Standardization and improvement of influenza surveillance: memorandum from a WHO/GEIG meeting.

Standardization and improvement of influenza surveillance: Memorandum from a WHO/GEIG meeting* The main objectives of influenza surveillance are early detection of influenza outbreaks and identifica- tion of the causative agent, collection and analysis of influenza morbidity and mortality data, and collec- tion of influenza virus isolates and analysis of their antigenic characteristics. The true morbidity and mortality from influenza are difficult to estimate in most countries on the basis of only reports of influenza-like illnesses. Analysis and comparisons of the impact of influenza in different countries are also difficult because surveillance systems and methods vary from one country to another. This Memorandum describes generally applicable systems that could improve influenza surveillance. Introduction Influenza virus is unique in that it continually changes its antigenic characteristics. These changes, which do not appear with any clear periodicity, affect all three virus types or subtypes currently cir- culating: A(HINI), A (H3N2), and B. This disposi- tion to change complicates the use of vaccines and other possible control measures and necessitates effective national and international surveillance pro- grammes. The main objectives of influenza surveillance are: to detect influenza outbreaks as early as possible and identify the causative agent (this should per- mit health providers to prepare for a possible increased demand on health care services and warn general practitioners so that they may carry out vaccinations among persons at risk who have not yet been vaccinated); to permit estimates of the impact of influenza by * This Memorandum is based on the report of a Consultation on the Standardization and Improvement of Influenza Surveillance sponsored by WHO and Groupe d'Etude et d'Information sur la Grippe (GEIG), which met in Monaco on 25 September 1991. The participants were: Dr M. Aymard, Lyon, France; Dr P. Chakraverty, London, England; Dr P. Glezen, Houston, TX, USA; Dr C. Hannoun (Chairman), Paris, France; Dr A. Kendal, Atlanta, GA, USA; Dr P. Saliou, Marnes, France; Dr R. Snacken, Brussels, Belgium; Dr M. Sprenger, Rotterdam, Netherlands; Dr B. Stroobant, Brussels, Belgium; Dr P. Tumova, Prague, Czechoslovakia; and Dr J. Watson, London, England. WHO Secretariat Mrs K. Esteves and Dr Y. Ghendon (Secretary), Microbiology and Immunology Support Services, World Health Organization, Geneva, Switzerland. Requests for reprints should be addressed to Microbiology and Immunology Support Services, World Health Organization, 1211 Geneva 27, Switzerland. A French translation of this article will appear in a later issue of the Bulletin. Reprint No. 5243 collection and analysis of influenza morbidity and mortality data; - to provide influenza virus isolates from out- breaks and sporadic cases for analysis of their antigenic characteristics which are required for a decision on the variants to be included in influenza vaccines; - to detect population groups at increased risk of complications from influenza, which are not cur- rently specified in recommendations for the use of influenza vaccine. Past experience shows that it is difficult to esti- mate the true morbidity and mortality from influenza in most countries on the basis of reports of influenza- like illnesses or acute respiratory diseases. Without laboratory confirmation, such illnesses may be erro- neously attributed to influenza viruses. At the same time, diseases and deaths induced by influenza viruses, where a physician was not contacted, or occurring after the acute symptoms have subsided will be underreported. Analysis and comparisons of the impact of influenza and the benefit of influenza vaccination in different countries are also difficult because surveillance systems and methods vary from one country to another. The laboratory element of influenza surveillance appears to be globally more uniform and available than the statistical observations. However, both labo- ratory services and capabilities to isolate and charac- terize respiratory viruses other than influenza virus need to be expanded in many parts of the world. In an effort to address these issues and to identi- fy generally applicable systems that might improve influenza surveillance information, representatives from six countries discussed current systems used at the national, regional or community level. Their findings and general recommendations are described below. Bulletin of the World Health Organization, 70 (1): 23-25 (1992) . World Health Organization 1992 23 Memorandum Influenza surveillance systems Groups of physicians (sentinel physicians), volun- teering to report weekly on the extent of influenza- like morbidity among their patients, are an important source of information in several countries. The num- ber of physicians participating in such surveillance systems ranges from about 40 to several hundred. Their interest and experience accumulated over sev- eral years are providing results with a high degree of year-to-year consistency despite the range of symptoms presented. Using patient contacts as the denominator and acute respiratory illness/influenza-like illness as the numerator, a statistical model can be developed of morbidity baselines which permit the clear demons- tration of epidemics that exceed the expected base- line. The approach for