ASSESSING THE COMPARABILITY OF MORTALITY STATISTICS

International Variations in Dementia and Alzheimer Disease Diagnosis and Certification Habits and Their Associations With Dementia and Alzheimer Disease Mortality: A Cross-Sectional Study of 38 Countries

OBJECTIVE

To examine international variations in national diagnosis and certification habits prefer recording dementia (D) versus Alzhiemer disease (AD) as the underlying cause of death (UCOD) and their associations with mortality rates of dementia and AD.

METHODS

We calculated proportions of D/D+AD and AD/D+AD deaths as proxies of national diagnosis and certification habits. Pearson correlation coefficients (r) were estimated to assess the associations of proportions with the mortality rates of dementia or AD among adults aged 75 to 84 years across 38 countries.

RESULTS

The countries with a high preference for recording dementia as the UCOD were Taiwan and Latvia with proportion of D/D+AD deaths of 92% and 88%, respectively, and those with a high preference for recording AD as the UCOD were Slovenia, Turkey, and Poland with proportion of AD/D+AD deaths of 100%, 99%, and 89%, respectively. The r values for the proportions and mortality rate for dementia and AD were 0.67 (95% CI: 0.44-0.81) and 0.46 (95% CI: 0.16-0.68), respectively.

CONCLUSION

We identified a small number of countries with obvious natonal diagnosis and certification habits preferring dementia or AD and had moderate effects on international variations in the mortality rates of dementia and AD.

Seizure-related deaths in children: The expanding spectrum

Although seizures are common in children, they are often overlooked as a potential cause of death. Febrile and nonfebrile seizures can be fatal in children with or without an epilepsy diagnosis and may go unrecognized by parents or physicians. Sudden unexpected infant deaths, sudden unexplained death in childhood, and sudden unexpected death in epilepsy share clinical, neuropathological, and genetic features, including male predominance, unwitnessed deaths, death during sleep, discovery in the prone position, hippocampal abnormalities, and variants in genes regulating cardiac and neuronal excitability. Additionally, epidemiological studies reveal that miscarriages are more common among individuals with a personal or family history of epilepsy, suggesting that some fetal losses may result from epileptic factors. The spectrum of seizure-related deaths in pediatrics is wide and underappreciated; accurately estimating this mortality and understanding its mechanism in children is critical to developing effective education and interventions to prevent these tragedies.

Evaluation of Concordance Between Original Death Certifications and an Expert Panel Process in the Determination of Sudden Unexplained Death in Childhood

IMPORTANCE

The true incidence of sudden unexplained death in childhood (SUDC), already the fifth leading category of death among toddlers by current US Centers for Disease Control and Prevention estimates, is potentially veiled by the varied certification processes by medicolegal investigative offices across the United States.

OBJECTIVE

To evaluate the frequency of SUDC incidence, understand its epidemiology, and assess the consistency of death certification among medical examiner and coroner offices in the US death investigation system.

DESIGN, SETTING, AND PARTICIPANTS

In this case series, 2 of 13 forensic pathologists (FPs) conducted masked reviews of 100 cases enrolled in the SUDC Registry and Research Collaborative (SUDCRRC). Children who died aged 11 months to 18 years from 36 US states, Canada, and the United Kingdom had been posthumously enrolled in the SUDCRRC by family members from 2014 to 2017. Comprehensive data from medicolegal investigative offices, clinical offices, and family members were reviewed. Data analysis was conducted from December 2014 to June 2020.

MAIN OUTCOMES AND MEASURES

Certified cause of death (COD) characterized as explained (accidental or natural) or unexplained, as determined by SUDCRRC masked review process.

RESULTS

In this study of 100 cases of SUDC (mean [SD] age, 32.1 [31.8] months; 58 [58.0%] boys; 82 [82.0%] White children; 92 [92.0%] from the United States), the original pathologist certified 43 cases (43.0%) as explained COD and 57 (57.0%) as unexplained COD. The SUDCRRC review process led to the following certifications: 16 (16.0%) were explained, 7 (7.0%) were undetermined because of insufficient data, and 77 (77.0%) were unexplained. Experts disagreed with the original COD in 40 cases (40.0%). These data suggest that SUDC incidence is higher than the current Centers for Disease Control and Prevention estimate (ie, 392 deaths in 2018).

