Objective This study seeks to assess the performance of a community-based surveillance system (CBSS), developed and implemented in seven rural communes in Cambodia from 2000 to 2002 to provide timely and representative information on major health problems and life events, and so permit rapid and effective control of outbreaks and communicable diseases in general.
Methods Lay people were trained as Village Health Volunteers (VHVs) to report suspected outbreaks, important infectious diseases, and vital events occurring in their communities to local health staff who analysed the data and gave feedback to the volunteers during their monthly meetings.
Results Over 2 years of its implementation, the system was able to detect outbreaks early, regularly monitor communicable disease trends, and to provide continuously updated information on pregnancies, births and deaths in the rural areas. In addition, the system triggered effective responses from both health staff and VHVs for disease control and prevention and in outbreaks.
Conclusion A CBSS can successfully fill the gaps of the current health facility-based disease surveillance system in the rapid detection of outbreaks, in the effective monitoring of communicable diseases, and in the notification of vital events in rural Cambodia. Its replication or adaptation for use in other rural areas in Cambodia and in other developing countries is likely to be beneficial and cost-effective.
The World Health Organization asserts that ‘effective communicable disease control relies on effective response systems, and effective response systems rely on effective disease surveillance’ (WHO 2000). Routine health facility based disease surveillance systems, such as those on which most developing countries depend, could provide neither a complete nor a representative picture of health problems in the communities because patients who cannot get access to public health facilities or who choose not to use them are not reported by these systems. In order to overcome this limitation of facility-based health information systems, community-based surveillance systems (CBSS), based upon a network of lay people involved in the systematic detection and reporting of health-related events from their community, have been employed in a variety of settings. The operational characteristics and the performance of these CBSS have varied. For example, the type of data collected has varied, depending on the objectives of each CBSS. It has included the incidence of specific infectious diseases such as Guinea worm infection (Cairncross et al. 1999), yaws (Anselmi et al. 1995), malaria (Ruebush et al. 1994; Ghebreyesus et al. 2000), and tuberculosis (Balasubramanian et al. 1995); pregnancy outcomes (Ahluwalia et al. 1999); nutritional status of children (Valyasevi et al. 1995); and vital events (Jaravaza et al. 1982). CBS systems have been shown to provide useful information for monitoring disease control programmes (Cairncross et al. 1996; Ghebreyesus et al. 2000; Howard-Grabman 2000). However, these programmes have each focussed on a single disease and thus did not maximize the value of scarce resources available at the peripheral level. A CBSS targeting all common diseases and vital events would be more appropriate and resource-efficient (Manderson 1992; Cairncross et al. 1996). In this paper, we report the experience of a CBSS in rural Cambodia, including its development and feasibility, its performance in terms of sensitivity and positive predictive value, and its usefulness to local health services.
Materials and methods
A CBSS has been developed and implemented since July 2000 in seven rural communes located in four provinces in Cambodia (Figure 1). The four provinces chosen for this pilot study were among those in the Border Malaria Control Project funded by the European Commission, which provided financial support for the surveillance system. The seven communes comprised 52 villages and had a total population of about 30 000 inhabitants in the year 2000. They were served by four health centres, each with a catchment population of 7000–10 000, and by four referral hospitals, each serving about 10 such catchments. Other providers of health care included traditional healers, private practitioners, Traditional Birth Attendants, and drug sellers.
The events to be reported by the CBSS were identified through discussion with health staff and Village Health Volunteers (VHVs) based on their public health importance, severity and potential for an outbreak as well as the existence of a control programme. They included malaria, chronic cough, acute severe diarrhoea, measles and haemorrhagic fever, and births and deaths. A standard case definition was used throughout the system to collect data at the village level. It was adapted from case definitions used at the health centre level:
Suspected malaria: Any person with high and intermittent fever associated with chills. Separate episodes were considered as different cases.
Suspected measles: Any child (under 15 years) with fever and maculo-papular rash and any of the following: cough, runny nose, or red eyes.
Severe acute diarrhoea: Any person aged five years or more with acute watery diarrhoea of more than three motions a day and severe dehydration characterized by sunken eyes and intense thirst. Here too, separate episodes were considered as different cases.
Haemorrhagic fever: Any child with high and persistent fever of abrupt onset, associated with maculo-papular rash and petechiae/gingival bleeding/bloody stool.
