Zur Validität der Herz-Kreislauf-Mortalitäts-Statistik der ehemaligen DDR

It has been known already for a long time that in the GDR the mortality rates for coronary heart disease (CHD) and cerebrovascular accidents (CVA) had been considerably underestimated. Instead of these diagnoses very often such general conditions like atherosclerosis and hypertension have been coded as underlying cause of death. We carried out, therefore, two validation studies in order to check whether and to what extent violations of the WHO coding rules were responsible for that. In the first study all hospital deaths which occurred in the GDR between 1985 and 1989 have been compared with the corresponding data of the official mortality statistics (record-linkage-database). In the second study 4.154 death certificates have been manually checked and recoded. Among the hospitalized patients who died from an acute myocardial infraction (AMI) the AMI was coded as underlying cause of death at the death certificate only in 57 % (men) and 54 % (women), respectively. Among cases of CHD these proportions were 66 % and 62 %, respectively, and among cases of CVA 46 % and 44 %, respectively. In the second study among those deaths with AMI as one of the three possible diagnoses at the death certificate AMI was coded as underlying cause of death in men in 46 % and in women in only 30 %. For CHD these proportions were 71 % and 59 %, respectively, and for CVA 44 % and 46 %, respectively. Both studies confirm that in the GDR the selection rules recommended by WHO have often been ignored when coding the death certificates of death cases from AMI, CHD and CVA. Based on the results of the two studies the following correction factors for the official mortality rates are proposed for men and women, respectively: AMI 1.8/2.3; CHD 1.5/1.6; CVA 2.2/2.3.     

Multiple-cause coding of death from myocardial infarction: population-based study of trends in death certificate data

BACKGROUND: Data on long-term trends in mortality are generally unavailable for multiple-cause coding of deaths. We wanted to know whether multiple-cause coding of deaths for myocardial infarction contributes much to the interpretation of death certificate data on mortality rates for this condition. METHODS: We analysed all causes of death on death certificates in the former Oxford health service region from 1979 to 1998. RESULTS: Of 69,333 death certificates that included myocardial infarction as a cause of death, it was the underlying cause of death in 93.6 per cent. The ratio of 'mentions' to 'underlying cause' was broadly similar over the study period, during which time there were substantial falls in mortality rates. There were significant changes to the ratios, associated with timing of changes to coding rules; but their effects were small. The ratio of mentions to underlying cause was similar in men and women and in different age groups. CONCLUSION: The underlying cause of death was a robust and almost complete measure of certified deaths for myocardial infarction.     

Geographic distribution of deaths due to sentinel health event (occupational) causes

This report describes the Compressed Mortality File available from the National Center for Health Statistics that can be used to easily and efficiently generate annual mortality rates for geographic areas as small as counties for any period from 1968 to 1985. Several ways of presenting geographic variation in mortality rates due to potentially work-related deaths and changes in these rates over time are discussed for the 15-year period from 1969 through 1983. Causes of death that are potentially work-related were identified using the sentinel health events (occupational) [SHE(O)] concept. Data are given for nine diagnostic groups of occupationally related disorders, and maps are presented for bladder cancer, acute myeloid leukemia, and pneumoconioses. Significant changes in age-adjusted mortality rates were noted for pneumoconioses and acute myeloid leukemia that could not be due to changes in the disease coding of death certificates. Racial differences in mortality rates due to pneumoconioses may be due to differences in employment patterns. The use of SHE(O) codes to search the Compressed Mortality File may be helpful in identifying areas for public health concern, even if only as a monitoring signal for subsequent time periods. This file also provides an easy way to generate reference population mortality rates for epidemiologic studies.     

Coronary heart disease death and sudden cardiac death: a 20-year population-based study

Trends in out-of-hospital coronary heart disease (CHD) death, a surrogate for sudden cardiac death (SCD), are important for understanding the decline in CHD mortality. Little is known about out-of-hospital CHD death without prior CHD diagnosis, the definition of unexpected SCD. The authors analyzed secular trends in CHD death and unexpected SCD over a 20-year period (1979-1998) to examine the association between prior CHD and SCD and to test the hypothesis that in-hospital deaths declined more than SCDs. The yearly decline in CHD mortality rates was 5.3% for in-hospital deaths and 1.8% for out-of-hospital deaths (p = 0.001). Among all SCDs, the proportion of unexpected SCD was 49%. Mortality rates for both unexpected SCD and SCD with prior CHD declined over time, but unexpected SCD declined at a slower rate than SCD with prior CHD (p = 0.001). The relative odds of prior CHD were higher among persons with SCD than among controls, but there was a modest decline in the magnitude of the association. Thus, during the past 20 years, the decline was greater for in-hospital CHD deaths than for SCDs. Since approximately half of the SCDs were unexpected and rates of these deaths declined less over time than rates of SCD with prior CHD, primary prevention is becoming increasingly more important in sustaining the decline in CHD mortality.     

