Neonatal Sepsis, Evaluation, Bangladesh

Three to four MNCS programme upazilas will be purposively selected for this study in consultation with GoB and partners. The likely districts from where the three to four upazilas will be selected include Sunamganj, Kishoreganj, Bandarban and Sherpur. However, this may change depending on GoB/UNICEF decisions about MNCS scale-up. The upazilas will be selected considering presence of NGO partners with the ability to implement the intervention, start of MNCS in the upazilas by 2009 and no overlap with other similar programmes research projects. Twenty unions (average population of about 20,000) will be identified from the selected upazilas and randomly assigned to intervention or comparison. Unions with substantial urban populations will be excluded. In order to achieve balance in randomisation across upazilas, the investigators will make efforts to select an even number of unions from the selected upazilas.

Study design

A cluster-randomised design will be used. Of the 20 selected unions in the selected sub-districts/upazilas, 10 will be randomly assigned to intervention and 10 to comparison, that is, about 200,000 population (~40,000 households) in each arm. In both intervention and comparison unions, the common MNCS intervention package will be uniformly implemented, which includes training of community-based health workers on essential newborn care. Details of the MNCS intervention package and delivery systems can be found in Annex I. All the MNCS upazilas have also already implemented facility-based IMCI.

The objective of randomisation is to ensure balance between intervention and comparison areas in terms of population and geographical characteristics and coverage of health interventions and programmes. The sampling units (unions) will be stratified according to upazila. If any union has a total population of less then 12,000 (the minimum required to obtain an adequate sample of neonates) they will either be excluded or if such exclusion results in fewer than 20 unions being available for randomisation then two such small unions will be merged and considered as a single sampling unit. Equal number of unions for intervention or comparison will be selected from each upazila. Within each upazila, unions will be paired on distance from the upazila health complex and members of each pair be randomly selected for intervention or comparison.

In the intervention unions, in addition to routine MNCS services, services for the management of neonatal sepsis will be made available in the community, targeting all pregnant women and newborns born within the study period. Community-based management of neonatal sepsis will be provided by three types of care providers: community health workers, village doctors and community clinic care providers.

The intervention and evaluation will continue for at least 18 months. Evaluation will be ongoing and include rolling household surveys to assess intervention coverage and care-seeking and referral compliance practices. In addition, two special surveys will be conducted to assess provider performance, referral compliance, quality of care at the different levels of care, and incremental program and household costs. The first special survey will be conducted at 6 months after the implementation of the intervention and then again at 12 months. The first round of rolling survey will also include a sample of newborns born in the previous 12 months to assess baseline care-seeking patterns and coverage for neonatal sepsis.

Inclusion criteria

• development of danger signs of neonatal sepsis during neonatal period

Exclusion criteria

• developing signs not in neonatal period