CONCLUSIONS AND RELEVANCE

To our knowledge, this is the first comprehensive masked forensic pathology review process of sudden unexpected pediatric deaths, and it suggests that SUDC may often go unrecognized in US death investigations. Some unexpected pediatric deaths may be erroneously attributed to a natural or accidental COD, negatively affecting surveillance, research, public health funding, and medical care of surviving family members. To further address the challenges of accurate and consistent death certification in SUDC, future studies are warranted.

Assessment of Relative Utility of Underlying vs Contributory Causes of Death

Importance

In etiological research, investigators using death certificate data have traditionally extracted underlying cause of mortality alone. With multimorbidity being increasingly common, more than one condition is often compatible with the manner of death. Using contributory cause plus underlying cause would also have some analytical advantages, but their combined utility is largely untested.

Objective

To compare the relative utility of cause of death data extracted from the underlying cause field vs any location on the death certificate (underlying and contributing combined).

Design, Setting, and Participants

This study compares the association of 3 known risk factors (cigarette smoking, low educational attainment, and hypertension) with health outcomes based on where cause of death data appears on the death certificate in 2 prospective cohort study collaborations (UK Biobank [N = 502 655] and the Health Survey for England [15 studies] and the Scottish Health Surveys [3 studies] [HSE-SHS; N = 193 873]). Data were collected in UK Biobank from March 2006 to October 2010 and in HSE-SHS from January 1994 to December 2008. Data analysis began in June 2018 and concluded in June 2019.

Main Outcomes and Measures

Death from cardiovascular disease, cancer, dementia, and injury. For each risk factor-mortality end point combination, a ratio of hazard ratios (RHR) was computed by dividing the effect estimate for the underlying cause by the effect estimate for any mention.

Results

In UK Biobank, there were 14 421 deaths (2.9%) during a mean (SD) of 6.99 (1.03) years of follow up; in HSE-SHS, there were 21 314 deaths (11.0%) during a mean (SD) of 9.61 (4.44) years of mortality surveillance. Established associations of risk factors with death outcomes were essentially the same irrespective of placement of cause on the death certificate. Results from each study were mutually supportive. For having ever smoked cigarettes (vs never having smoked) in the UK Biobank, the RHR for cardiovascular disease was 0.98 (95% CI, 0.87-1.10; P value for difference = .69); for cancer, the RHR was 0.99 (95% CI, 0.93-1.05; P value for difference = .69). In the HSE-SHS, the RHR for cardiovascular disease was 0.94 (95% CI, 0.87-1.01; P value for difference = .09); for cancer, it was 1.01 (95% CI, 0.94-1.10; P value for difference = .75).

Conclusions and Relevance

Risk factor-end point associations were not sensitive to the placement of data on the death certificate. This has implications for the examination of the association of risk factors with causes of death where there may be too few events to compute reliable effect estimates based on the underlying field alone.

The incomparability of cause of death statistics under "one country, two systems": Shanghai versus Hong Kong

BACKGROUND

Valid and comparable cause of death (COD) statistics are crucial for health policy analyses. Variations in COD assignment across geographical areas are well-documented while socio-institutional factors may affect the process of COD and underlying cause of death (UCD) determination. This study examines the comparability of UCD statistics in Hong Kong and Shanghai, having two political systems within one country, and assesses how socio-institutional factors influence UCD comparability.

METHODS

A mixed method was used. Quantitative analyses involved anonymized official mortality records. Mortality rates were analyzed by location of death. To analyze the odds ratio of being assigned to a particular UCD, logistic regressions were performed. Qualitative analyses involved literature reviews and semi-structural interviews with key stakeholders in death registration practices. Thematic analysis was used.

RESULTS

Age-standardized death rates from certain immediate conditions (e.g., septicemia, pneumonia, and renal failure) were higher in Hong Kong. Variations in UCD determination may be attributed to preference of location of death, procedures of registering deaths outside hospital, perceptions on the causal chain of COD, implications of the selected UCD for doctors' professional performance, and governance and processes of data quality review.

CONCLUSIONS

Variations in socio-institutional factors were related to the process of certifying and registering COD in Hong Kong and Shanghai. To improve regional data comparability, health authorities should develop standard procedures for registering deaths outside hospital, provide guidelines and regular training for doctors, develop a unified automated coding system, consolidate a standard procedure for data review and validity checks, and disseminate information concerning both UCD and multiple causes of death.

Using multiple-cause-of-death data as a complement of underlying-cause-of-death data in examining mortality differences in psychiatric disorders between countries

BACKGROUND

Little is known about the mortality differences in psychiatric disorders (PD) between countries according to multiple-cause-of-death (MCOD) data.