Trends in fatal coronary heart disease among people aged 25-74 years in Oita City,Japan, from 1987-1998

Serum cholesterol has been increasing in recent years in Japan. There is concern that risk of coronary heart disease (CHD) may be increasing too, but there is little information on validated fatal CHD trends in the Japanese population. We identified 1,056 deaths from heart disease and other deaths possibly hiding CHD from death certificates of residents aged 25-74 years in Oita City, Japan in 1987-1988, 1992-1993, and 1997-1998 (mean population, 273,000 in 1997-1998). We validated 994 of them by medical record review and physician interviews, classifying them into definite fatal acute myocardial infarction (AMI) and possible fatal AMI or CHD death based on Monitoring Trends and Determinants in Cardiovascular Disease project's criteria. Sudden death was defined to estimate the number of CHD sudden deaths. In men, age-adjusted mortality rates due to validated fatal CHD remained quite stable over 10 years (25.3 per 100,000 [95% CI, 15.0-35.5] in 1987-1988 to 24.2 per 100,000 [95% CI, 16.1-32.3] in 1997-1998). When 50% or all sudden deaths were included as fatal CHD, the rates for men tended to decline. This was due to decreasing out-of-hospital deaths in connection with a declining CHD death rate among men aged 65-74 years, whereas in-hospital CHD deaths were level. In women, the rate of validated fatal CHD was highest in 1992-1993, but the 1997-1998 rate was similar to the 1987-1988 rate. We did not find that fatal CHD rates increased in Oita men and women from 1987-1998. Rather, out-of-hospital fatal CHD tended to decline in Oita men.     

Could the coronary heart disease mortality rates decline be artefactual?

The recent decline in Coronary Heart Disease (CHD) mortality rates has been attributed to reduction in risk factors and to improved management. In this article, we review whether artefacts of classification could have played a role as well. Knowledge and information on disease, competing causes of death, death certification accuracy and completeness, advancing age of the population with multiple conditions as well as death certificate coding practice could have affected secular trends of CHD mortality rates. However, the lack of noticeable shifts within the cardiovascular category or with another category makes it difficult for the CHD decline to be artefactual.     

Nosological coding of cause of death

Death certificates representing 766 decedents who had participated in the Hypertension Detection and Follow-up Program (1973-1979) at one of 14 US centers were given to three nosologists for purposes of coding underlying cause of death. Analyses examined interobserver variability among the three nosologists as well as intraobserver variability for each of the three nosologists. All three nosologists agreed on a three-digit International Classification of Diseases, Adapted (ICDA) code in 90.2% of the cases and at least two out of three agreed in 99.7% of the death certificates examined. Agreement rates improved when disease codes were collapsed into broader categories utilized in the Hypertension Detection and Follow-up Program. When particular disease classifications (e.g., cerebrovascular, ischemic heart disease, myocardial infarction, and neoplasms) were examined, three out of three agreement rates were highest for neoplasms (97.8%) and lowest for myocardial infarction (86.5%). Similarly, two out of three agreement was highest for neoplasms (98.5%) and lowest for myocardial infarction (88.0%). Intranosologist agreement rates were based on a recoded 20% sample of death certificates. Agreement rates for three-digit ICDA codes ranged from 94.8% to 96.1% for the three nosologists. The agreement rates for the general disease categories ranged from 96.7% to 97.4%.     

The burden of diabetes-related mortality in France in 2002: an analysis using both underlying and multiple causes of death

Aim To describe the burden of diabetes-related mortality in France. Methods Underlying and multiple causes (all causes listed) of death were extracted from the 2002 French national mortality registry. Death rates were standardized on the age structure of the European population. Results Diabetes was reported as the underlying cause of death in 11,177 certificates (2.1%), and as multiple causes in 29,357 certificates (5.3%), giving a ratio (multiple/underlying causes) of 2.6. When diabetes was a multiple cause, the mean age at death was 75 years in men, 81 years in women. The age-standardized mortality rates were 41.0/100,000 in men, 24.6/100,000 in women. The excess mortality observed in men (men/women ratio = 1.7) decreased with age. Geographic differences were observed: higher rates in the North-East, lower rates in the West of the country. In certificates mentioning diabetes, the most frequent cause of death was diseases of the circulatory system (76%). Coronary heart diseases, foot ulcers and renal diseases were more likely to be mentioned in certificates referring to diabetes than in those that did not. Discussion The use of multiple rather than underlying causes of death more than doubled diabetes-related mortality rates. While probably still under-estimated, the burden of diabetes-related mortality corresponds to a high proportion of the total mortality, especially in men. Geographic differences partially reflect disparities in diabetes prevalence. Causes more frequently associated with diabetes include coronary heart disease and complications related to neuropathy and nephropathy.     

