Since 2006, CARE has been working across Southeast Asia to create a number of context-specific community-based surveillance (CBS) models to increase early detection of highly pathogenic avian influenza (H5N1) and other emerging infectious diseases. This 18-page paper illustrates CARE's approaches, discussing the pros and cons of each model. The aim of the document is to highlight elements common to all forms of community-based surveillance, as well as context-specific differences, and to offer suggestions for replication.

CARE's CBS models share 3 common objectives:

1. CBS improves sensitivity - the ability of the surveillance system to catch all cases - by increasing awareness of a broad case definition.
2. CBS moves case detection and reporting along the spectrum from most passive to a more active system.
3. CBS links local mobilisation for case detection, early warning, and response to the formal surveillance system.

The models can be classified according to the following characteristics:

• Intensive, active surveillance focusing solely on avian and human influenza (AHI) - A community surveillance network involves volunteers trained and supervised by a "community surveillance coordinator" - who is a village health worker (VHW) or animal health worker (AHW). The volunteers extend the role of VHWs and AHWs by conducting regular household visits looking for suspected avian influenza cases in poultry and humans. (These visits also create a space for interpersonal communication to promote behaviour change for prevention and improve willingness of households to report). In addition to behaviour change and detecting current outbreaks, volunteers may discover events that were not reported (i.e. "cold" cases) which alert community leaders to significant underreporting in the community. If there is a suspected case, the volunteer immediately reports this to the surveillance coordinator or the commune avian influenza (AI) committee; if there are no suspected cases, zero reporting and additional information on behaviour change at household level is submitted during a monthly meeting between the surveillance coordinator and the volunteer network. CARE explains that "AHI is a complex disease for surveillance purposes because there may be multiple channels of reporting: hotlines, human health department, animal health department, community leaders, and the like. The CBS model streamlines these different reporting lines and ensures that human and animal health sectors both receive information and collaborate during suspected outbreaks. The model also simplifies and clarifies reporting channels for households and community leaders."
  
  

  CARE explains that intensive models are best in areas with high outbreak risk and limited government support for building systems with more long-term sustainability and relevance (such as an integrated disease surveillance model). For active surveillance, households must be within walking distance for volunteers. A community investing significant human resources in surveillance will enjoy greater benefits if they are also trained in prevention and response activities and can fully participate. The model may be more sustainable if activities are implemented under certain conditions and not year-round.




• Events-based systems designed to detect and respond to rare events of public health significance more broadly - Event-based surveillance does not replace a formal, indicator-based surveillance system; rather, it allows surveillance systems to cover a broader population base while using fewer financial and human resources than an expanded indicator-based system would require. Event-based surveillance models can make use of the media, health facilities, hotlines and/or community structures - for example, national media campaigns to encourage communities to be on the lookout for unusual clusters of sickness or disease outbreaks in animals and humans. Although this system would not be considered a form of CBS on its own, it could be extended through outreach to grassroots organisations and/or community and religious leaders.
  
  CARE's first illustration of this approach comes from Cambodia, where a village surveillance team (VST) comprised of the village chief, the AHW, and the VHW. This team is trained to report on events suspected to be linked to avian influenza in poultry and/or humans. The VST members are expected to incorporate informal surveillance activities into their routine. For instance, an animal health worker may ask about poultry deaths in the community when checking on the health of other animals. CARE indicates that a successful VST requires that the community have a high level of awareness of the team's role in surveillance and reporting. CARE Cambodia addressed this issue by launching the VSTs with a high profile community event: a participatory drama about the dangers of AI followed by a town-hall-style discussion and an introduction of the VST members and their responsibilities. Follow-up village discussions were designed to increase AI awareness and promote the VST's role and responsibilities as the point of contact. The key element is increasing the interaction between surveillance volunteers and the community.
  
  CARE Cambodia developed a monitoring tool used on a regular basis in partnership with local government to identify gaps within the overall surveillance system. Each VST member is asked the following questions once a month: Did you conduct any surveillance activities this month? Were there any events meeting the case definition? If yes, did you report them? If yes, to whom did you report? What was the response? By collecting this information, CARE and local government can look for patterns pinpointing potential gaps in reporting from the community or response from the district or province. It also allows identification of gaps by sector.
  
  The potential for sustainability is a particular strength of the model implemented in Cambodia, CARE explains. Because activities are minimal and consistent with volunteers' everyday routine, it is more likely they will continue participating in the surveillance system with little incentive. Another key strength of the model is reports are filtered through people with a higher skill set, which has the potential to reduce the number of reports requiring outbreak investigation and reduces the costs of maintaining the system (such as refresher training courses, which would be required if volunteers were central). Additionally, networks that are built can be used for any novel virus or unusual outbreak, ensuring systems will not be outdated when new threats emerge.
  