statistical modelling is similar to that used over many years for mortality in the USA (1) and permits the definition of an epidemic threshold. Laboratory investigation of a subset of ill patients can be organized by providing the physi- cians with portable kits for throat or nasal swab spe- cimen collection which are mailed to a central labo- ratory. A combination of direct antigen detection and rapid culture methods permits quick and efficient re- sponses to the physicians. About 30-50% of speci- mens collected by trained surveillance physicians can be positive for influenza and other respiratory viruses including respiratory syncytial virus (RSV). It is desirable that the surveillance system pro- vides a measure of the severity of the influenza. A simple measure can be obtained through analysis of weekly mortality reports for an entire region or for major cities. The reports can be either influenza and pneumonia deaths (selected by a trained worker in the office that receives death certificates), or, since influenza epidemics are an almost unique event lead- ing to excess mortality, the total number of deaths. The total number of deaths is often available sooner and more readily. After about five years of data col- lection, statistical baselines can be established and major epidemics of mortality readily detected. Another system which might be introduced generally in the future is collection of reports of hos- pital admissions for respiratory illnesses. As more hospitals install effective computer systems to moni- tor patient admissions and discharges such data may be quickly made available without jeopardizing the patients' privacy. Other approaches to surveillance used in certain countries include recording the number of emergency hospital beds and home visits by emergency physi- cians, statistics on sales of pharmaceuticals, and data on social security sickness claims, or absenteeism from school or work. Although such measures may be effective in an individual country sociocultural differences preclude their general implementation. Recommendations All countries (at least in the European region) should explore the possibility of implementing (if not al- ready in place) a minimum of two illness surveil- lance systems, such as: - weekly reporting of acute respiratory diseases by a group of physicians who have received special instructions and training, and for whom virology laboratory support is provided; - weekly reporting of morbidity with total mortali- ty and, where possible, also pneumonia and influenza mortality. A proposed set of data to collect is presented in the Annex. At present the case definition of influenza for sentinel physicians is based on clinical diagnosis only: an acute febrile (e.g., temperature 238°C) res- piratory illness, usually accompanied by prostration and cough or myalgia or sore throat. During out- breaks a sufficient number of clinical cases should be confirmed by laboratory diagnosis to identify the primary etiologic agent(s). At least one laboratory in each country should analyse the samples collected by the sentinel physi- cians for influenza. Representative influenza virus isolates from each country should be sent to the WHO Collaboratoring Centre for further antigenic analysis. Other acute respiratory viruses (RSV, para- myxoviruses, adenoviruses), should also preferably be identified to discriminate outbreaks of influenza from other respiratory viral illness. The above systems should be operated in a consistent fashion for several years, so that the data can be accumulated and used to establish baseline values by simple modelling techniques like those developed by Serfling (2). To enhance its value all information from the unifled influenza surveillance systems should be made available to all interested parties. The World Health Organization should therefore consider estab- lishing an electronic database which can receive the above information by direct data transfer when pos- sible (or transfer from computer disk in other cases), and which can be accessed by all participating coun- tries. References 1. Lul, K.-L. & Kendal, A.P. Impact of influenza epi- demics on mortality in the United States from 24 WHO Bulletin OMS. Vol 70 1992 Standardization and improvement of influenza surveillance October 1972 to May 1985. American journal of public health, 77: 712-716 (1987). 2. Serfling, R.E. Methods for current statistical analy- sis of excess pneumonia-influenza deaths. Public health reports, 78: 494-506 (1963). Annex Data elements for influenza surveillance A. Data which should be gathered from sentinel physicians: (1) Number of cases of influenza-like and other acute respiratory illness/week, by age group: infants (<lyear), preschool children (1-4 years), children (5-19 years), adults (20-59 years), and elderly (.60 years). (2) Total number of encounters by physician with patients. (3) Total number of weekly deaths from all causes by age. (4) Number (or %) of positive isolations of influenza virus by type and subtype. (5) Number of deaths per 106 persons (from any cause). (6) Number of deaths by cause: influenza and pneumonia, heart failure, cerebrovascular diseases, and poorly defined. B. Data which are less recommended: (1) School and industrial absenteeism. (2) Sale of medicines (e.g., cough suppressants, etc.) for illnesses related to influenza. (3) Admissions to emergency wards. (4) Admissions to hospital for pneumonia. WHO Bulletin OMS. Vol 70 1992 25