Chronic cough: Any person with cough for more than 21 days. VHVs have to report the same case every month until the patient is cured or has died.
Cluster of cases: A group of five or more similar cases occurring unusually closely together in any village within a week.
Data on these events were collected by VHVs and reported to data collation and analysis teams based in health centres. VHVs reported immediately any clustering of cases (more than five cases within a week) and deaths because of acute diarrhoea to the data collation and analysis team. Health centre staff considered such events to be potential outbreaks and reported them immediately to the staff based at operational district offices who investigated and took measures to control outbreaks. The VHVs also reported every month the total number of cases of each event included in the surveillance system using a monthly report form (Figure 2). This was done in a monthly meeting of VHVs and health centre staff in which the data were collated and analysed, and remedial actions discussed. The collated data were reported to the team in the Operational District office who gave feedback and investigated any unusual increase in the number of cases and neonatal deaths.
At least one VHV was enrolled per village, either selected by health centre staff or elected by the villagers. Health staff involved in the CBSS included three staff at each health centre, two at each Operational District and one at each Provincial Health Department.
A series of 3-day initial training workshops was held for both VHVs and health staff at each project site shortly before the implementation of the system. It was followed by a monthly half day of refresher training separately for VHVs and health staff and further training in collation and analysis of data for the health staff. The training of VHVs focused on disease recognition using a syndromic approach and on methods for prevention. Slides and videocassettes were used to train VHVs to recognize diseases and events accurately, and to contribute better to disease prevention and control.
A household survey was conducted in July 2001 in all villages in three of the seven communes, in order to obtain village-based data to validate VHVs’ case reports of diseases and other health-related events during the preceding month (for cases of disease) and year (for vital events). The CBSS standard case definitions were used. Causes of deaths were assessed by verbal autopsy based on the standard CBSS definitions; deaths of infants under 28 days were investigated for neonatal tetanus using a standard verbal autopsy questionnaire, and so were deaths suspected to be because of measles. Each case and vital event reported during this survey was then matched to the CBSS data using household identifiers. Matching of reports was carried out in the field, when the VHVs were present for clarification if needed. Survey and outbreak investigation data were taken as the reference value. Cases reported by VHVs that matched those detected by the household survey or an outbreak investigation were considered as ‘true positive’ for the estimation of the sensitivity and positive predictive value of the CBSS.
Table 1 shows the total number of cases of malaria, measles, severe diarrhoea and haemorrhagic fever as well as person-months with chronic cough, as reported by the CBSS from September 2000 to August 2002. Only a third of malaria, chronic cough, and haemorrhagic fever cases, just over a quarter of severe diarrhoea cases, and less than one in 20 cases of measles had contacted a health facility.
Table 1. Number of cases of diseases* reported by CBSS, Cambodia September 2000 to August 2002
Treated at health facility, n (%)
Treated at home, n (%)
Total cases, n
* For chronic cough, the table shows person-months with cough. For malaria and diarrhoea, separate episodes are considered as different cases.
The monthly incidence of malaria, severe diarrhoea, measles and haemorrhagic fever reported by the CBSS from September 2000 to August 2002 is shown in Figures 3–5, with comparison between the total cases (including cases treated at home and health facilities) with those treated at health facility alone. With the exception of July 2001, the total monthly incidence of malaria had steadily declined from over 250 cases in September 2000 to around 100 cases in August 2002. The decline of malaria cases reported by the CBSS as treated at health facility alone over the same period was substantially less marked than that the total malaria incidence, which also reflects a relative increase of the use of public health facility for malaria treatment. The CBSS data also show that a relatively small number of cases of haemorrhagic fever, measles, and severe diarrhoea had contacted a health facility.
CBSS data show that 95% of births and deaths occurred at home. Home deliveries were assisted almost exclusively by Traditional Birth Attendants (TBAs) and 90% of perinatal deaths occurred at home. Most deaths because of malaria, chronic cough, diarrhoea, haemorrhagic fever and measles also occurred at home (Table 2).
Table 2. Place of births and deaths reported by CBSS, pilot study communes, Cambodia September 2000 to August 2002
At health facilities, n (%)
At home, n (%)
Causes of death
All other causes
The infant and under-five mortality rates, calculated from data generated by the CBSS from all project communes in the second year of project implementation, were 72.9 and 89.0 per 1000 live births respectively, slightly lower than those of the previous year which had been 80.0 and 107.5 per 1000 live births respectively (Table 3).