Trends in coronary heart disease mortality and morbidity in Auckland, New Zealand, 1974-1986

Routine mortality statistics show that coronary heart disease (CHD) death rates have declined consistently in Auckland men since 1968; in women, death rates declined between 1968 and 1986 but since 1981 there may have been a reduction in the rate of decline. Data from CHD registers conducted in Auckland, New Zealand in 1974, 1981, and since 1983 as part of the WHO MONICA Project, have been used to investigate the validity and reasons for the decline in the age group 35-64 years. In Auckland age-standardized sudden coronary death rates in men declined by approximately 2% per year between 1974 and 1986; there was no apparent decline in women. There was also an indication of a decline in age-standardized definite myocardial infarction rates but again only in men; 28 day case fatality in patients with a definite myocardial infarction has not changed significantly in the period 1981-1986. These results validate the mortality trends based on death certificates and in particular the differing recent trends in men and women. The decline in CHD mortality in men without a concomitant change in case fatality and the lack of recent decline in women, suggest that changes in the natural history of the disease rather than treatment are responsible for the mortality trends. Since disease events are rare in absolute numbers, long-term monitoring of coronary heart disease in large population groups will be necessary to usefully study disease trends, particularly in women.     

Causes of death among persons reported with AIDS.

OBJECTIVES. This study describes causes of death in persons with acquired immunodeficiency syndrome (AIDS) and assesses the completeness of reporting of human immunodeficiency virus (HIV) infection or AIDS on death certificates of persons with AIDS. METHODS. AIDS case reports were linked with death certificates in 11 local/state health departments; underlying and associated causes of death were available for 32,513 persons with AIDS who died. RESULTS. HIV/AIDS was designated as the underlying cause of death for 46% of persons with AIDS who died between 1983 and 1986 and 81% of persons with AIDS who died since 1987 (the year specific coding procedures were implemented for HIV/AIDS). Most other underlying causes of death were conditions within the AIDS case definition (notably Pneumocystis carinii pneumonia), pneumonia, infections outside the AIDS case definition, and drug abuse. Unintentional injuries, suicide, and homicide were less common. HIV/AIDS was listed as underlying or associated on 88% of death certificates from 1987 to 1989; reporting varied primarily by HIV exposure category and time between diagnosis and death. CONCLUSIONS. Physicians and other health care professionals should realize their critical role in accurately documenting HIV-related mortality on death certificates. Such data can ultimately influence the allocation of health care resources for HIV-infected individuals.     

National trends and coding patterns in fall-related mortality among the elderly in the United States

Because good information on deaths caused by a fall would be important for prevention policies, we analyzed the influence of coding differences on variability in state-level fall death rates in the elderly. We examined state differences in the number of cause of death codes on death certificates, death certifiers, completeness of E-coding, and indicators of specificity of coding. We found that state-specific fall mortality rates ranged from 13.9 to 140.4 in people aged 65 years and above. States employing a coroner to investigate injury deaths had 14 per cent fewer recorded fall deaths than those where a medical examiner conducted the investigations. Each unit increase in the median number of cause of death codes was associated with a 10 per cent increase in the number of falls. For each 1 per cent increase in the use of unspecified codes for the underlying cause of death, the number of falls dropped by 2 per cent. Current fall mortality data do not appear to identify all instances of falls. Variability in unintentional fall-related death rates among states may be partly explained by death certification coding practices. Standardization of coding and training for documentation of fall events and death certificate reporting could help uncover the actual fall mortality burden in the elderly.     

