**Background:** Vision impairment affects an estimated 19 million children globally, with common preventable causes like uncorrected refractive error and eye infections. In Zanzibar, about 22,000 children (5% of 6–12 year olds) need treatment or spectacles, but school eye health programmes are often ad hoc and short-term due to NGO funding. The government aims to integrate eye health into existing school nutrition programmes (SFP) to improve sustainability. This study compared the performance of an integrated model (IM) versus a vertical model (VM) of school eye health delivery in Zanzibar, using the RE-AIM framework (Reach, Effectiveness, Adoption, Implementation, Maintenance).
**Methods:** A non-randomised interventional comparative study was conducted from April 2017 to October 2017 in 19 schools across Unguja and Pemba Islands. The IM (9 schools, n=6,257 children) integrated eye health into the existing school feeding programme, while the VM (10 schools, n=5,721 children) delivered eye health as a stand-alone intervention. Teachers (30 per model) received 2-day training: VM teachers were trained on distance vision screening using a modified Snellen chart (6/12 cut-off) and identifying obvious eye diseases; IM teachers additionally received training on measuring height/weight, identifying nutrition issues (BMI <18 or >30 kg/m²), and health education on balanced diet, sanitation, and deworming. Screening was conducted by teachers, and children who failed were referred to vision centres for examination and management (spectacles, medication, or referral). Performance was assessed using RE-AIM indicators: reach (proportion screened), effectiveness (follow-up rate, spectacle compliance during unannounced visit, full-time wear), adoption (schools starting within 2 months and completing within 6 months), implementation (screening validity: proportion of failed screenings confirmed by optometrist, and proportion of passed screenings confirmed), and maintenance (continuation after funding ended). Semistructured interviews were conducted with 36 stakeholders (MoH, Ministry of Education, teachers) to understand challenges and outcomes.
**Key Results:**
- **Reach:** IM achieved 96% screening coverage (5,992/6,257; 95% CI 92.1%–100%) vs VM 90% (5,142/5,721; 95% CI 85.7%–95.8%). High coverage was due to effective coordination and high school attendance. Screening peaked in April and July, with no screening in June due to monsoon flooding.
- **Effectiveness:** IM had higher voluntary follow-up rate (63.6%, 77/121; 95% CI 57.5%–69.7%) vs VM (46%, 46/100; 95% CI 35%–57%). Main reasons for not attending were financial constraints and lack of awareness. Spectacle compliance during unannounced visit was 71% (22/31) in IM vs 13.3% (4/30) in VM. Full-time wear was low: 4.5% (1/31) in IM vs 0% in VM. Reasons for part-time wear included teasing by friends (IM: 33.3%, VM: 66.7%) and not feeling the need to wear them often (IM: 30.6%).
- **Adoption:** All IM schools (100%) started screening within 2 months and completed within 6 months. In VM, 70% started within 2 months (3 schools delayed due to syllabus), and 80% completed (2 schools did not complete, citing it was 'too late to catch up').
- **Implementation:** Screening validity: 76% of children who failed screening in IM had vision <6/12 or obvious eye disease when re-examined by optometrist, vs 58% in VM. All children who passed screening in both models had vision ≥6/12 and no obvious disease (100% specificity). Delayed implementation post-training reduced validity.
- **Maintenance:** No schools in either model continued screening after funding ended (0% in both). Reasons included no budget allocated and no directive to resume.
**Clinical Implications:** The integrated model outperformed the vertical model across most RE-AIM indicators, demonstrating better reach, effectiveness, adoption, and implementation. The IM was also previously shown to be more cost-effective (cost per child screened: US$1.23 vs VM; cost per child identified: US$24.76). These findings support integrating school eye health into existing nutrition programmes to improve resource use and sustainability. However, low spectacle compliance and cessation after funding highlight the need for tailored health education (e.g., addressing teasing, cultural beliefs) and sustainable financing through government mainstreaming. The study recommends strong coordination, timely implementation post-training, and co-development of sustainability plans with local stakeholders.