Computer networking as a tool for public health surveillance: the French experiment.
Surveillance and control of communicable diseases require the following steps:
* timely collection of relevant and appropriate epidemiologic data;
* timely and meaningful statistical and mathematical analysis of these data; and
* rapid dissemination of the results to all who need to know (including decision makers, health-care providers, public health officials, such as the reporters of data, and the general public).
Computer networking is useful in accomplishing these three steps that encompass surveillance. It connects all members of the system, including staff in geographically remote facilities, and makes available these resources:
* large epidemiologic data bases, which would otherwise be inaccessible to all but a few of the participants in the surveillance system;
* powerful data-management capability and highly specialized software for epidemiologic and modeling applications; and
* communications facilities (e.g., electronic mail and computer bulletin boards) that allow ready interaction between the diversified participants in the system (e.g, general practitioners [GPs], staff in biological laboratories, hospital staff, and officials in public health offices).
Beginning in 1984 in France, we started to develop a National Information System devoted to surveillance of communicable diseases and based on computer networking. One experience with this system is described in more detail below.
DEFINITION AND HISTORY
In 1983, the INSERM (Institut National de la Sante et de la Recherche Medicale--the French equivalent of the U.S. National Institutes of Health) and the DGS (Direction Generale de la Sante--the National Department of Health) jointly sought to promote new initiatives to enhance the quality of epidemiology and surveillance of communicable diseases in France. The notifiable diseases reporting system, which included more than 30 diseases, was producing results that were not uniformly reliable.
My office issued a report that suggested setting up a computer network aimed at linking all the partners involved in the surveillance, control, and epidemiology of communicable diseases not currently in the reporting system (e.g., urethritis among males) or for which data obtained through the notifiable diseases reporting system were not satisfactory (e.g., for measles) . This report proposed building a national information system and stated three goals:
* to improve surveillance by incorporating new data providers (sentinel GPs) and by promoting wide access to existing surveillance data bases;
* to provide support for an alarm system; and
* to provide support for epidemiologic investigation by giving investigators
access to the data bases and software they needed.
The French Communicable Diseases Network (FCDN, and in French: Reseau National Teleinformatique de surveillance et d'information sur les Maladies Transmissibles, RNTMT) was established on November 1, 1984.
DESCRIPTION OF THE FRENCH COMMUNICABLE DISEASES NETWORK
The FCDN was conceived in 1983 as a national communicable disease information system with a surveillance component. The basic idea was to use computer networking to link all health-system partners who were responsible for recording surveillance data and who might be interested in using data on communicable disease.
This concept was based on several assumptions. One is that effective surveillance (and control) of any communicable disease takes into account all the participants who must be in contact through the system (e.g., GPs to detect clinical cases, staff in diagnostic laboratories to validate the cases, epidemiologists to analyze the data, and health officers with the authority to make decisions).
Another assumption was that the networking of all these partners would create new dynamics. For example, within the network, different local health officers would compete to provide better data from their region because their results could be compared in real time with those of all other participants in the system. Sentinel GPs could know the results of the notifiable disease reporting system in real time (unlike the general population of GPs, who probably do not receive the epidemiologic bulletins that contain this information). With this information, these health-care providers may stimulate interest in the notifiable disease reporting system among their colleagues.
Eight years into this project, the system continues to operate under these assumptions. It currently links 1,500 partners, has accumulated 200 megabytes of data, and provides its users with 5,000 videotext pages of information.
The French Communicable Diseases Network partners (general)
The 1,500 users of the FCDN access the system through the host computer located at the URBB (Unite de Recherches Biostatistiques et Biomathematiques). Most of the sentinel GPs now use the MINITEL, a terminal supplied by the French Telephone Company and widely used throughout France. These participants in the system can be classified into three groups, as described below.
The first group comprises persons who conduct the surveillance on which the FCDN is based. The first component of the system is operated by the URBB and links approximately 500 sentinel GPs. These sentinel physicians monitor seven communicable diseases and the results of tests for human immunodeficiency virus (HIV). The second component of the FCDN is operated by the National Department of Health, Office of Communicable Diseases. It links the 100 departmental health offices that collect reports of the 20 currently notifiable diseases in France . The offices forward these reports to the National Department of Health by filing electronic forms each week. This means that the data base is constantly being updated. The third component of the system is operated by the National Health Laboratory. This laboratory uses the network to link 59 hospital laboratories that monitor listeriosis, Haemophilus infections, and meningococcal infections. Each week, these laboratories fill out computer-based forms reporting any isolates of the organisms under surveillance .
The second group of network partners can be described as "information providers." Although they do not use the network facilities for surveillance or epidemiologic activities, they can provide information of interest to other network users. Examples of members of Group 2 are the National Veterinary and Food Research Center (in French: CNEVA), which regularly updates an electronic bulletin on rabies; several national reference centers; and the European center for acquired immunodeficiency syndrome (AIDS) epidemiology, which provides specific information on AIDS.
