Sentinel Community Surveillance (SCS)

TO BE REVISED

Sentinel Community Surveillance is a special type of information system based on a combination of quantitative and qualitative techniques. They generate information from a selection of sentinel sites, chosen according to established criteria (purposive sampling). Questionnaires are applied to the total population within each site (ideally a community or neighbourhood). This is cross-checked with and informed by a combination of focus group and key informant interviews. Finally, preliminary findings are fed back to local institutions involved, as well as representatives of the communities concerned.

This is useful when access is limited, or the volume of information from all sites is unmanageable or unnecessary. Representation of information will depend on careful analysis of criteria for selecting sites.

SCS were conceived in Central America in 1984 as a capacity building development process for producing accurate, detailed and actionable data rapidly and at a low cost. It focuses on the use of those data in local and national planning by stimulating informed dialogue. This dialogue occurs not only in the sites but, via a well-informed communication strategy, trough the domain represented by the sites. This may be a municipality, a city, a state, a number of provinces or an entire country.

While developed in the health sector, SCS has been refined and used in a wide range of development concerns: education, food security, environment, landmines, transport, justice.

The table summarise some key features of SCS:

Representativeness of
Sentinel sites / Sentinel sites are not pilot areas to test special interventions, but they are intended to be representative of conditions and services. They are different only in that they are a concentration of measurement resources.
Cross syntheses of methods / Sentinel sites are a concentration of measurement resources. Household survey is the methodological cornerstone, but other qualitative and quantitative data allow fine tuning and insights. Analysis undertaken in sentinel sites includes: analysis of existing data, household surveys, rapid assessment techniques.
Manageability / Sentinel sites are chosen so that data are manageable in volume and concept. Data should be kept simple and sparse, increasing in quantity and sophistication as the capability increase
Potential links with existing system / Sentinel systems are not managed as a parallel information system: their strength lies in the potential links with existing information systems. Whatever data is lacking or in unreliable may be better investigated in the sentinel sites.
Cyclical repetition / By returning only to the same panel of community after a given time interval, savings can be made from monitoring everywhere at the time. The ability to repeat measurement in the same place makes impact estimation relatively straightforward
Logistical efficiency / SCS usually yields larger sample sizes that the same resources invested in random samples (it is more efficient to measure all households of a community than sparse ones)
Focus on whole communities / SCS involves large, contiguous parts of a community (i.e. not only seven household per cluster, as in other surveys). Inside a sentinel site all households are covered, therefore SCS data are handled as mini-universes. Such and in-depth study of a community allow the analysis of factors affecting impact and the link between qualitative and quantitative data that would not be possible otherwise.
Building up local expertise / After few years SCS will have developed substantial expertise in the design and in conducting special studies in a variety of areas. The capacity building is not a one-off training, but the establishment of a learning curve linked to practice

The choice of sentinel sites

How should the sample be chosen to optimise the balance between logistic and scientific considerations? How can we be sure that indicators obtained by a sample will be representative of the whole community?

In general the selection of sites is based on purposive selection, and this allow the application on SCS also in data-poor countries, where there is no up-to-date sampling frame.

To minimise potential error and bias consider:

Potential for error / Accept that choice of site may be not be representative and account for an error in the results
Number of sites / the larger the number of sites, the most likely are to be representative. The largest feasible number is desirable
Stratification / stratification increases the chances of the proportions to be correct (for example: six sites from the cities, two from minor towns… this is very similar to the procedures used in multistage random sampling)
Additions and changes / Selections should not be seen as definitive: inclusion of additional sites, for example, may permit adjustments of the selection. Also remember that periodically a changeover with new sites should take place
Clarity of criteria / Criteria for selection need to be clearly specified
Multiple views / The selection should be done by a multi-disciplinary group, with all members having experience of the area to be worked
Validation trough other studies / There is also the possibility of validation or comparison with other studies investigating fewer variables but using a different methodology

Note: In a rapidly changing situation, such as civil war, a site may change characteristics (population, economy, status vis-à-vis centre-periphery) thus changing its eligibility according to sampling criteria and its representation of a given region or type of population. Access to a site may also be lost. Data, particularly for time series monitoring, must then be interpreted with great care.

A typical cycle

Sentinel Community Surveillance (SCS) - Page 2/2