  Though the H5N1 virus has not yet affected Latin America, in preparation for a possible introduction of the virus in the region, efforts are underway to build event-based surveillance in Nicaragua. Goals include establishing clear reporting mechanisms, defining triggers for reporting (e.g., an excess number of Severe Acute Respiratory Illness (SARI) cases), and ensuring a broad understanding and awareness of those triggers. CARE began work by establishing rapid response capacity, a key aspect of which was bringing animal and human health teams together. CARE created simplified training materials on epidemiology and surveillance. As CARE staff began training programs, they also began working with local volunteers and trainees to define trigger indicators at the local level and implement broader communications campaigns with partners. CARE explains that training on triggers and reporting mechanisms should reach public and private health workers, pharmacists, coroners, traditional healers, religious leaders, community leaders, veterinarians, and other animal health and husbandry professionals. Also, a critical component of such a model is building high levels of awareness in the community on the importance of reporting to their local health care provider.
  
  CARE suggests that an event-based model is particularly suited to situations where:
  • There is a need to cover a large geographic area where no cases or infrequent outbreaks of AI have been recorded;
  • There are very limited resources for maintaining a surveillance system;
  • Achieving the highest level of sensitivity is not the goal because outbreak response capacity cannot keep pace with reporting;
  • There is outbreak response capacity, but surveillance systems do not reach broadly or deeply to detect emerging threats early on;
  • Private sector and/or community health structures have not yet been reached with key messages about influenza surveillance;
  • In-depth training for prevention and behaviour change communication (BCC) is not necessary because the risk of outbreaks is very low; or
  • Standard case definitions have not been circulated to health care providers.

  
  
  

• Integrated surveillance models targeting multiple diseases to improve detection of infectious disease outbreaks - The model piloted by CARE in Lao PDR is designed for early detection of outbreaks for multiple diseases, but the community-level component is focused on events rather than on indicators. The system trains community volunteers to report significant events based on 3 categories of infectious diseases based on recognition of symptoms: single cases, clusters, and unusual numbers. Village volunteer teams (VVT) of 4-7 people are formed. Possible members include the village head (who often serves as the team leader), a representative of the Women's Union, the village veterinary volunteer, community leaders, the village health volunteer, a traditional healer, or a teacher. Reporting and tally forms for VVTs are simple, picture-based, and useful in low literacy areas. In peri-urban areas it may be possible for volunteers to do regular household visits. In rural areas, VVTs organise small gatherings and community events to discuss health issues and conduct surveillance activities.
  
  Under the pilot model, when VVT members are notified of an outbreak they complete a "pink" form which includes basic case details represented in picture form. The village volunteer then takes this form and reports the situation to the village chief, who completes a "green" form that notes the total number of cases. Copies of both the pink and green forms are sent to zonal-level staff, who then verify the outbreak report and complete a "white" form, which is sent to the district for response. A key aspect of the model is ensuring all levels report immediately for support to the next level on the flow chart. Given the time that response measures may take, it is important to strengthen capacity of village volunteers and zone staff to take initial response measures - control, containment, and communications.
  
  CARE indicates that there are multiple challenges to an integrated system, such as the fact that it is difficult to train volunteers on information about multiple diseases, and integration requires close coordination between multiple departments and ministries. Factors that may be more suited to an integrated, events-based model include: low risk of AI outbreaks, high risk of other infectious diseases, and rural areas with low population density.

Summary of lessons learned:

1. While a highly sensitive system might seem ideal, increasing sensitivity also increases cost and responsibility of local government to investigate suspected cases. Can local government support the system? Will it be able to respond to all reports if the system is very sensitive?
2. Standard operating procedures (SOPs) for suspected outbreaks that detail actions community leaders should take are an important aspect of prevention and containment. Training for surveillance volunteers and community leaders on community-level response measures and containment strategies could be vital in preventing spread of disease.
3. Geographic targeting strategies at national, provincial, and district levels can conserve resources. At the community level, targeting can reduce volunteer activities – for instance, by deciding to visit only high-risk households or to conduct household visits only during high-risk season. If a BCC strategy is employed, volunteers should be matched appropriately to households in order to ensure maximum receptiveness to the messages.
4. Sustaining volunteer effort is an important aspect in designing a locally appropriate CBS model. The more that volunteer activities overlap with their regular duties and habits, the more likely the surveillance activities will be sustained with limited resources. Volunteer motivation is crucial; volunteers are often primarily motivated by the opportunity to provide a useful service or information to their community. Ensuring opportunities for refresher trainings and recognition from the community greatly improves sustainability of volunteer effort.
5. For most event-based models, it is clear what type of information needs to be communicated from the community to the formal system during a suspected outbreak. In the absence of an outbreak, one danger to guard against is that district level departments may attempt to integrate paperwork and activities at the community level with the formal system, resulting in overly burdensome costs (human and financial resources) compared to the value added.

CARE concludes that "[t]here is no single model appropriate for all contexts; rather, the principles of community-based surveillance can be successfully applied in any community around the world to increase the ownership and capacity of communities to detect and respond to disease outbreaks." A series of questions for replications, provided in a blue-tinted text box in the final 2 pages of the document, is designed to help guide practitioners considering which model to use, and how to proceed.

Email from Whitney Pyles to The Communication Initiative on February 13 2009.