Misclassification of coronary heart disease in mortality statistics. Evidence from the WHO-MONICA Ghent-Charleroi Study in Belgium

STUDY OBJECTIVE: To validate the Belgian vital statistics for coronary heart disease (CHD) on the basis of an independent acute myocardial infarction (AMI) register, carried out as part of the WHO-MONICA project. DESIGN: Records of fatal cases of AMI in the WHO-MONICA register were individually linked to the corresponding death certificates. SETTING: Since 1983, the WHO-MONICA Collaborating Centre Ghent/Charleroi registers all fatal and non-fatal AMI in the age group 25-69 years in two geographical areas, Ghent in the northern Dutch speaking part and Charleroi in the southern French speaking part of Belgium. Registration is done according to the MONICA protocol. The official vital statistics in Belgium are published on a yearly basis. They are essentially a reflection of the "underlying" causes of death, coded according to the 9th revision of the International Classification of Diseases (ICD). The study was undertaken in the period 1983-1991. MAIN RESULTS: Out of a total of 741 (Ghent) and 934 (Charleroi) well documented MONICA fatal cases of AMI, 492 (66.4%) and 641 (68.6%), respectively, were officially labelled as CHD (ICD code 410-414); 438 (59.1%) and 385 (41.2%), respectively, were officially labelled as AMI (ICD code 410). A substantial fraction of the MONICA AMI cases--27.1% in Ghent and 38.2% in Charleroi--was coded as "other forms of CHD" (ICD 411-414) or as "other forms of heart disease" (ICD 420-429). The remaining MONICA AMI cases--13.8% in Ghent and 20.6% in Charleroi--were classified in either very aspecific (for example, atherosclerosis, ICD 440) or totally unrelated ICD codes (for example, neoplasm, ICD 140- 239). CONCLUSIONS: It is concluded from the results in this paper that a substantial part of all deaths caused by CHD in Belgium are labelled with incorrect ICD codes and are therefore misclassified in the official mortality statistics for Belgium. This is partly caused by a "drainage" of cases towards less specific CHD related ICD categories. A considerable fraction, however, seems to be absolutely misclassified.       

Mortality from coronary heart disease and acute myocardial infarction--United States, 1998

Despite improved clinical care, heightened public awareness, and widespread use of health innovations, coronary heart disease (CHD) remains the leading cause of death in the United States, and the decline in rates from CHD that began during the 1960s slowed during the 1990s. This report provides national and state-specific death rates for CHD and for acute myocardial infarction (AMI). During 2001, approximately 1.1 million persons are expected to have a CHD event. Prevention remains the key strategy for reducing CHD mortality.     

Causes of death in Huntington disease as reported on death certificates

Huntington disease (HD) is a late onset autosomal dominant neurological disorder. Two hundred fifty-three death certificates of HD-affected individuals from four midwestern states were examined to determine the completeness of reporting HD on the death certificates. Overall, 66% of death certificates indicated HD as a primary or contributory cause of death. There was significantly better reporting on more recent death certificates, but even since 1979, 16% did not report HD. The implications to those researching HD family histories and to the accuracy of mortality rates are discussed.     

Insulin treated diabetes mellitus: causes of death determined from record linkage of population based registers in Leicestershire, UK.

STUDY OBJECTIVE--Analyses of causes of mortality in people with diabetes using data form death certificates mentioning diabetes provide unreliable estimates of mortality. Under-recording of diabetes as a cause on death certificates has been widely reported, ranging from 15-60%. Using a population based register on people with diabetes and linking data from another source is a viable alternative. Data from the Office of Population Censuses and Surveys (OPCS) are the most acceptable mortality data available for such an exercise, as direct comparison with other published mortality rates is then possible. DESIGN--A locally maintained population-based mortality register and all insulin-treated diabetes mellitus cases notified to the Leicestershire diabetes register (n = 4680) were linked using record linkage software developed in-house (Lynx). This software has been extensively used in a maintenance and update cycle designed to maximise accuracy and minimise duplication and false registration on the diabetes register. Deaths identified were initially coded locally to the International Classification of Diseases, 9th revision (ICD9), and later a linkage was performed to use official OPCS coding. Mortality data identified by the linkage was indirectly standardised using population data for Leicestershire for 1991. Standardised mortality ratios (SMR) were estimated, with 95% confidence intervals. Insulin dependent diabetes (IDDM) was defined as diabetes diagnosed before age 30 years with insulin therapy begun within one year of diagnosis. All other types were considered non-insulin dependent diabetes (NIDDM). Analyses were performed for the whole sample and then for the NIDDM subgroup. Results from these analyses were similar and therefore only whole group analyses are presented. MAIN RESULTS--A total of 370 deaths were identified for the period of 1990-92 inclusive - 56% were in men and 44% in women, median age (range) 71 years (12-94). Approximately 90% of deaths were subjects with NIDDM. Diabetes was mentioned on 215 (58%) death certificates. The all causes SMRs were significantly raised for men and women for all ages less than 75 years. Ischaemic heart disease (ICD9) rubrics 410-414) accounted for 146 (40%) deaths - 41% of male and 38% of female deaths. Male and female SMRs were significantly raised for the age groups 45-64, 65-74, and 75-84 years. Cerebrovascular disease (ICD9 rubrics 430-438) accounted for 39 (10%) deaths and the SMR for women the external causes of death (ICD9 rubrics E800-E999) were also significantly raised overall and in age groups 15-44 and 45-64 years. This was not true for men, although numbers of deaths in this category were small for both men (4) and women (9). CONCLUSION--Record linkage has been used successfully to link two local, population based registers. This has enabled an analysis of mortality in people with diabetes to be performed which overcomes the problems associated with using as a sample, death certificates where diabetes is mentioned. The mortality rates and SMRs estimated should more accurately reflect the true rates than would be possible using other methods. The persisting excess mortality identified for people with diabetes is of a similar magnitude and attributable to similar causes as has been reported elsewhere in population based studies.     