The last group of network partners includes persons who read and use the information provided by other participants. The 100 departmental hygiene engineers are such partners.
The sentinel general practitioners
The sentinel GPs participate on a voluntary and unpaid basis. As of January 1, 1992, there were 500 of them. The original sentinel GPs, in 1984, participated in a system that covered only influenze . Since that time, sentinel GPs have been recruited in a collaborative effort between FCDN staff and the French medical journals.
The FCDN regularly publishes its results in the most widely read French medical journal (Le Quotidien du Medecin). Readers of the journal may then be stimulated to join the FCDN; priority is given to applying physicians in regions/departments in which sentinel GPs are currently underrepresented. When a sentinel GP is recruited, his/her demographic characteristics are recorded (e.g., age, seniority in medical practice, individual or group practice, and place of practice). The characteristics are further used to allow weighting in order to provide surveillance results standardized according to the demographic characteristics of the overall population of approximately 50,000 French GPs. This weighting is necessarily dynamic, because the demographic pattern for the GPs varies over time. For instance, when a sentinel GP takes a holiday, the geographic structure of the set of sentinel GPs changes, and a correction factor must be applied.
In practice, sentinel GPs are required to connect to the FCDN at least once a week. This frequency of contact is necessary in order to identify inactive sentinel GPs (versus active sentinel GPs who access the FCDN to report "zero" cases, as appropriate).
The FCDN is based on a videotext server and a relational data base management system. The videotext server is the front end of the information system; it allows for information exchange, data entry, and synthetic information retrieval. The relational data base management system allows the user to manage and consult the surveillance data base. Workstations with sophisticated graphics capabilities provide representational material, particularly on the dynamics of epidemics [4,5].
The primary software packages used in the FCDN include VTX-VALU for the videotext applications, ORACLE for data base management, MATHEMATICA for devising graphics tools, and SAS for statistical analyses.(*1) Details of the organization of the system are provided elsewhere . The current FCDN system is the third to have been implemented during the 7-year history of the project. Progress in computer technology has been so rapid that the system has had to continue to evolve. However, because FCDN operates 24 hours every day and because all changes in the system must be easily assimilated by participants, the computer team constantly faces a difficult situation.
Data collected by the surveillance operators of the FCDN are entered through the user terminals via videotext data entry applications. These applications include automatic checking for consistency according to the rules established by FCDN epidemiologists. All files are updated daily. Special software has been devised to raise the level of security of the system to protect against such problems as electrical failures.
EXAMPLES OF RESULTS
Three different examples of surveillance activities conducted with the FCDN are described below. The first report category, measles, provides an example of a condition with a clearly stated mandate for control and prevention. The second, HIV testing by GPs, constitutes a "continuous watch" of a public health problem of great interest to decision makers. The third, influenza-like illness (ILI), is being monitored but without a clear definition of the objectives of the surveillance.
Figure 1 shows the overall results of surveillance for measles in the FCDN; note marked seasonality, with peaks in late spring . The number of cases of measles per year was approximately 150,000 in 1991. (In the same time frame, the number of births per year in France was approximately 700,000.) These figures show that efforts to promote measles vaccination clearly need to be continued. These somewhat unsuccessful results might well be related to the perception in France that measles is a trivial disease. Our survey of a panel of GPs showed that measles ranked eighth in the list of communicable diseases when GPs were asked whether they believed that surveillance of these diseases was important for public health. This attitude was strongly correlated with age, with older GPs according much less importance than younger GPs to the surveillance of measles. This finding also supports the value of a national information system (including measures of the health professionals' attitudes) to allow surveillance data to be interpreted correctly.
HIV testing and surveillance by sentinel general practitioners
Since March 1987, sentinel GPs have reported on their use of testing for HIV (8). Data indicate a) whether the patient or the GP originates the request for testing, b) the reasons given for testing, c) age, d) gender, and e) results of the test.
The system has shown the prevalence of the use of HIV testing by GPs; approximately 25% of all HIV tests performed in France since 1987 have been ordered by GPs. This figure excludes routine HIV testing done in blood banks and was computed as the ratio between the total number of HIV tests prescribed by all GPs--as extrapolated from the numbers for sentinel GPs--and the total number of HIV tests sold in the same period.
The FCDN system has shown a continuous trend for more HIV testing in the "general population," beyond the bounds of the usual risk groups, which in 1987 accounted for almost all use of HIV tests. At present, >50% of HIV tests are ordered by a GP after the patient requests testing; approximately equal numbers of males and females were tested in 1991. At present, women are even more likely to be tested than men. The system has also been able to detect "epidemics of HIV tests." Such an "epidemic" occurred in November and December 1991, after numerous articles appeared in the media about the risk of HIV after blood transfusion.
