Multilevel onsite training and mentorship model to accelerate early childhood cancer diagnosis in Northwest Ethiopia: A quasi-experimental mixed method study

BACKGROUND: Childhood cancer survival in low-income countries remains below 20%, with late recognition and delayed referral commonly reported. In Northwest Ethiopia, care providers often lack the knowledge, skills, and support systems for recognizing early warning signs of pediatric malignancies. Community caregivers often present only after symptoms become advanced. The University of Gondar Comprehensive Specialized Hospital piloted a quality improvement initiative aimed at achieving earlier diagnoses. This study reports whether a multilevel onsite training and mentorship model could improve early recognition, referral practices, and timely diagnosis of childhood cancer in Northwest Ethiopia. METHODS AND FINDINGS: From January 2024 through September 2024, a quasi-experimental pre-post, mixed-methods study design was employed across three tiers of care: primary (e.g., health centers), secondary (general hospitals), and tertiary (specialized hospitals). Eighteen primary-level and 29 secondary-level clinicians completed intensive, on-site training (Ten and Seven days, respectively), while one thousand twenty health extension workers received pictorial outreach modules. A 6-month mentorship program combined monthly onsite visits with remote supervision. Mentees included general practitioners, nurses, health officers, and health extension workers mentored by a multidisciplinary team that included, for example, pediatric haemato-oncologists. Knowledge, attitude, and practice surveys and clinical chart reviews were conducted at baseline and 6 months post-implementation to evaluate patient-journey intervals. These quantitative assessments were integrated with qualitative interviews and focus groups grounded in the consolidated framework for implementation research. The intervention utilized specific curricula focusing on clinical recognition, referral protocols, and documentation. In this study, the participants had a 100% response rate. Median knowledge scores at primary and secondary levels rose from 54.6 (95% CI: 36.4, 63.6) to 90.90 (95% CI: 81.8, 100.0) and from 36.36 (95% CI: 27.3, 45.5) to 90.91 (95% CI: 61.4, 93.7), respectively. Median practice proficiency increased 87.5 (95% CI: 78.7, 100) to 100.00 (95% CI: 100.0, 100.0) at primary level and 68.75 (95% CI: 50.0, 87.5) to 93.8 (95% CI: 86.7, 100.0) at secondary level. The median interval from symptom onset to first health contact fell by 9.3% (from 27.0 days (95% CI: 16.0, 33.3) to 24.5 days (95% CI: 15.0, 32.0)), and the time from first contact to confirmed diagnosis decreased by 54.2%. Treatment initiation interval increased by 11.9%, reflecting ongoing infrastructural constraints. Qualitative findings underscored the roles of supportive leadership, diagnostic supply limitations, cultural beliefs, and referral coordination in shaping outcomes. This quasi-experimental pre-post design without a control group limits strong causal inference, especially in the presence of potential confounders like parallel public-health initiatives and seasonal variations in care-seeking. CONCLUSIONS: A context-tailored, multilevel training and mentorship model was associated with improved provider capacity and reduced diagnostic delays in Northwest Ethiopia. While the initiative demonstrated high fidelity and adaptability in conflict-affected settings, achieving timely treatment requires further investment in diagnostic infrastructure. These tools and protocols are well-positioned for national scale-up and integration into routine continuing medical education.