Study protocol : improving newborn survival in rural southern Tanzania : a cluster-randomised trial to evaluate the impact of a scaleable package of interventions at community level with health system strengthening

Child mortality has declined substantially in many countries including Tanzania, but newborn mortality remains high and around 3 million babies die every year in the first 28 days of life. Community-based approaches with home visits in the first week of life have shown great potential to reduce newborn mortality. INSIST aimed1 to develop, implement and evaluate an integrated, two-part strategy that combines interventions at community level with health system strengthening in rural Southern Tanzania to reduce newborn mortality. The community intervention focused around interpersonal communication through home visits in pregnancy and the early neonatal period by a village-based "agent of change". Key messages focused on hygiene during delivery, immediate and exclusive breastfeeding, and identification and extra care for babies born small because of low birth weight or prematurity. Extra care for babies born small included skin-to-skin care for small babies and referral to hospital for very small babies. The community intervention was implemented in six poor rural districts in Southern Tanzania, with 65 of the 132 wards within these districts randomized to receive the community intervention. In addition, a health system quality-improvement package was implemented in all health facilities of one district. Data collection for the evaluation included i) a baseline household survey in 2007 of all 243,000 households in 5 of the 6 study districts to estimate baseline mortality and prevalence of newborn care behaviours, ii) an adequacy survey in 2011 in a representative sample of 5,000 households to estimate coverage of home visits and prevalence of newborn care behaviours, and iii) an endline household survey in 2013 in a representative sample of 200,000 households to estimate newborn and maternal mortality and prevalence of newborn care behaviours. The final analysis was based on "intention to treat", comparing newborn