Monitoring HIV testing by the French sentinel GPs certainly qualifies as "surveillance," because it is clearly a "continued watchfulness" and because its aim is to detect--and interpret--trends. These data have been quite useful--from a public health point of view--during an ongoing scientific and political debate on the desirability of mandatory testing for HIV, at least in such situations as premarital or preobstetrical delivery check-ups. The system showed what tests were actually done and trends in the test results in prenatal and perinatal situations.
Surveillance for influenza-like illness (ILI)
Since the FCDN system began operating, sentinel GPs have reported approximately 140,000 cases of ILI. The eight outbreaks that occurred in the same period are graphed in Figure 2 (national data). The FCDN announces the place and time of each year's outbreak, describes the pathway of spread, evaluates the total number of patients, and does cost estimates (e.g., in terms of lost work days) for the National Health Insurance System.
The two French World Health Organization (WHO) influenza reference centers describe the nature of the virus that is currently circulating. Specific surveys have also been done through the FCDN network, in collaboration with the National Health Insurance System, to evaluate vaccine coverage, because most vaccines are administrated by GPs. Since an outbreak of ILI actually reflects a mixture of outbreaks caused by different agents, strategies to control ILI should take all these agents into account. However, if a surveillance system is to be used to provide an analysis of an outbreak by virus, even more than 140,00 reported cases would be needed in order to obtain meaningful description of the epidemic by age, place, and other descriptive factors. Such surveillance would definitely be costly and would probably be difficult to perform with a reasonable degree of confidence in the representativeness of the results.
We have ascertained that sentinel GPs would not be willing to perform the necessary diagnostic tests on all of their patients with ILI. Even if they did so, it would be difficult to make immediate use of the results--e.g., it would not be possible to modify the current vaccine during an epidemic even if we did identify and characterize a new type of virus not included in the vaccine. More precisely, our detailed, quantitative description of an epidemic would be no more useful than the qualitative observation that a particular virus is circulating.
Finally, our experience with monitoring ILI has shown that the interesting, practical result of such surveillance is rapid access to information. Timely information is important for GPs, who claim that knowing in advance about outbreaks in their region alerts them to the need for diagnosis and helps them to plan for dealing with the influenza outbreak. The data base on ILI has been shown to be useful in terms of straightforward epidemiologic research that may lead to practical warnings to GPs. A recent project, for example, demonstrated the temporal relationship between ILI and meningococcal disease , which occurred more frequently and more severely after an outbreak of ILI.
DISSEMINATION OF INFORMATION
Computer networking allows rapid dissemination of results to all network users; indeed, this characteristic was one of the reasons we chose this technology. Our experience shows that network users need to receive reports from the system on paper and not just in electronic form. In addition, surveillance results needed to be more broadly disseminated than just among network users.
Network users can access roughly 5,000 pages ("screens") of information. These pages are organized hierarchically and can be accessed through successive menus with multiple choices, or directly with keywords. Most, but not all, of the pages can be read by all users. For example, the National Department of Health may want to redistribute information specifically to its departmental health officers. Technically, each page of information can be indexed on the corresponding passwords of the user, or groups of users can be allowed to read it.
An electronic bulletin board contains surveillance bulletins and information bulletins. The FCDN system includes three electronic surveillance bulletins corresponding to the three groups of surveillance operators.
For each of eight health problems under surveillance, the sentinel GPs' electronic bulletin plots incidence over time for the most recent 28 weeks and the geographic distribution of the incidence observed during the previous week. Incidence rates are weighted to the national distribution of GPs. The mean delay between collection and distribution of data is 8 days.
The notifiable diseases electronic bulletin is updated each week. For each French department, it gives the number of cases reported during the previous week and in the current year for 16 of the 20 notifiable diseases.
The data base for the other notifiable diseases is maintained on the host computer and can be accessed by health professionals and downloaded on their personal computers. Because these conditions are uncommon, we have not maintained an electronic bulletin for them.
The hospital laboratory network redistributes, in a similar form, surveillance data on time and place associated with listeriosis, Haemophilus infections, and meningococcal infections.
Surveillance is an important component of the information system in public health. It is the reason the system was conceived as providing general public health information, which includes the following categories:
* administrative information: e.g., recommended schedules for various vaccinations, lists and address of all the centers that provide confidential, free testing for HIV, and copies of relevant legislative acts.
* epidemiologic information: e.g., the national department of health and the WHO AIDS Collaborating Centre for Europe have provided a base of approximately 1,000 videotext pages of information. Similarly, the National Veterinary and Food Research Center updates a bulletin from French veterinary laboratories that deal with cases of animal rabies.
* continuing education information: this section covers announcements for meetings on communicable disease epidemiology and now includes information specifically oriented toward GPs in the field of "rare and severe communicable diseases" (e.g., anthrax).
