Background:

Minimizing healthcare waste is essential to sustain current healthcare systems, and de-implementation strategies aim to reduce low-value care.

Objective:

To assess the effectiveness of different types of de-implementation interventions.

Methods:

We searched MEDLINE and Scopus for randomized trials comparing de-implementation interventions to placebo, no intervention, or another de-implementation intervention in primary care. We screened the titles and abstracts, and full-texts, extracted the data, and assessed the risk of bias independently and in duplicate. We used 5 predefined intervention categories: provider education, audit and feedback, patient education, decision aids, and laboratory system interventions. We conducted a random-effects meta-analysis, and assessed evidence certainty using GRADE approach.

Results:

We screened 12,113 abstracts and included 110 studies (published between 1983 and 2023). Table 1 includes the pooled estimates and evidence certainty for different types of de-implementation interventions. Prespecified sensitivity and subgroup analyses did not suggest meaningful differences.

Conclusions:

Pooled evidence suggests that achieving a meaningful impact on low-value care use usually requires a de-implementation intervention with multiple strategies. Very low certainty evidence suggests that provider education has a slight impact on low-value care use. Low certainty evidence suggests that audit and feedback has a slight, and patient education moderate impact on low-value care use. We found moderate certainty evidence that provider education combined with audit and feedback decreases low-value care use slightly. Our results, besides knowledge on the local prevalence of low-value care, are helpful when deciding on the de-implementation of low-value care.