The quality and utility of death certificate data.

Abstract

Despite the impressive strides in health information, mortality statistics are the only source of data at the national, state, and local level that is consistently available and that satisfies the basic criteria of coverage and continuity. The Neither the accuracy of the reporting as defined by the study methods nor the interpretation of the results appears to be very heartening. The study determined that only 65 per cent of the observed underlying causes named on the death certificates fell into the groups that were defined as indicating good agreement. Using site-specific cancer as categorized by the three-digit ICDA codes, the study found substantial variability in accuracy of underlying cause among the separate site-specific cancer categories. In discussing the sources of disagreement, the study identified the physician or coroner as the most common source for discrepancy, and concluded there is a need to raise the level of awareness of physicians and coroners to pay greater attention to entries of cause of death.While this detailed cancer specific study is largely directed toward epidemiological applications, it raises a number of broader issues about the quality, use, and evaluation of current cause of death information.Any measure of "accuracy" used to evaluate death certificate information is wholly dependent on the performance requirements determined by a specific user. An epidemiologist following a relatively rare disease will have an exacting requirement for accuracy, whereas the community health planner may well be satisfied with the relative rank or magnitude of a disease specific entity. The epidemiologist will use the death certificate as a point of departure; and the planner will use it as an appropriate end point.Upgrading the system and its accuracy is a variable thing, since the death certificates serve such diverse user audiences. On the one hand, access to the certificate becomes a major obstacle to the epidemiologic clinician, or disease specific investigator who would not rely on the certificate except as a starting point for a detailed study. This situation is illustrated in the cancer mortality findings in that site-specific cancers appearing in the poor accuracy groups (Groups 2-4) are relatively rare. Any rigorous study of rare phenomena would invariably turn toward a comprehensive case finding approach without sole reliance on either hospital or death records.On the other hand, to the community health planner, the intended use of mortality data may be to characterize time trends in broad disease groupings and demographic shifts. Using larger aggregations of specific 3-digit ICDA groupings will tend to lessen errors in accuracy. The proportional mortality (Table 8) of Percy, Stanek, and Gloeckler give some assurance that the hospital and death certificate results are

Road Traffic Deaths in Brazil: Rising Trends in Pedestrian and Motorcycle Occupant Deaths

“…This analysis is limited by the quality of data collection and reporting to the SIM national database. The accuracy of death certificate and hospital-coded death data has long been recognized as limited (Glasser 1981;Lu et al 2001), in particular with road traffic deaths (Harris 1990;Hijar et al 2012;Lapidus et al 1994). Previous studies from Brazil have shown that cause-ofdeath coding in the SIM database can be limited; when reviewing coded deaths from the southern city of Porto Alegre, Fajardo et al (2009) found that nearly 13 percent needed to be recoded by researchers.…”

Stroke Incidence and Case Fatality in Southern Greece

“…More importantly, it is well known that the quality of routine mortality statistics for stroke shows considerable differences across countries. 13,14 Approximately 45% of cases with stroke as the underlying cause of death did not meet the WHO criteria, according to a study from Auckland. 15 Similar to our substudy, only 52% of stroke deaths in the Tripoli area could be confirmed as first-ever stroke.…”

Stroke Registry of Tartu, Estonia, from 1991 through 1993

“…However, although we paid particular atten- tion to find the cases occurring outwith hospitals, difficul ties may always [ 17] arise in identifying cases amongst the older population more often treated at home, as well as for those who die immediately after the ictus and are not seen by a physician. Many authors [6,22] have suggested the possibility of an overestimation o f the stroke mortality rate when based only on official routine statistics. A recent study from Auckland established that about 45% of stroke cases which stated stroke as the underlying cause of death failed to meet the W HO criteria [23], Presumably, we were able to overcome this point by including only the patients with a certain clinical history of stroke noted in medical documents.…”