Dissemination through paper bulletins and journals
Quarterly, sentinel GPs receive news of the network by mail, including copies of reprints and, eventually, details on specific surveys in which they have participated.
Efforts have been made to ensure the dissemination of surveillance results of all network participants. Periodically (roughly, quarterly), an update of the surveillance activities of the network is published in the Bulletin Epidemiologique Hebdomadaire, a journal published by the National Department of Health and distributed to all public health professionals.
Dissemination to all French GPs--not just sentinel GPs--is essential not only because GPs are both the source and the target of the public health problems monitored by the FCDN but also because it provides the opportunity to recruit new sentinel physicians into the surveillance system.
As a result of an agreement with Le Quotidien du Medecin, the journal most widely read by GPs, a summary of results by FCDN is published weekly. Finally, the FCDN issues a quarterly bulletin (Sentinelles), which is distributed free of charge to all GPs in France.
In the past 8 years, the FCDN has proved that computer monitoring can be a useful and efficient tool for public health surveillance. Each year, the FCDN has been able to detect influenza outbreaks without delay, to monitor changes in the incidence of measles and mumps, to describe changes in physicians' and patients' attitudes toward HIV testing, and to provide new results or new epidemiologic data on diseases being monitored.
As long as surveillance data continue to accumulate on the hard disks of the host computer (presently, the size of the surveillance data base is >200 megabytes), the system can allow many more health professionals to have inexpensive access to these data, just by downloading the information to their own personal computers.
Some of the initial goals of FCDN have not yet been met, and it may be that more effort must be placed on educating potential participants in the system about the benefits and efficiency of the surveillance system. Substantial human resources would certainly be required for this task. Another problem FCDN has experienced is that although some borders that should be crossed are not, others were crossed that should not have been--e.g., civil servants in different departments have used the convenient e-mail to contact one another directly instead of following the standard operating policy of transmitting information hierarchically. A less desirable example is when a network participant disseminates an "alarming" observation (a physician encountering such a case is supposed to refer it to the local health officer, who is responsible for its validation). The present system of information and/or communication needs to be monitored and reassessed constantly in order to forestall problems.
In order to continue to grow and remain effective, managers of the system should focus their efforts on three points:
* Placing more emphasis on the education of users of the system and on the mainstreaming of the FCDN system into the general public health system.
* Implementing decision-support systems oriented toward the surveillance of rare but severe communicable diseases. Such systems would include modules to help physicians recognize these diseases and know what follow-up actions are appropriate (most often, referral to specialized hospitals and timely dissemination of information about the condition).
* Improving the quality of surveillance data, in terms of collection techniques, analysis, and effective dissemination of information.
(*1) VTX-VALU (Digital Equipment Corp.); ORACLE (Oracle, Inc.); MATHEMATICA (Wolfram Research, Inc.); SAS (SAS Institute, Inc.).
[1.] Valleron A-J. Project pour le developpment de la recherche et de l'information sur les maladies transmissibles. Rapport a la Direction Generale de la Sante. Paris, 1984.
[2.] Hubert B, Laporte A, Lepoutre A, et al. Le point sur la surveillance des maladies transmissibles en France. B.E.H. 1991:36;155-6.
[3.] Menares J, Reyes E, Martin-Bouyer G. Surveillance de la grippe par les medecins correspondants. France: Septembre 1980--Juin 1981. Rev Fr de Sant Publ 1982:19;77-87.
[4.] Garnerin Ph, Vidal J, Valleron A-J. A visualization tool for representing epidemic spread. Mathematica J 1991:1;59-61.
[5.] Carrat F, Valleron A-J. Epidemiologic mapping using the "Kriging" method. Application to an influenza-like illness epidemic in France. Am J Epidemiol 1992 (in press).
[6.] Garnerin Ph, Valleron A-J. The French Communicable Diseases Computer Network: a technical view. Comput Biol Med 1992:22;189-200.
[7.] Mary M, Garnerin Ph, Roure C, Villeminot S, Swartz T-A, Valleron A-J. Six years of public health surveillance of measles in France. Int J Epidemiol 1992:21;163-7.
[8.] Massari V. Fagnani F, Landman R, Valleron A-J. Prevalence of HIV infection and cost of medical follow-up for asymptomatic seropositive followed in general practice in France. Health Policy 1991:19;159-66.
[9.] Hubert B, Watier L, Garnerin Ph, Richardson S. Meningococcal diseases and influenza-like syndrome: a new approach to an old question. J Infect Dis (in press).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Proceedings of the 1992 International Symposium on Public Health Surveillance; A.-J. Valleron and Ph. Garnerin speeches|
|Publication:||Morbidity and Mortality Weekly Report|
|Date:||Dec 1, 1992|
|Previous Article:||Informatics in public health surveillance: current issues and future perspectives.|
|Next Article:||Disease surveillance in China.|