQR 0.5–16.0%) (49 RCTs)
Yes13 16–33
C†: median percentage change relative to baseline control‡=1.3% (IQR 1.3–11%)13 RCTs; low-moderate risk of biasNot applicableA&F with or without other strategies vs no strategy:
median percentage change relative to baseline control‡=1.3% (IQR 1.3–28.9%) (21 RCTs)
 Physician reminderShojania et al, 200976
Computer reminder (delivered at the point of care)
Improvement in process adherenceD*†: median RD¶=5.7% (IQR 2.0–24%)18 RCT/quasi-randomised designModest (range: small to large)Computer reminders with other strategies vs other strategies alone:
D†: median RD¶=1.9% (IQR 0.0–6.2%; n trials not reported)
Yes13 19 21 22 25 28 29 37 40 49 58 60 66–75 77
C†: not reportedNot applicableC†: not reported 
 EOVO'Brien et al, 200750Professional practiceD*†: median RD§¶=5% (IQR 3–6.2%)19 RCT; low-moderate risk of biasSmall (range: small to modest)EOV with or without other strategies vs no strategy:
D†: median RD§¶=5.6% (IQR 3–9%; 28 RCTs)
Yes13 21 22 26–28 32 35 37 40 45–49
C†: median adjusted change=23% (IQR 12–39%)15 RCTs; low-moderate risk of biasNot applicableC†: median adjusted change=21% (IQR 11 to 41%; 17 RCTs)
 Educational meetings and workshops (including continuing medical education)Forsetlund et al, 200912Compliance with desired practiceD*†: Median RD‡§=6% (IQR 2.9–15.3%)19 RCTs; low-moderate risk of biasModest (range: small to moderate)Educational meetings with or without other strategies vs no strategy:
D†: median RD‡§=6% (IQR 1.8–15.9%) (30 RCTs)
Yes19 21 22 25 31 32 35 38 41 42 47 51–64
C†: median adjusted % change relative to the control group 10% (IQR 8–32%)5 RCTsNot applicableC†: median adjusted % change relative to the control group 10% (IQR 9–24%) (8 RCTs)
 Local opinion leadersFlodgren et al, 2011a65Compliance with desired practiceD*†: median RD¶§=9% (IQR −15 to +38%)5 RCT; high risk of biasModest and variable (range from negative, no effect, to small and large effects)
Unclear due to inconsistent and limited evidence
Local opinion leaders alone or together with other strategies vs no intervention or other strategies alone
D†: median RD¶§=12% (IQR 6–14.5%; 15 RCTs)
Mostly consistent: mixed effects21 22 24 32 37 40
C†: not reportedC†: not reported
 Printed educational materials (majority studies disseminated passively)Giguère et al, 201244Professional practiceD*†: median RD¶=2% (IQR −0.6 to 29%)7 studies; low qualitySmall and variable (range: negative, no effect, to small and large effects)Mixed but mostly consistent13 21–24 26 27 32 35–43
C†: SMD 13% (IQR 16–196%)3 studies; low or very low quality
Organisational-level strategies
 Revising professional rolesNo benchmark review identified28 61 69 78 83 84
 FacilitationBaskerville et al, 201282Compliance with desired practiceSMD†=0.56 (95% CI 0.43 to 0.68; z=8.76; p<0.001; I2=20%)
OR=2.76 (95% CI 2.18 to 3.43; non-significant heterogeneity, p=0.19)
20 RCTs and 3 CCTs (1398 participants); high qualityEffective (consistent)Not applicableYes78–81
Context-level strategies
 Financial strategiesScott et al, 201191Professional behavioursAll types of financial incentives, provided by primary care physicians
Uncertain (no combined/overall effect size)
Authors’ conclusion: different financial interventions had positive but modest and variable effects on a small number of outcome measures of quality of healthcare (7 studies)
7 studiesVariable
High uncertainty
Not applicableYes. Some subsequent reviews presented positive results and some showed no effect or mixed results20 23 78 80 85–90
 Regulatory strategiesNone identifiedNot applicableNot applicableNot applicable
 Multifaceted strategiesNo benchmark review identifiedMultifaceted strategies likely to be more effective27 32 36 48 51 52 60 61 71 78 105–113
Multifaceted less or just as effective/unclear12 13 17 19 20 34 50 65 76 104
 Tailored strategies to identified barriersBaker et al, 201092Compliance with desired practicePooled adjusted OR†=1.54 (95% CI 1.16 to 2.01) from the Bayesian analysis
Pooled OR=1.52 (95% CI 1.27 to 1.82) p<0.001 from the classical analysis
12 RCTs (2189 participants; moderate quality)Not applicableNot applicableNo other review identified
  • *Based on dichotomous data (intervention vs no intervention) from the benchmark review. Overall effect is described using the definition proposed by Grimshaw et al13 (see Methods).

  • †D, dichotomous; C, continuous; SMD, standardised mean difference.

  • ‡Weighed according to the number of health professionals (number of practices, hospitals, communities) participating in the study.

  • §Adjusted for baseline differences in the outcome.

  • ¶Unweighted or unclear weighting/adjustment.

  • A&F, Audit and feedback; CCT, controlled clinical trials; EOV, educational outreach visits; OR, odds ratio; RCT, randomised controlled trial; RD, risk difference.