Table 3. Infant and under-5 mortality rates, all pilot study communes, Cambodia 2000–2002
Number of live births
* Rate per 1000 live births.
Year 1 (September 2000 to August 2001)
Year 2 (September 2001 to August 2002)
Detection of outbreaks
From August 2000 to September 2002 two outbreaks of malaria were detected in a commune; seven outbreaks of severe acute diarrhoea were reported in two communes; 10 outbreaks of measles were detected in three communes; and two clusters of haemorrhagic fever cases were reported in one commune. All these, with the exception of one measles outbreak, were confirmed to be true outbreaks.
The information obtained by the CBSS drove local health staff to respond rapidly and investigate outbreaks. For instance, in Chan Mul commune, when reports of a suspected outbreak of measles were brought to the attention of the commune health centre, a team of health centre staff immediately went to the affected villages to verify the reports and subsequently undertook necessary control measures as well as further investigations of the outbreak.
Accuracy of the reported cases of common syndromes and vital events
The sensitivity of VHVs’ reporting of cases of communicable disease ranged from 65% for malaria to 93% for measles. The positive predictive value (PPV) of VHVs’ report on communicable diseases ranged from 82% for severe diarrhoea to 90% for measles. Concerning vital events, 76% of women 7–9 months pregnant and 82% of births in the three surveyed communes were reported by VHVs (Table 4). The CBSS detected 95% of all deaths in the three communes during the year prior to the survey.
Table 4. Sensitivity and positive predictive value (PPV) of the CBSS, household survey of three communes, July 2001
Cases detected in survey*
Cases reported by CBSS
True positive cases
* Gold standard for estimating sensitivity and PPV of CBSS: Household Survey for malaria, chronic cough, and pregnancy (recall period 1 month) and outbreak investigations for measles (1 year).
Measles (7/2000 to 6/2001)
Pregnancy (7–9 months)
A large proportion of cases of major infectious diseases did not seek medical treatment at public health facilities and were therefore not reported by the health facility-based surveillance system. Health facility data therefore could not allow the detection of all measles, severe diarrhoea or haemorrhagic fever outbreaks. They also could not capture all births and deaths in the rural areas because the majority of these vital events occurred at home. In contrast, the CBSS captured more comprehensive and representative data for major communicable diseases and detected disease outbreaks more frequently and more rapidly than the routine disease surveillance system.
An additional factor is that in Cambodia, health staff sometimes inflate the number of cases they have treated in their routine monthly reports, in order to receive additional quantities of drugs and supplies, either for private practice or to replace date-expired stocks. Surveillance data collected by the CBSS are not subject to such bias.
Most importantly, the CBSS provided a forum for information sharing and joint decision-making between local health staff and VHVs, which led to better collaboration and active community participation between these two sets of key players in disease control and prevention. The system also empowered the local health staff and community in disease surveillance and outbreak response, based on which they could take effective follow-up and corrective action.
A tendency for malaria and diarrhoea incidence to decline with time can be seen in Figures 3 and 4. Both tendencies can also be seen in the health facility data. This, and the maintained level of reported incidence of measles and haemorrhagic fever, supports the view that any such decline is not attributable to a fall in sensitivity of the CBSS. It would be tempting to attribute the decline to improved preventive interventions and outbreak response, but such year-on-year variations are common in infectious disease surveillance. A longer time series, or reliable data from non-CBSS communes, would be required to confirm that the CBSS had contributed to a sustained decline in disease incidence.
Factors underlying the performance of the CBSS
The high performance of the Cambodian CBSS is linked to the importance of events monitored, the system design and its key players. Events monitored by it were multiple, important, relevant and relatively easy for local people to identify. The diseases or syndromes to be reported were locally the most important communicable diseases in terms of severity, burden or epidemic potential. They are all targets of national control programmes, and effective control measures are available. Births and deaths, on the contrary, constitute important and relevant information needed for appropriate planning of disease control and prevention activities as well for monitoring infant and child mortality, as birth and death registration are not available in Cambodia. All these events have elicited the interest of health staff and VHVs, who are the end users of the data they collect.
