Surveillance is fundamental to public health decision-making and subsequent action. Sustained and integrated global epidemiological surveillance has been weak in Madagascar, similar to that in other developing countries . Sentinel surveillance systems offer advantages over passive surveillance, which is known to have limitations due to incomplete reporting . The major objective of sentinel surveillance was to identify illness clusters early, before diagnoses are confirmed and reported to public health agencies, and to mobilize a rapid response, thereby reducing morbidity and mortality. To ensure efficiency and accuracy, sentinel systems require strong communication systems.
GPs participated in sentinel surveillance on a voluntary basis; therefore, the system is not representative of the whole country. To ensure better representation, the number of participating GPs should be increased. On the all 1500 health care centers directed by an physician, only 13 were included in the network. However, it was also important to ensure that sentinel GPs are easily accessible to surveillance staff. All sentinel centers were located in urban areas where population density was higher and where outbreak impact should be very dramatic. They also had been chosen to give an estimate picture of the different climate trend of Madagascar. The estimated population covered by all sentinel sites is 500.000 inhabitant representing as of 2008 3% of the Malagasy population.
Supervision of the surveillance system was carried out by MOH as planned. Stakeholders were made aware of the advantages and limitations of sentinel surveillance systems. Sentinel surveillance systems may enhance collaboration among health ministry services and health-care practitioners. However, they do not replace traditional public health surveillance, nor are they substitutes for direct medical analysis, in which the physician reports unusual or suspect cases of public health importance.
Establishing sentinel surveillance was a difficult process, even if it was less costly and used fewer resources than passive surveillance. The problem, among others, in establishing such a system involves connecting GPs to the sentinel system and coordinating their work. We selected mobile phone communication for daily reporting and paper forms for weekly reporting, increasing the human resources required to carry out this type of surveillance. In the future, with the increase in mobile communication in Madagascar, in particular that involving 3G internet, sentinel surveillance could be based on electronic records delivered directly from the practitioners.
Although the cost of data transmission on a daily basis is minimal, at less than 1US$ per month per sentinel center, the costs and maintenance of the system require better quantification, both in terms of resources spent (time of the SGPs,...) and person-hours involved in responding to system alerts. Furthermore, the importance of malaria in Madagascar and other sub-Saharan countries requires that a rapid malarial test be linked to fever syndrome surveillance.
The mobile phone with its low cost and universal availability has been recognized as an important piece of communication technology in health care . However, published studies on using mobile phones are still limited  to countries where limited resources and remote locations are the main reasons for using mobile phone.
Sentinel surveillance systems seek to use existing health data in real time (daily) to provide immediate analysis and feedback to those charged with the investigation and follow-up of potential outbreaks. However, data relating to these various syndromes are nonspecific by nature. For illnesses that are self-limiting and of a short duration, resolving the causes of the syndrome is not usually of direct benefit to the patient, is not a priority for the clinician, and is not always feasible with current technology. The advantage of using syndromic data for outbreak detection is speed of response . Experiences to date indicate that this advantage may only be theoretical. The time required to conduct investigations and retrieve diagnostic and epidemiological information might negate the advantage of quick data acquisition. The absence of sustained syndromic signals usually provides greater reassurance that an outbreak does not exist than the information obtained by immediate investigation.
The rapidity of the system, although excellent compared with traditional surveillance systems, needs to be improved. This Malagasy sentinel surveillance system did detected fever outbreaks in areas where traditional surveillance system didn't. It also highlighted the difficulty in building epidemiological baselines without historical references.
Systematic methods for determining the expected number of visits on a particular day require historical data to create baselines in trimmed-mean seasonal and autoregressive integrated moving average (ARIMA) models. Time series methods are an important tool to provide alarm thresholds. These forecasting models could produce results with good accuracy in their predictions of the data .