Intervenciones personalizadas para abordar los determinantes de la práctica

Tailored implementation strategies are frequently recommended to improve healthcare professional practice. Tailoring involves the selection and design of strategies to address context-specific barriers (referred to as determinants of practice) to best practice recommended in clinical guidelines. Improvements in practice are thought to be more likely if implementation strategies are selected to address identified practice determinants. This is an update of a review published in 2015.

To assess the effect of tailored implementation strategies, compared with a non-tailored strategy or no strategy, in improving healthcare professional practice. Secondary objectives were to assess whether the effects of tailored implementation strategies differ according to whether theory, evidence of the effectiveness of strategies, and input from stakeholders were involved in the tailoring process, and to assess whether the effects of tailored implementation strategies differ according to setting (high- or low-income country).

We searched CENTRAL, MEDLINE, Embase, two other databases and two trials registers from 2014 to 5 March 2024. We performed a forward citation search for papers citing the previous update. We did not apply any restrictions on date of publication, publication status or language.

We included randomised controlled trials (RCTs), including cluster-RCTs, that compared tailored implementation strategies with strategies not tailored to address determinants of health professional practice, or no strategy. We excluded studies of tailored strategies targeting behaviour change among patients only.

The primary outcome was healthcare professional practice assessed using measures of adherence to recommended practices or guidelines in a healthcare setting.

We used the Cochrane risk of bias tool version 2 (RoB 2) to assess risk of bias in the studies.

Review authors (working in pairs) screened all citations, extracted data, and assessed risk of bias independently and in duplicate. A third review author resolved disagreements. We performed meta-analyses using random-effects models for the primary outcome using the most conservative estimate of effect where multiple outcomes were reported. We also performed meta-analyses using the least conservative estimate in sensitivity analysis. Where data were unsuitable for pooling in meta-analyses, we conducted a narrative synthesis using a vote-counting approach. We assessed heterogeneity using the I² statistic and the certainty of the evidence for the main comparison using GRADE.

The previous version of this review included 32 studies. In this update, we excluded five of those studies as they no longer met the review's eligibility criteria; they did not meet the definition of tailoring or focused on patient outcomes only. We added 52 new studies (including two from trials registers), bringing the total number of included studies to 79. Participant numbers were available for 41 of these studies (n = 25,630); participant counts were not reported or were unclear in the other 38 studies.

Overall, 35 studies with a total of 5015 healthcare professionals examined the effect of a tailored implementation strategy on professional practice compared to a non-tailored strategy. In total, 44 studies with a total of 20,615 healthcare professionals examined the effect of a tailored strategy on professional practice compared to no strategy. We judged 41 of the 45 studies included in the meta-analysis as free from high risks of bias across all domains. Bias in the measurement of the outcome was the most common domain judged at high risk of bias among the remaining studies. Overall, the certainty of the evidence in the main comparison (tailored vs non-tailored) was moderate, reflecting visual inconsistency and high heterogeneity between study results. We found that tailored implementation strategies probably lead to a slight improvement in professional practice compared with non-tailored strategies (odds ratio (OR) 1.49, 95% confidence interval (CI) 1.26 to 1.75; 23 studies, 2600 participants, moderate-certainty evidence). An OR greater than 1 indicates a small improvement in professional practice compared with no strategy. Some studies included effect measures for more than one outcome. Sensitivity analysis pooling the least conservative effect measures that were reported showed an OR of 1.74 (95% CI 1.3 to 2.24. For tailored implementation strategies compared with no strategy, we report a pooled OR of 1.34 (95% CI 1.17 to 1.53; 24 studies, 8250 participants). Some studies included effect measures for more than one outcome. Sensitivity analysis pooling the least conservative effect measures that were reported showed an OR of 1.51 (95% CI 1.28 to 1.79; P < 0.0001).

Despite the increase in the number of new studies identified, our overall finding is like that of the previous review. Tailored implementation strategies probably result in a slight improvement in professional practice.

Dr Sheena McHugh received funding from the Irish Health Research Board (HRB-RL-2020-004) that helped to support the conduct of this review.

Registration: Zenodo, 7104299, via DOI: 10.5281/zenodo.7104299 Protocol available via DOI: https://doi.org/10.5281/zenodo.7104299 Previous versions available via: DOI: 10.1002/14651858.CD005470.pub3, DOI: 10.1002/14651858.CD005470.pub2, and DOI: 10.1002/14651858.CD005470.