9. Epidemiological Information Systems

There can be no doubt that an information system is vital for the proper functioning of any programme. Information systems have been designed to provide the managerial and epidemiological data necessary to monitor the impact of interventions on the malaria problem and the implementation of programmed activities. Such information systems have led to the belief that the main causes of changes in the epidemiological situation are the control interventions themselves, and that the only response to an unexpected deterioration in the situation is to intensify those interventions. A broader look at concomitant processes might have suggested a more critical determinant factor and a more effective control policy.

While information is essential for the management of any control activity, it must be kept in mind that the objective is control and not the acquisition of information for its own sake, no matter how interesting it may be. During an epidemic, the first priority should be the appropriate care of the cases, which in most instances will require an improvement in the accessibility and efficiency of diagnostic and treatment facilities. This must take precedence in the distribution of resources and an important aspect to consider is the optimization of the deployment of microscopical diagnostic facilities.

At present, most of the established antimalarial programmes are reorganizing their information systems in response to the emphasis on disease management required by the Global Malaria Control Strategy, in contrast with the almost exclusive attention previously given to parasite infections during the eradication era. There is therefore a need for certain redefinitions of reportable variables and a search for relationships between the new and the old variables. This is particularly important for trend analyses, which will require the establishment of relationships between old and new time series of not fully comparable variables. It will also be necessary to collect information from general health services that had not previously been received by malaria programmes.

The new emphasis on disease management requires the separate monitoring of 'clinical malaria' and 'laboratory confirmed cases' reported by formal health services and by lay community health workers. It can be assumed that reported 'clinical malaria' will be more or less equivalent to the 'slides collected' previously by formal health services and by community workers. Both time series can be compared to test their hypothesis and to see how previous definitions of normality may be translated to the varaibles used in the new systems. New time series will have to be started for the new data collected on the main forms of severe malaria, treatment failures and deaths.

An information system should not be limited to the routine reporting and subsequent analysis of data. It is essential that all echelons of the health services should be aware of the importance of particular indicators of risk defined for each area, and that they should be required to communicate abnormalities in those indicators to the level capable of generating an appropriate response as soon as possible. Routine information and trend analyses should then be supplemented by properly planned epidemiological surveys or specific studies conducted by the specialized services to confirm, evaluate or study particular problems or situations, especially when an emerging epidemic is suspected.

All echelons of the services reporting epidemiological information should be considered full participants in the overall process of malaria control. Peripheral services should receive not only consolidated reports but also be able to exchange information with neighbouring areas and receive technical assistance when necessary, so that they appreciate the usefulness of reporting. This should never become a burden and a cause of non-compliance or inaccurate reports. It is also essential to stress, reinforce or stimulate an interest in understanding time and spatial variability and to avoid the tendency to rely only on averages of scattered observations, to extrapolate to large areas or to disregard past information. It is unfortunately still very common for epidemiological analyses to be limited only to the comparison of current data with past year indicators. It is also common to find services that neglect or destroy past reports.

The renewed interest in malaria epidemics has made malariologists aware of the fact that most epidemics are owing to, or greatly influenced by, meteorological or social determinants (Kouznetzov, 1989; Nájera and Beales, 1989), yet most antimalarial services still fail to monitor these variables since the indicators of risk for their epidemic-prone areas have not been determined. Thus in most of the recent descriptions of malaria epidemics, and even in those where exceptionally heavy rains or population displacements are recognized as the main cause, those concerned fail to analyse the details of that relationship, or to show any interest in monitoring such determinants. As a result, they continue to build so-called warning systems based on the classic malariometric variables.

The recognition that the risk of malaria epidemics is in most cases determined by ecological and social variables should lead to their monitoring. Most of the variables concerned are carefully studied by other public services, such as those responsible for agriculture, public works or economic development. Their monitoring therefore requires the involvement of a variety of collaborating partners through the establishment of truly functional intersectoral links. As examples of these situations, graphs 13-15 show how known epidemic years in Khartoum and Wad Medani in Sudan are linked to early deviations from normal Nile river levels and rainfall. Epidemics seem to have been caused by the floods resulting from the joint effects of river overflow and excessive local rainfall, the latter being of particular importance in 1988 in Khartoum and in 1975 in Wad Medani.

9.1. Identification of indicators of epidemic risk

Although a considerable number of determinants may precipitate an epidemic, in any particular situation, most of them are interrelated. Therefore, there is seldom more than one major risk factor or at most only a few with a high predictive value. Nevertheless, in areas of varied and isolated ecological features, such as highland valleys, or of highly concentrated socioeconomic development, there may be large differences in epidemic potential and risk factors between neighbouring areas. The identification of local epidemic-prone areas and risk factors should be based on the retrospective analysis of the malariometric indicators, including the history of past epidemics and the records of potential risk factors suggested by the ecology of the area.

In most antimalarial programmes, the main epidemic risks are known, even if the epidemiological services may not have the recorded information to document that knowledge. For example, in some areas, a major event such as the opening of a road, the beginning of a colonization programme, or the appearance and spread of drug resistance, marks the introduction of new risk factors. These factors may sometimes have been identified several decades previously and their monitoring may have been terminated, as in the case of the Punjab, where a sophisticated epidemic forecasting system was discontinued following the adoption of the national malaria control campaign in the 1950s. In such situations, it is generally possible to obtain the missing records from the relevant services, as well as any information which may indicate an important change in the epidemiological conditions in the area. A simple statistical analysis will then indicate the validity or otherwise of the original epidemiological hypotheses. In areas where there are no reliable epidemiological records but reasonably accurate records of deaths or hospital records, it may be possible to study the correlation between mortality and potential risk factors, as was done in the Punjab in the first quarter of this century (Christophers, 1949).