The effect of inaccuracies in death certification and coding practices in the European Economic Community (EEC) on international cancer mortality statistics

An investigation into the effect of national death certification and coding practices on published mortality statistics in eight EEC countries is reported. Doctors in each country were asked to complete specimen death certificates for a bank of written case histories. Certificates from each country were coded by their own offices and then by a WHO reference centre. Within and between countries, discrepancies occurred both in the doctors' diagnoses and in the codes assigned to certificates. At an international level these differences had serious implications for the comparability of mortality data for cancers of the cervix and uterus, and for mesotheliomas.     

Birthweight and coronary heart disease in a cohort born 1857-1900 in Melbourne, Australia

BACKGROUND: The widely observed association between birth size and risk of later coronary heart disease (CHD) has not been examined in an impoverished pre-20th century birth cohort. METHODS: Birth weights and maternal characteristics, for births between 1857 and 1900 in a charity hospital, were recorded from preserved ledgers. Names were linked to death certificates to determine age and cause of death. Death with CHD was coded using specific criteria, and survival analysis methods were used to relate risk of CHD to birth weight, allowing for competing causes of death and adjusting for potentially confounding maternal factors. RESULTS: Death certificates were traced for 8,584 (53%) of 16,272 registered live-births. Survival analyses were confined to 2,938 subjects (1,572 male, 1,366 female) who survived beyond age 40, since none of the 486 CHD cases was recorded earlier. CHD risk increased with time, but there was no evidence that it was related to birth weight, in men or women. CONCLUSIONS: We did not replicate findings in more recent cohorts. This may represent a true lack of association in a historical cohort who we believe remained impoverished through their early life. However, we acknowledge the strong possibility of misclassification of cause of death by the person filling in the death certificate and/or our coding criteria, and temporal change in diagnostic criteria for CHD. We cannot exclude the possibility that low birth weight babies 'programmed' in utero for later CHD were more likely to die in infancy, in this cohort with a high infant mortality rate.     

Multiple codification of the causes of death: from dying "of" to dying "from"]

Multiple-cause mortality data is examined in the Valencian Region. In addition to coding the underlying cause of death (UCD), all causes of death which appeared mentioned on death certificates (MCD) were coded according to preliminary rules established by the Mortality Statistics Office. Specific diseases were selected to explore mortality patterns. The average number of conditions coded per death certificate was 2.7. Two or more conditions on the lowest used line appeared in 33.8% of all medical certificates. Septicaemia, high blood pressure and arteriosclerosis stand out among the conditions more often coded as MCD than coded as UCD. Exploring for mortality patterns a statistical association between coronary heart disease and diabetes emerged (p less than 0.0001). Multiple-cause mortality coding allow to discriminate mortality patterns and show a new magnitude to some specific causes of death.     

Diabetes mellitus as cause of death

Certification and coding of diabetes mellitus as a cause of death were investigated by sending a random sample of 300 physicians a set of 6 case histories. Of these, 228 (76%) participated in the study by completing a death certificate for each of these cases. The certificates were subsequently coded by the Central Bureau of Statistics. The main finding was that doctors varied enormously in the way in which diabetes mellitus was mentioned on the death certificate: not at all, as a contributory cause of death, or as an underlying cause of death. Coding removes some of the inconsistencies, but induces additional variation: a higher age of the deceased is associated with a lower probability of having diabetes mellitus coded as the underlying cause of death, and a higher probability of not receiving a code of diabetes mellitus at all. It is concluded that the cause-of-death registration does not provide an accurate picture of the contribution of diabetes mellitus to the cause-of-death pattern of the Netherlands. This is due, amongst other things, to the conceptualization of causes of death on which the registration is based. On the other hand, changes in certification and coding practice within the current system may already lead to some improvement.