Data collection and validation
Following the decision by local health officials to participate in the initiative, the research team worked together with province and city health office staff to identify, collate and map sources of data that could be used in the analysis [12]. This included data sources and previous analytical work at the local, regional and national level in relation to population demographics, mortality rates and causes, intervention coverage, and local health system structure and costs. We identified and reviewed routine health information systems, special surveys and studies, technical documents, guidelines and previous analytical work. Other sources of data included national agency publications, such as national health insurance reports, health budget/expenditure, policy issuances, guidelines, and protocols. The initial round of data collection took place between February and December 2009; however, data collection and validation was ongoing throughout the implementation process to ensure the best and most appropriate data for the analysis were used. The limited data available at local level was a significant challenge, particularly in relation to mortality rates and causes, health service coverage parameters, and health system costs.
While mortality rates were generally available at the province /city level (with the exception of neonatal mortality, for which data were much scarcer), estimates were typically drawn from vital registration or death review systems that are recognised to suffer from underreporting. Data on causes of death among neonates (from delivery to one month of age) and children aged 1 to 59 months were available at the national level only. We sought expert validation of the available mortality data to verify whether these estimates were considered locally representative.
In reviewing health service coverage for critical MNCH interventions, indicators were defined across parameters of supply, demand/use and quality. Local data were generally available for the supply side parameters, such as the proportion of health facilities with a continuous supply of relevant commodities or in which trained staff were located. Most of these data were drawn from local health information systems, such as the Field Health Service Information System which records provision of health services. However, there were little or no reliable local data available for indicators of demand/access and quality, which require information on population use of health services. For example, health programs record and report on how many children with acute respiratory infection (ARI) are seen and treated with antibiotics, but there is no information available on the total number of children in the population who experience an ARI, which is needed in order to estimate the coverage of antibiotic treatment of childhood ARI. We relied on regional and national data sets to substitute for local data. For example, we used regional estimates from national surveys such as the household-based National Demographic and Health Survey and Family Planning Survey. Generally, these estimates are representative down to the regional level, not to the province/city level. Where possible, triangulation of data from different sources was undertaken to validate inputs for use in the analysis, and we sought expert validation to verify the estimates used.
It was difficult to obtain information on services provided at public hospitals. Management of many hospitals is outside of the Provincial or City Health Office, and public health officials (e.g. Provincial, City and Municipal Health Officers) often do not have ready access to the information systems of such hospitals. In addition, little information could be obtained on the private sector, which plays a particularly significant role in provision of MNCH services in urban areas. For example, four of the five facilities offering basic emergency obstetric care in Pasay City were operated by private providers. This demonstrates the difficulty faced by cities in planning services and assessing the need for quality improvement initiatives when a significant number of providers are private and the local government has no access to data from these facilities and limited regulation over their practice. For a number of indicators we gathered data directly from hospital records.
Information on the cost of health system inputs is critical for the development of sound plans, but many costs were not routinely available to LGUs for their use in planning. We gathered cost information from a wide range of sources, including distributor/suppliers’ price lists and consulting key informants regarding prices of some commodities.
The problem-solving approach
Problem-solving workshops were facilitated by the IC team and representatives from the DOH regional health offices. Between two and three problem-solving workshops were held in each LGU over a five-month period. Each workshop was one to two days in duration. Participants included program managers from the provincial/city health office, municipal health officers, health centre personnel, district nurse supervisors, sanitary inspectors and other key personnel. At these workshops, local government officials participated in structured discussion to identify the key constraints to scaling-up MNCH intervention coverage and develop a range of strategies to overcome those constraints.
The participation of stakeholders from all levels of the local health system provided rich discussion, drawing on varied experiences and perspectives. For example, municipal health workers provided the perspectives of those operating on the ground and their contributions acted as a “reality check” on the feasibility of proposed strategies or increased coverage targets. The participation of regional staff as facilitators proved valuable in the development of strategies and solutions as they were able to refer to experiences in other provinces in solving problems similar to those being discussed; and to explain the content of recent national policies and protocols. A number of scenarios that included different combinations of the strategies defined during the workshops were modelled using the decision-support tool. This provided an estimate of cost of each scenario and the anticipated impact. These results were then used by local officials to determine which combination of strategies represented the “best buys”, that is, the scenarios likely to deliver the greatest anticipated impact given the resources required.
Adoption of recommended strategies
The majority of the strategies recommended by the IC were accepted by the three LGUs and the IC team assisted in translating the strategies into the format required for inclusion in local plans and budgets, including identification of different funding sources. The IC team then verified that these strategies had indeed been incorporated into the plans and budgets. A subsequent follow-up review of the initiative validated that the majority of strategies had been included in the 2011 plans of the three LGUs, and changes in their budgets had been made to reflect these strategies [13]. However, this review also revealed that a key strategy developed by Pasay City for which there had been initial enthusiasm and high expectations, was no longer being supported.