Impact of Point-of-Care Rapid Diagnostic Tests on Antibiotic Prescription Among Patients Aged <18 Years in Primary Healthcare Settings in 2 Peri-Urban Districts in Ghana: Randomized Controlled Trial Results

The study addresses inappropriate antibiotic prescribing for acute febrile illnesses in primary care settings with limited diagnostics, a key driver of antimicrobial resistance (AMR). Ghana has high antibiotic prescription rates, particularly in primary care, amid reports of multidrug resistance. Nonadherence to prescriptions also contributes to resistance and poor outcomes. Evidence suggests that point-of-care testing (POCT) can reduce diagnostic uncertainty and improve antibiotic stewardship, especially for respiratory tract infections. The objective was to evaluate whether a package combining POCT, a clinical diagnostic algorithm, and training/communication (T&C) reduces antibiotic prescriptions and maintains clinical outcomes in children and adolescents with nonsevere acute febrile illness in two peri-urban districts in Ghana.

The paper contextualizes AMR as a global health challenge with substantial mortality and highlights Ghana’s high antibiotic prescribing rates and documented multidrug resistance. Evidence from prior studies indicates that POCTs (e.g., CRP, respiratory pathogen tests) can guide antibiotic use and reduce inappropriate prescribing, potentially curbing resistance. Studies have assessed POCT effectiveness for respiratory infections, and WHO initiatives (e.g., AWaRe classification, global AMR action plan) provide stewardship frameworks. The literature also identifies common reasons for nonadherence to antibiotics and suggests behavioral strategies to improve adherence.

Design: Open-label, prospective, parallel 1:1 randomized controlled trial conducted in four public health facilities in Shai-Osudoku and Ningo-Prampram districts, Greater Accra Region, Ghana, from September 2020 to September 2021. Participants: Children and adolescents aged 6 months to <18 years with nonsevere acute febrile illness (>37.5°C or history of fever within 7 days), willing to provide specimens and available for follow-up. Severely ill patients requiring hospitalization were excluded. Informed consent/assent obtained. Randomization: Centralized, block sizes of 64, 96, 128; site-specific lists concealed from clinicians and applied sequentially by field research officers. Open-label allocation to intervention or standard care. Intervention: Package comprising (1) pathogen-specific POCTs (influenza A/B, RSV, Streptococcus pneumoniae urine antigen, group A Streptococcus antigen, Salmonella Typhi IgM, malaria Pf/Pan antigen), (2) non–pathogen-specific tests (CRP, full blood count with WBC and differential, urine dipstick), (3) a clinical diagnostic algorithm guiding antibiotic decisions based on POCT results, CRP, and WBC, and (4) training and communication materials for prescribers and caregivers developed from baseline qualitative research to enhance adherence. Control: Routine standard-of-care with clinical assessment; labs requested per national guidelines (commonly malaria Ag and WBC); prescriptions per results and clinician judgment. Separate prescribers and rooms were used to minimize contamination between arms. Endpoints: Primary—(a) proportion with favorable clinical outcome at day 7 (alive and resolution of presenting symptoms), (b) proportion prescribed any antibiotic at day 0 (or until day 7). Secondary—healthcare worker adherence to algorithm; patient/caregiver adherence to prescriptions (pill count ≥90% and qualitative interviews); adverse events (AEs) and serious AEs (SAEs). Follow-up: Day 7 clinic or telephone visit for symptom resolution and adherence verification; additional investigations as needed for persistent/worsening symptoms. Sample size: Planned n=2766 to detect 30% relative reduction in antibiotic prescriptions (control 55%) with 80% power, alpha 0.05, 4% precision and 15% LTFU. Actual enrollment n=1512 (52%). Analytic precision adjusted from 4% to 6% while maintaining power and alpha as planned. Statistical analysis: R v4.2.1. Descriptive statistics; chi-square tests; estimates of proportions with 95% CIs; relative risk (RR) and absolute differences with 95% CIs for primary outcomes. Prespecified subgroup analyses by age, sex, and quarter; exploratory analyses by respiratory vs nonrespiratory diagnoses and malaria status. No multiplicity adjustment. Ethics approvals obtained from Oxford (Ref 52-19) and Ghana Health Service (Ref GHS-ERC 014/07/19).

