Table 2. Specific questions - sub-study designs and methods.

Study design and methodology	Study site; study populations; sampling procedures; and data collection procedures
1. Does co-design of job aides and communications skills training support improve adoption of innovations and the quality of medical and nursing care
Rapid cycle co-design meetings with hospital staff of job aides (e.g. to support better documentation of vital signs observations and feed prescribing), structured team based case review (TCR audit) tools and to adapt an existing communications training will be used with pilot testing in these sites for an initial six months. (Completed) Currently - Introduction of new job aides and TCR audit tools is being progressively extended to MPN hospitals in the closing part of Phase 1 in preparation for Phase 2 when feedback on key indicators and MPN peer engagement will be used to promote their use. The adoption of job aides and their early effects on process measures of quality of care will be evaluated in Phase 2 to inform continued efforts to promote adoption.	Study sites: We will seek 2 - 3 MPN hospitals as volunteers for the rapid-cycle co-design activities prior to progressive pilot implementation in the remaining 8 - 9 hospitals. Study populations: Nurses and medical staff will be identified by hospitals to work with researchers to co-design job-aides and mortality audit tools and subsequently implement these. For piloting adoption of job aides, medical records of NBU admissions in the pilot period will be sampled. Procedures: Hospital records for NBU admissions for whom a co-designed job aide should be used / completed will be used to ascertain adoption and use these pilot studies data on up to 30 cases in 2-3 hospitals, allowing further redesign of job aides as needed. Notes and results of group meetings to co-design TCR audit tools and communications training will be used to assess the design process and improve tools prior to wider scale implementation as part of a Human Centred Design approach.
2. Description of common modifiable factors in providing effective feeding to sick babies.
We will promote use of the structured neonatal TCR audit approach co-designed in Phase 1 across the MPN during Phase 2 and 3 to review late deaths occurring on the NBU (those ≥ 3 days post-natal age) to identify modifiable factors occurring during the NBU stay. TCR audit reports will be collated as part of MPN activities and aggregate reports generated on modifiable factors from across the 11 hospitals. Pooled data will be used to describe the nature and frequency of modifiable factors across hospitals.	Study sites: This work will include all 11 MPN hospitals. Study populations: Hospitals will be asked to review late post-natal deaths (those ≥ 3 days post-natal age at time of death) each month (noting that the Ministry of Health expects all neonatal deaths to be reviewed). Data collection procedure: Deidentified data will be collected from hospitals’ structured neonatal TCR audit reports. We aim to collect data on 250 cases as the basis for a report summarizing findings from all hospitals.
3. How do important process of care quality indicators and mortality change over the period of MPN participation
Building on Phase 1 and the earlier establishment of the information system we will track and develop feedback systems on quality of practices such as feed prescribing, monitoring (e.g. assessment with pulse oximetry) and mortality outcomes and progressively use this to focus hospital teams’ attention on their performance in providing quality care and achieving good survival outcomes. We will aim to further drive local learning on how to improve care through online and MPN meetings between hospitals and promote the use of the information from locally conducted neonatal TCR audits that identify factors that can be modified to improve outcomes.	Study sites: This work will include all 11 MPN hospitals. Study populations & sampling: We will examine medical records for all NBU admissions in Phases 1, 2 and 3 and especially those meeting criteria of being a vulnerable baby (either preterm birth (<37 weeks gestation) or low birth weight (<2500g). Data collection procedure: All NBU records from which routine de-identified data are currently being captured and we will use the quality indicators developed (e.g. on feeding practices and correct antibiotic use) to provide monthly mortality reports and three monthly summary reports on quality of care indicators. We will track the change in aggregate and hospital specific performance and mortality outcomes over time.