The system design, including two-way flow of information, instant feedback, local use of data, and simplicity as well as its decentralized management contributed to the success of the operation of the CBSS. The monthly feedback meeting is a central feature of the Cambodia CBSS and is crucial for its success. It enables information to be fed back to all participants of the system and decisions to be made to address identified issues within the same day of data reporting and minimizes the related workload of the health staff as well. This process challenges all participants to take necessary remedial action together, the results of which can be closely monitored by them at the next monthly feedback meeting. The system has, therefore, overcome constraints that have hampered many previous CBS systems including delayed feedback and non-participation of local health staff and communities in data analysis, decision-making and action-taking. Additionally, the feedback meeting provides an opportunity for continuing training of VHVs and health staff, thereby contributing to the improvement of the system.
The Cambodia CBSS, in which data are actively collected through periodic home visits, yields a higher proportion of cases reported than passive data collection surveillance systems. The use of tally sheets to record and report events is most appropriate for semi-literate VHVs in remote areas and helps to reduce errors in data collection. Furthermore, a CBS system which is developed with local participation and locally managed is likely to be more effective than vertically-run programmes and projects developed and implemented by national, international or non-governmental organizations.
Finally, another contributing factor in the success of CBSS is the motivational mechanism to maintain VHVs’ voluntary work. It included travel costs, per diem and food given to VHVs who attended their monthly meeting with health staff at the health centre, free medical care, training, rewards and recognition as well as the immediate response of health staff to the information reported by VHVs and ‘instant’ feedback of information to them. As a result, VHVs were willing to travel quite frequently to collect data in their own village as well as to travel by foot or motorbike to the health centre, located typically 4–11 km from their home, for the monthly meeting, which almost all regularly attended.
The VHVs were sometimes unable to collect data from people living too far away from their house, especially those temporarily moving to camp by their farm land during the rice-planting and harvesting seasons. For instance, the majority of households not visited by VHVs in June 2001 were located further than 1.5 km from their house. Also, the response of health staff to the information generated by the CBSS was not yet optimal in many project communes because of low salaries and inadequate funding for health services.
Costs and sustainability
The annual cost of the Cambodia CBSS was about US$0.5 per capita including occasional visits from Phnom Penh for training, supervision, and evaluation. This cost would be lower by half if the system were operated by the District alone; that is, without the research and development input by one of us (SO). It appears to be lower than that of many similar systems running in developing countries to date, because of its use of existing health infrastructure and staff, which costs much less than projects run by non-governmental organizations where additional staff and facilities have to be funded (O'Neill 1993; Cairncross et al. 1997). The amount of staff time required to manage the system was relatively small; 1 day per month for the monthly meeting, plus a half day of training, and time spent on outbreak response, if any. The VHVs spent 3–4 half days per month on home visits for data collection, in addition to their day at the monthly meeting.
In 2004, the system proved to be replicable as it has been implemented, with support from Save the Children Australia, throughout an operational district of 100 000 inhabitants. It is being implemented in two more districts, and a fourth is planned for early 2005.
The Cambodian CBSS has many attributes that could make it more viable – that is sustainable – than other CBSSs. First, VHVs and health staff have the capability to run and manage the system by themselves, with little technical or supervisory support from the central level. Second, the system is built on the existing health system and resources, following the Ministry of Health policy and strategy to strengthen the Operational District structure. It gives the Operational District and health centre a mechanism to fulfil their role in disease prevention and control in the communities. Third, the Cambodian CBSS comprises mechanisms to maintain VHVs’ motivation to continue the CBSS. The mechanisms include continuous training, supportive supervision, health care benefits, work recognition, ‘instant feedback’, and involvement in data analysis and decision-making.
We conclude that a community based surveillance system run by VHVs and local health staff is feasible and that this system can produce useful information for monitoring trends and to identify potential outbreaks of common infectious diseases.
This study would not have been possible without financial assistance from the European Commission and Cambodia's Ministry of Health for which we are very grateful. We wish to thank in particular Dr Frederick Gay, Dr Andrew Corwin, Ms Sarah McFarlane, Dr Kyi Minn, Prof Felicity Cutts, Ms Susanne Wise, Dr Gertrud Schmid-Ehry and many other people for their encouragement and inputs during the development and refinement of the CBSS in Cambodia.