- Enrollment: 1512 participants randomized (intervention 761; control 751); 53.5% male; 76.3% aged <5 years; median age 2 years (Q1–Q3: 1–4). - Clinical presentations: Fever 94.4% (1428/1512), cough 64.0% (968/1512). Post-test respiratory diagnosis in 73.4% (1110/1512). - POCT results (intervention arm unless stated): Malaria Ag positive 11.4% (84/738; control 11.2% [78/696]); influenza A/B positive 10.0% (76/758); group A Streptococcus 4.6% (35/758); RSV 2.8% (21/755); S. pneumoniae urine Ag 5.1% (21/412); Salmonella Typhi IgM 1.6% (12/759). CRP <20 mg/L in 83.5% (627/751); WBC <11,000/µL in 71.0% (540/761). - Primary outcomes: • Favorable clinical outcome by day 7: 99.7% (759/761) intervention vs 99.4% (747/751) control; very low unfavorable outcomes (0.3% vs 0.5%). • Antibiotic prescription overall: 38.7% (294/760) intervention vs 43.4% (323/745) control; RR 0.89 (95% CI 0.79–1.01). - Subgroups with significant reductions in antibiotic prescriptions: • Age <5 years: RR 0.86 (95% CI 0.75–0.98). • Nonmalaria (mRDT-negative) patients: RR 0.85 (95% CI 0.75–0.96). • Respiratory diagnosis: RR 0.84 (95% CI 0.73–0.96). • By site, Shai-Osudoku: RR 0.81 (95% CI 0.71–0.93); St. Andrew’s showed no overall reduction (RR 1.16; 95% CI 0.90–1.50). - Adherence to antibiotic prescriptions: • By pill count: 89.3% (117/131) intervention vs 93.6% (160/171) control; P=0.263. • By qualitative interview: 94.5% (277/293) intervention vs 96.0% (308/321); diff −1.4% (95% CI −5.1 to 2.3; P=0.527). • Adherence to no-antibiotic prescriptions: 99.8% (466/467) intervention vs 99.5% (421/423) control. - Healthcare worker adherence to algorithm: Instances of discordance noted—52 patients received antibiotics contrary to algorithm guidance; higher proportional prescribing at Shai-Osudoku. - Safety/withdrawals: Four participants (2 per arm) hospitalized on day 1 (3 high malaria parasitemia; 1 tonsillitis with elevated WBC); all discharged after 48 hours and had favorable outcomes by day 7. AEs were limited to persistent symptoms in a few cases; no mortality reported.

The diagnostic package combining POCTs, a clinical algorithm, and T&C maintained excellent clinical outcomes and reduced antibiotic prescriptions overall with borderline statistical significance, and achieved significant reductions among key subgroups: children under 5, mRDT-negative (nonmalaria) patients, and those with respiratory symptoms/diagnoses. These findings support the role of POCT-enabled stewardship in primary care settings with limited laboratory capacity, aligning with similar trials in Burkina Faso and Uganda. Despite reductions, antibiotic prescribing remained above the WHO-recommended <30% benchmark for primary care, underscoring ongoing stewardship needs. High adherence to prescriptions in both arms suggests good caregiver engagement and potentially some contamination of messaging between arms; nonetheless, the strategy appears safe and effective for targeted reductions, particularly when bacterial infection is unlikely by testing and biomarkers (CRP/WBC).

Integrating POCTs with a clinical diagnostic algorithm and tailored training/communication at the primary care level in Ghana reduced antibiotic prescribing without compromising patient outcomes, with the largest benefits among children <5 years, nonmalaria febrile patients, and those with respiratory syndromes. Policymakers in low- and middle-income countries should incorporate simple, affordable POCTs into primary care to support evidence-based antibiotic use. Future research should optimize algorithm adherence, evaluate cost-effectiveness, assess broader respiratory viral seasons (including SARS-CoV-2 where relevant), and investigate strategies to meet WHO prescribing targets.

- Under-enrollment: Only 1512 of planned 2766 participants were recruited (52%), reducing precision and power relative to original assumptions; analytic precision adjusted from 4% to 6%. - Low positivity of respiratory pathogen tests likely influenced by COVID-19 context (school closures, lockdowns), potentially reducing detectable impact of pathogen-specific testing. - SARS-CoV-2 was not included in testing; respiratory etiology may be undercharacterized. - Potential contamination between arms (participant and prescriber interactions) may have attenuated observable differences. - Variability in healthcare worker adherence to the algorithm and site-level differences in prescribing behavior could affect generalizability.