In other areas, there may be no epidemiological records, but meteorological services may have kept detailed records for many decades, and the ecology of the area may suggest the most probable risk factors. In general, the impact of malaria epidemics is sufficiently serious to remain in people's memory for some years, so that it may be possible to determine whether the years of remembered epidemics were also years of exceptional meteorological or socioeconomic conditions (graphs 13-15).

9.2. Field investigations

In principle, malaria epidemics should be foreseeable, at least from the time of occurrence of the causative disturbances. The occurrence of an epidemic may therefore be considered as:

1. a failure to understand the local epidemiology of the disease, and particularly the interrelation of ecological and socioeconomic processes;
2. a failure to appreciate the consequences of certain economic development or health policies;
3. a failure to translate existing knowledge into appropriate monitoring and preventive activities.

Every epidemic provides an opportunity to refine epidemiological knowledge, detect changes in the geographical limits of the areas considered at risk and, in areas with a risk-monitoring system, find its weaknesses. The follow-up of epidemic control should therefore include the mobilization not only of the resources of the health services, but also those of other sectors and research institutions for the support of epidemiological studies to improve the capacity to forecast and prevent epidemics and the organization of epidemic preparedness.

The recognition of an area as subject to a particular category of epidemic risk gives some indication as to the set of risk factors which should be monitored. However, the further analysis of the local variability of malaria experience and the monitored indicators may show important variations in epidemic risk which should serve to refine the accuracy of the forecasting system. Those events which are not in line with expectations, and in particular unforeseen epidemics, should receive the most intense study. Such studies should be seen not only as means of improving the accuracy of the system and of adapting it to changing circumstances, but also as opportunities for extending intersectoral collaboration to obtain timely information on relevant variables and to improve the speed and relevance of preventive activities.

Historical research plays a major role, not only in helping to provide the most solid link with the studies and tested hypotheses of the past, but also in providing, for some specific areas, historical data which may help to identify associations of past epidemics with certain ecological or social variables, thus suggesting possible etiological relations, and therefore playing a role in the search for useful indicators of epidemic risk.

The study of actual epidemics may also show the importance of certain vectors, hitherto considered as secondary, in the genesis of some epidemics, as compared with the main vectors in the area concerned, e.g., A. annularis in parts of Bangladesh, rather than A. philippinensis (Rosenberg & Maheswary, 1982; Maheswary et al., 1993), or A. albitarsis in some epidemics in Colombia, rather than A. darlingi.

9.3. Geographical information systems

Since epidemics are catastrophic events, their study requires the precise definition of both time and space. Both for the implementation of control interventions and for the understanding of the dynamics of risk, it is necessary to define as accurately as possible the geographical limits of each phenomenon. Most countries have good-quality topographical and political maps, showing the main physical features, the altitude, the hydrological network, the administrative boundaries, the location of villages and population centres, and national and secondary roads. They also have some form of geographical information system, providing information on agricultural areas, the distribution of main crops, irrigated areas, etc. Most of the malarious countries of the Americas and Asia made detailed geographical reconnaissances in support of the spraying operations of the malaria eradication campaigns, which in some cases provided the best available information at the time on population distribution in peripheral areas. In other countries, similar efforts have been made by other services, including the use of satellite information on surface water, vegetation and land occupation and use, of certain areas of interest to hydrological or agricultural services (Thomson et al., 1996).

Computer and communications technology provide cost-effective tools for the establishment of a dynamic information system. The existing geographical information systems (GIS) provide the programmes needed to give the required geographical dimension to epidemiological studies, and their use is spreading rapidly in most countries. Health programmes, particularly those for the elimination of dracunculiasis, the control of schistosomiasis, immunization and, in some areas, malaria control, have started to incorporate GIS in their information systems.

A computer-based GIS allows the collection of data linked to geographical location from different sources, and stores it in a form which permits subsequent analysis and synthetic presentation in map form. It is not only a system for the production of computerized maps but also, and more importantly, for the integration and spatial analysis of data from different sources such as population distribution (location of towns, villages, hamlets and road networks), the environment (physical features, land use, surface water), the location of health and other services (hospitals, dispensaries, health posts, schools, post offices, etc.), epidemiology (morbidity, mortality, drug susceptibility), meteorology (rainfall, temperature, humidity), agriculture (irrigation, main crops, land productivity), the socioeconomic situation and any other subject in which the data are linked to geographical location. Using these databases, it is possible to establish, confirm and monitor spatial and time correlations among the different data, and eventually formulate predictive hypotheses.

The study of epidemic potential should make the best possible use of any existing resources for GIS and stimulate their development where they do not exist. It is clear that GIS should be integrated in the general epidemiological services of the country. It should define the basic geographical unit of analysis of routine health data, the distribution of responsibilities for reporting, epidemiological analysis, and the communications system.

The epidemiologist responsible for malaria, whether in a specialized or an integrated service, should identify the sources of the ecological and socioeconomic data selected for monitoring. Through intersectoral collaboration, it will be necessary to ensure the automatic flow of data after agreement has been reached on the timing and points of reception as well as on their use and distribution.

The strengthening or the development of GIS for epidemic prevention and control will require financial, human and technical resources for the planning, acquisition, installation and maintenance of equipment and software, for training, supervision and evaluation, as well as for the verification, validation, correction and maintenance of related databases.