4. What is the optimum approach for risk adjustment of neonatal mortality so that variations in case-mix and case-severity can be accounted for when evaluating temporal trends in inpatient neonatal mortality
We will build on prior work undertaken in a single Kenyan hospital to develop two candidate prognostic scores for neonatal mortality and use data already captured to undertake external validation and improvement of these modelling approaches. The aim will be to develop a prognostic scoring approach that can be applied at individual patient level that enables us to undertake risk adjustment when tracking the long-term trends in survival outcomes in response to the MPN intervention.	Study sites: This work will include all 11 MPN hospitals. Study populations & sampling: We will use existing data on over 30,000 NBU to validate / revise a preferred prognostic model and then use this model to provide risk-adjusted estimates of mortality each month for the 11 sites for a total of 36 months across Phases 1, 2 and 3. These data will enable us to say with greater certainty whether NBU mortality rates are declining over the period of intervention and potentially in which sub-populations.
5. How do senior doctors and nurses’ social ties influence the performance of frontline staff and evolve within the MPN influenced by face to face and online interaction
Building on work to develop and implement the communications training we will progress to examine the relationships between medical and nursing staff within and between hospitals. This work will be informed by conduct of a realist review of social MPN analyses conducted on hospital staff and based on this proceed to empiric work involving in-depth interviews (IDIs) with senior hospital staff, small group discussions with frontline workers and family members, and episodes of non-participant observation of everyday practices. The aims of this data collection will be to explore changes in how MPN participants perceive their roles, their teamwork, the practices they employ to improve neonatal and family centred care and how these may all be mediated by the creation or strengthening of social ties resulting from MPN participation.	Study sites: Medical and nursing leads from all 11 MPN hospitals’ NBUs will be eligible for IDI and four hospitals representing maximum variation in performance on quality indicators being tracked will be identified for further IDI and small group discussions. In these four purposefully selected hospitals, non- participant observation will be conducted and front-line staff and family members will be invited for small group discussions based on convenience sampling but employing inclusion criteria to ensure diversity in respondents (e.g. based on age, qualification (for staff), gender and education (for family members) The aim is to conduct two small group discussions with staff (total four to six people) and families per site.
6. How does information being produced by the MPN reach senior advisory and advocacy group members and how may it be influential in fostering MPN growth and wider health system improvement
The MoH Technical Group of Experts is expected to meet three times monthly in Phases 2 and 3. In the last 3 months of Phase 3 IDI will be conducted with group members and key wider stakeholders. We will explore their opinions on the value of the information being created by the MPN, their experience of their work as a group, and how such information might be used in the wider health system context to promote improved NBU care. Activities or reports have been of value in helping support decision making.	This work will focus on the national level and at the level of county administrations where these have hospitals included in the MPN. In Phase 3 IDI will be conducted with members of the MoH Technical Group of Experts (n = 6 - 8), health care policy makers, key donor partner personnel (e.g. UNICEF, WHO) and selected senior county and hospital managers (total n = 6 - 10). Sampling procedures: Sampling will be purposeful with a snowballing approach used to identify relevant interviewees in national or county government, hospital management or donor organisations.
7. How can better performance measurement be integrated into national information systems to sustain improvement
We have conducted with the MoH (Informatics Division) a careful appraisal of existing neonatal data collection approaches within the national DHIS2 system revealing considerable weaknesses 110. We will draw on prior work 38 to use consensus development methods to develop a core set of preferred neonatal quality and outcome indicators for use at national level. We will follow this with use of human centred co-design approaches to improve the process of neonatal data capture and the DHIS2 tools that support it to enable long-term, large scale tracking of neonatal care and outcomes embedded in the national DHIS2.	This work will be conducted at the national level with the Ministry of Health and its Neonatal and Child Health Unit and its Informatics Division. A stakeholder meeting, linking to one of the proposed MPN meetings, will be held to review and suggest revisions to the current information procedures and tools. Draft quality and outcome indicators developed will be refined and agreed with key stakeholders though the Ministry of Health Technical Group of Experts. Based on this process we will revise existing paper tools and online DHIS2 data capture tools employing co-design workshops and ‘walk-throughs’ with health records information officers and clinicians to develop minimal viable products (MVPs) for initial testing. We will progress these MVPs so they can be incorporated into the DHIS2 system and potentially extend this by enabling distributed data capture through mobile applications.

TCR, team based case review; MPN, multi-professional network; NBU, newborn unit; IDI, in-depth interview; MoH, Ministry of Health; MVP, minimal viable product.