Inadequacies of Death Certification in Beirut: Who Is Responsible?
Sibai, Abla M., Nuwayhid, Iman, Beydoun, May, Chaaya, Monique, Bulletin of the World Health Organization
Death certification is a public health surveillance tool and a valuable source of information at the national and local levels. Among activities that benefit from the availability of cause-of-death statistics obtained from death certificates are the monitoring of the health of populations, the setting of priorities and the targeting of interventions (1, 2). Such statistics are also the keystone of much epidemiological study, directing the focus of research and complementing mortality data in follow-up studies (3, 4).
In order to facilitate cause-of-death documentation and to standardize reporting and coding practices among various countries, the United Nations and the International Institute for Vital Registration and Statistics periodically develop protocols and guidelines for the management, operation and maintenance of civil registration (5). Moreover, WHO and other organizations produce rules and guidelines for mortality and morbidity coding (6, 7). Nevertheless, death registration is fragmented and largely inadequate in most developing countries (8). In the Eastern Mediterranean Region, for example, only rive countries reported mortality data to WHO between 1985 and 1990. This amounted to a population coverage of only 16%, whereas in Europe and the Americas the corresponding values were 94% and 80%, respectively (9). Late registration, incomplete information and inaccurate data may also be factors in inadequate death registration. While several studies, largely conducted in the West, have focused on the validity of information (10-14) and coding practices (15-18) related to specific causes of death, only a few have examined omissions in data-reporting on death certificates (19, 20) and no information is available on the timing of registration in relation to the occurrence of death. The two latter problems remain issues of concern mainly in developing countries. Little, if anything, is known of the magnitude of these deficiencies or the reasons for them.
The present study was undertaken to evaluate death certification in Lebanon, with reference to its process, mortality documentation and registration as a first step towards identifying limitations and reducing deficiencies. The vital system in Lebanon has never been examined and its utility as a public health tool for planning and evaluation remains questionable. Lebanon is a middle-income country with a population of about 3.5 million. Data at the national level are scarce, the only census ever conducted having been in 1932. Until very recently there were no government statistical surveys or health statistics. The conflicts that ravaged the country between 1975 and 1991 exacerbated the chaotic nature of official documentation. There is virtually no information on causes of death. Even today there are no coding procedures or computerization for the information on death certificates.
The specific aims of the study were to assess the completeness of data-reporting on death certificates and to search for trends over the past 25 years, a period of transition from prewar to postwar conditions. We also investigated whether omissions in data varied with the demographic characteristics of deceased persons and with the circumstances surrounding death events.
Data were obtained from a systematic random sample of 20% of death certificates registered in selected years between 1974 and 1998 in Beirut. The certificates were retrieved, stratified by year of registration, from the files of the Statistics Department of the Ministry of the Interior. Initially, the years 1974, 1984 and 1994 were chosen to represent deaths registered in the prewar period, the war period and the postwar period respectively. Subsequently, the years 1997 and 1998 were added. The same sampling methodology, abstract form, coding manual and data entry file were used for all registration years. The final samples of approximately 500 for each year were considered adequate (maximum error 5%) for estimating proportions of missing data. …