**Background:** Low birth weight (LBW), defined as birth weight less than 2500 g, is a major contributor to neonatal mortality and long-term adverse health outcomes including neurodevelopmental problems, stunting, respiratory disorders, lower IQ, and adult-onset chronic diseases. Globally, approximately 15% of live births are LBW, with the highest burden in southern Asia and Sub-Saharan Africa. However, the true magnitude is likely underestimated due to poor data coverage and quality, including heaping of birth weight data (rounding to nearest 100g or 500g), use of inaccurate analogue scales, lack of standardized protocols, and incomplete recording. Accurate birth weight measurement is critical for guiding individual newborn care and for monitoring population-level progress toward global health targets including the Every Newborn Action Plan, Global Nutrition Plan, and Sustainable Development Goals.
**Methods:** The study was conducted at 22 health facilities in four Kenyan counties (Homa Bay, Siaya, Migori, Kisumu) and three facilities in Dar es Salaam, Tanzania (Amana and Temeke referral hospitals, Mbagala Rangitatu health centre). The prospective intervention arm collected data from October 2019 to February 2020 (4.5 months), targeting all neonates born during the period. Historical control data came from the same facilities for the same calendar months of the preceding year. The intervention package included: (1) provision of battery-operated digital scales (Seca 354) measuring in 10g increments, plus standard calibration weights; (2) three-day training for nursing staff on accurate measurement, calibration, and recording with 10g precision; (3) weekly supervision and feedback by health facility supervisors or research team members; and (4) mid-point half-day retraining. Data were collected on maternity ID, birth weight in grams, sex, and date of birth. In Kenya, data were entered using CommCare software; in Tanzania, data were entered into Excel. All data were double-entered and cross-checked. Statistical analyses included calculation of mean birth weight with 95% CIs, comparison using Student's independent t-test, and calculation of absolute difference and risk ratio for LBW prevalence, adjusted for clustering by health facility. A sensitivity analysis reallocated 25% of infants with exact birth weight of 2500g to the LBW category to account for digit preference.
**Key Results:** The prospective sample included 8441 newborns in Kenya and 4294 in Tanzania; historical data included 9318 in Kenya and 12,007 in Tanzania. Mean birth weight in Kenya was 3080g (SD 590) prospectively vs 3190g (SD 570) historically (difference 110g lower, 95% CI: 100-130). In Tanzania, mean birth weight was 2990g (SD 590) prospectively vs 3030g (SD 530) historically (difference 40g lower, 95% CI: 20-60). LBW prevalence in Kenya was 12.6% (95% CI: 10.9%-14.4%) prospectively vs 7.8% (95% CI: 6.5%-9.2%) historically, an absolute difference of 4.8 percentage points (95% CI: 3.2%-6.4%) and risk ratio of 1.61 (95% CI: 1.38-1.88). In Tanzania, LBW prevalence was 18.2% (95% CI: 12.2%-24.2%) prospectively vs 10.0% (95% CI: 8.6%-11.4%) historically, an absolute difference of 8.2 percentage points (95% CI: 2.3%-14.0%) and risk ratio of 1.81 (95% CI: 1.30-2.52). Sensitivity analysis adjusting for heaping at 2500g yielded essentially similar results: adjusted LBW prevalence in Kenya was 12.9% vs 8.5% (risk ratio 1.51, 95% CI: 1.29-1.77); in Tanzania, 18.5% vs 11.4% (risk ratio 1.62, 95% CI: 1.23-2.13).
**Clinical Implications:** This study demonstrates that routine birth weight records in Kenya and Tanzania substantially underestimate LBW prevalence—by nearly 5 percentage points in Kenya and over 8 percentage points in Tanzania. The intervention of providing digital scales, training, and supportive supervision is relatively low-cost and readily deliverable in low- and middle-income countries. More accurate identification of LBW infants at the individual level enables better targeting of care for vulnerable newborns, including thermal care, feeding support, and monitoring for complications. At the population level, improved data accuracy supports more precise tracking of progress toward global health targets and more appropriate resource allocation. The LBW prevalences reported (12.6% in Kenya, 18.2% in Tanzania) are higher than UNICEF's 2015 national estimates (11.5% and 10.5% respectively) but consistent with regional estimates for sub-Saharan Africa (12.2-17.2%). The findings are not nationally representative but the main conclusion—that routine data underestimate LBW and that a simple intervention can improve accuracy—is robust and generalizable to similar settings.