A systematic review and meta-analysis of 118 randomized clinical trials (N = 417,412) found that interventions to reduce potentially inappropriate prescribing in primary care were associated with a modest reduction in the number of medications prescribed but were not associated with significant changes in hospitalization, adverse drug reactions, or mortality. However, the absence of increased harm supports the safety of deprescribing.
The trials included community-dwelling adults and residents of long-term care facilities aged 65 years or older. The primary objective was to determine whether interventions targeting potentially inappropriate prescribing (PIP)—defined as medications with risks that may outweigh their benefits—could improve health outcomes. Certainty of the evidence was rated high for medication reduction and low to moderate—typically low—for other outcomes.
Patients who received a PIP intervention were prescribed approximately 0.5 fewer medications on average than those receiving usual care (standardized mean difference [SMD], −0.25; 95% CI, −0.38 to −0.13). This reduction was observed across both explicit interventions (e.g., those based on criteria like STOPP) and implicit interventions (e.g., clinician-led assessments using judgment rather than fixed rules).
No statistically significant differences were found for other outcomes. These included nonserious adverse drug reactions (risk ratio [RR], 0.92; 95% CI, 0.58–1.46), injurious falls (SMD, 0.01; 95% CI, −0.12 to 0.14), emergency department visits (RR, 1.02; 95% CI, 0.96–1.08), medical outpatient visits (SMD, 0.02; 95% CI, −0.02 to 0.07), or all-cause mortality (RR, 0.94; 95% CI, 0.85–1.04). While hospitalizations were slightly lower in the intervention group (RR, 0.95; 95% CI, 0.89–1.02), this reduction was not statistically significant and should be interpreted with caution.
Most interventions were pharmacist-led, delivered via clinical care teams, or incorporated software tools that flagged inappropriate medications. Examples included structured medication reviews and patient-specific deprescribing recommendations. Trial durations ranged from 1.5 to 24 months, with most lasting 6 to 12 months.
Quality of life did not differ significantly between groups (SMD, 0.09; 95% CI, −0.04 to 0.23). Among trials reporting this outcome, 73% used the EuroQol 5-Dimension (EQ-5D) tool. Variation in measurement instruments may have contributed to inconsistencies across studies.
Subgroup and sensitivity analyses did not reveal meaningful differences by patient characteristics, care setting (community vs long-term care), or intervention type (explicit vs implicit). No evidence of publication bias was detected.
Although these interventions reduced medication burden, they did not clearly improve other health outcomes but were deemed safe, with no evidence of increased harm. There was substantial heterogeneity across trials (e.g., I² = 90% for medication reduction), along with variability in outcome definitions and follow-up durations. The authors recommended that future studies report harms and benefits using standardized definitions to enable meta-analyses of key outcomes such as hospitalization, quality of life, and mortality.
Source: JAMA Network Open
