Principal findings

This evaluation of a pharmacist-led handshake stewardship program at a quaternary hospital in Ghana identified two gaps limiting stewardship effectiveness: an acceptance gap (28.4% of recommendations declined) and an execution gap (50.0% of accepted recommendations not implemented). While roughly three-quarters of recommendations were verbally accepted, only half were ultimately implemented, suggesting that prescriber agreement alone may be insufficient to ensure prescribing change without better workflow integration. Duration optimization was the primary stewardship need, accounting for 76.1% of interventions; antibiotic selection was generally appropriate at baseline (77.1% concordance), whereas prolonged prophylaxis (>24 h in 43.2% of cases) remained the dominant driver of potentially inappropriate antibiotic exposure. ^{Reference Rafati, Shiva, Ahmadi and Habibi14}

Comparison with existing literature

Our 71.6% verbal acceptance aligns with Hurst et al., whose evaluation of pediatric handshake stewardship reported 78% acceptance, confirming the model’s effectiveness for engagement. ^{Reference Hurst, Child, Pearce, Palmer, Todd and Parker10,Reference Morton, Curzake, Parente, Morrill, Gaitanis and Laplante13} However, our 50% implementation rate falls well below the 86% reported by MacBrayne et al in a US sustainability study. ^{Reference MacBrayne, Williams and Levek11} This discrepancy is consistent with resource constraints in LMIC settings, where 24/7 pharmacist coverage is often unavailable, so recommendations agreed upon verbally are frequently lost when patients are discharged before the change is enacted or when the prescriber is later unavailable to execute it. ^{Reference Pokharel, Raut and Adhikari4} The barriers we observed were therefore less about clinical disagreement than about the logistics of converting a verbal, face-to-face agreement into a documented order under limited staffing. Our 43.2% rate of prophylaxis exceeding 24 hours is also notably higher than the 22% reported by Elrefaei et al in a US quaternary hospital, consistent with Otieno et al.’s finding that prolonged duration is a persistent challenge in sub-Saharan African programs. ^{Reference Otieno, Campbell, Maley, Obinju Arunga and Otieno Okumu5,Reference Elrefaei, Nekidy and Nasef15}

Service- and procedure-level drivers

Stewardship outcomes also varied with service and procedural context (Table 4). Recommendations were less consistently accepted and implemented in emergency procedures and in high-volume Obstetrics and Gynecology (62.1% acceptance, 44.8% implementation) than in lower-volume services with more standardized prophylaxis practices, such as Orthopedic Surgery (100.0% acceptance, 83.3% implementation). This descriptive pattern is consistent with time pressure, heightened risk perception, and workflow disruptions in emergency and high-throughput settings, which may limit adherence to guidelines even when stewardship input is available. ^{Reference Gandra, Trett, Klein and Laxminarayan2} Comparable service-level gaps have been reported in resource-limited settings relative to higher-income contexts ^{Reference Pierce, Apisarnthanarak and Schellack17} and appear to reflect specialty-specific workflow and prophylaxis norms rather than resistance to stewardship principles. ^{Reference Hadi, Eltayeb and Balushi12} These observations suggest that the handshake model is most fragile precisely where it is most needed: in high-acuity, high-volume environments whose pace and unpredictability disrupt the real-time, face-to-face interaction the model depends on.

Implications for antimicrobial stewardship programs

These findings point to specific, feasible adaptations rather than generic calls for additional resources. Because excessive duration, not agent choice, was the dominant problem, embedding automatic stop orders for standard clean procedures within surgical order sets could prevent prolonged prophylaxis without requiring pharmacist input on every case. ^{Reference Dutcher, Yeager and Gitelman18} Because discharge timing was a leading cause of execution failure, routine predischarge pharmacy review of surgical prophylaxis would allow recommendations to be enacted before the patient leaves. And because prescriber unavailability repeatedly blocked implementation, empowering pharmacists to execute preagreed, protocolized duration limits through electronic stop orders rather than relying on repeated face-to-face verbal orders would decouple implementation from the physical presence of the prescriber. ^{Reference Yeh, Dahri and Legal19,Reference Bechir and Bechir20}

Strengths and limitations

Strengths: This study provides real-world LMIC data with granular documentation of the specific barriers to stewardship implementation, complete capture of stewardship interactions during the audit period, and a focus on implementation outcomes that are often overlooked in favor of clinical endpoints alone.

Limitations: Several limitations should be considered. First, the single-center retrospective design limits generalizability. Second, we excluded 18 encounters (7.3%) because of incomplete documentation, which may introduce selection bias; incomplete documentation could occur disproportionately during high-volume periods, emergency cases, or understaffed shifts, meaning our sample may underrepresent the most chaotic clinical scenarios in which stewardship failures are most likely. For the same reason, our implementation gap (50.0%) may partly reflect underreporting of implemented changes rather than true implementation. Third, as a descriptive analysis with a limited sample size, we did not perform multivariable adjustment for potential confounders; differences between emergency and elective procedures should therefore be interpreted as associative rather than causal. Fourth, the data set was limited to surgical prophylaxis and did not capture medical antibiotic use. Finally, this study evaluated process and implementation outcomes rather than clinical endpoints such as surgical site infections or readmissions. Although clinical outcomes are the ultimate measure of stewardship success, prolonged prophylaxis beyond 24 hours is a recognized marker of potentially avoidable antibiotic exposure, ^{Reference Aluisio, Waheed and Cameron21,Reference Branch-Elliman, Strymish, Itani, Wyatt and Gupta22} and evaluating implementation fidelity is a necessary first step in program evaluation, particularly in settings where robust clinical outcome surveillance may be limited.

Generalizability

Findings are generalizable to similar quaternary hospitals in LMIC settings implementing pharmacist-led handshake stewardship models. The predominance of duration-related prescribing gaps and workflow barriers to implementation likely reflects systemic challenges common across resource-limited surgical settings. However, service-specific patterns may vary based on local prescribing cultures and procedural case mixes. Our findings regarding the impact of emergency procedures on stewardship outcomes are particularly relevant for LMIC settings where emergency surgical volumes often exceed those in high-income countries. ^{Reference Pokharel, Raut and Adhikari4,Reference Aluisio, Waheed and Cameron21}

In conclusion, the antimicrobial stewardship recommendations delivered through handshake stewardship are frequently accepted but inconsistently implemented. Excessive duration of surgical prophylaxis, not antibiotic choice, represents the principal prescribing gap. Emergency procedures and workflow disruptions substantially undermine stewardship effectiveness. Future stewardship efforts in similar settings should prioritize embedded system solutions, such as automatic stop orders and workflow-aligned interventions, to translate accepted recommendations into sustained prescribing change. These findings support the growing evidence that effective stewardship in LMIC settings requires context-specific adaptation rather than direct implementation of high-income country models.