OBJECTIVE

To compare mortality differences in PD between Taiwan and the US according to MCOD and underlying-cause-of-death (UCOD) data and factors associated with the reporting of PD and assigning PD as the UCOD.

METHOD

MCOD data of Taiwan and the US for years 2003 through 2005 were used for analysis. Deaths per 100,000 population for various PD by age and sex were calculated for each country. Mortality rate ratios between Taiwan and the US were computed to examine the extent of mortality differences between the countries. Odds ratios in reporting PD and assigning PD as the UCOD by age and sex for each country were estimated according logistic regression model.

RESULTS

According to UCOD data, the PD mortality was 3.6 per 100,000 population in Taiwan and 21.9 per 100,000 population in the US, a sixfold difference. The mortality differences increased according to MCOD, which was 10.3 per 100,000 population in Taiwan and 115.4 per 100,000 population in the US, an 11-fold difference. Exception dementia/Alzheimer's disease, the mortality differences between the countries increased in schizophrenia, mood disorder/depression, use of alcohol and use of drug according to MCOD data compared with those according to UCOD data. The percentage in reporting PD among all deaths in the US (13.9%) was higher than those in Taiwan (1.4%); however, the percentage in assigning PD as the UCOD in Taiwan (35%) was higher than those in the US (19%).

CONCLUSION

MCOD data could be used as a complement to UCOD data to provide more information (such as percentage of reporting PD and assigning PD as the UCOD) in interpreting mortality differences in PD between the countries.

An international comparison study indicated physicians' habits in reporting diabetes in part I of death certificate affected reported national diabetes mortality

BACKGROUND AND OBJECTIVE

Physicians may find it confusing to decide whether to report diagnoses in part I or part II of the death certificate. The aim of this study was to contrast differences in diabetes mortality through a comparison of physicians' habits in reporting diabetes in part I of death certification among Taiwan, Australia, and Sweden.

METHODS

A cross-sectional, intercountry comparison study. We calculated the proportion of deaths with mention of diabetes in which diabetes was reported in part I of the death certificate and the proportion of deaths with mention of diabetes in which diabetes was selected as underlying cause of death.

RESULTS

We found that half of the differences in reported diabetes mortality among Taiwan, Australia, and Sweden were due to differences in reporting deaths with mention of diabetes anywhere on the certificate, and half due to differences in proportion of deaths with mention of diabetes in which diabetes was reported in part I of the death certificate.

CONCLUSION

Differences in the reporting of diabetes in part I of the death certificate among physicians in Taiwan, Australia, and Sweden was one of the factors that affected differing reported diabetes mortality in Taiwan, Australia, and Sweden.

Use of multiple causes of death in the analysis of occupational cohorts--an example from the oil industry

OBJECTIVES

To examine the efficacy of routine examination of multiple causes of death occurring on death certificates in cohort studies, with an example from the oil industry.

METHODS

The underlying and multiple causes were coded for all notified deaths from a cohort of 35,000 men employed at eight oil refineries in the United Kingdom. Matrices of the frequencies of underlying causes by contributory causes were analysed for the total population and by subgroups defined by refinery, occupation, age, and calendar period of death.

RESULTS

Over 75% of the 10,128 certificates had two or more causes but this varied by disease. Many ratios of mentions of total to underlying causes were similar to those of England and Wales. Ratios for cancer of the larynx and pneumonia were lower, indicating possible over-reporting of these diseases as the underlying cause. Investigation of an excess of pneumonia deaths at one refinery indicated possible miscoding of the underlying cause or the wrong position of pneumonia on some certificates, particularly in combination with malignancy and stroke.

CONCLUSIONS

Routine analysis of multiple causes of death can provide useful additional information in cohort studies.

Review of the validity of national coronary heart disease mortality rates

Epidemiologic studies of coronary heart disease are heavily dependent on national mortality rates. The diagnostic error for the coronary heart disease is substantial but unquantifiable and is conservatively at least +/- 30%. When this error is superimposed on innumerable errors and omissions in the compilation of monocausal mortality rates, the reliability of such vital statistics currently precludes their use for scientific purposes.

A possible artefactual component in specific cause mortality gradients. Social class variations in the clinical accuracy of death certificates

This paper investigates one possible avenue of artefactual influence on the production and/or concealment of social class gradients in specific cause mortality rates, namely, the possibility of social class biases in the accuracy of diagnosis of cause of death and the systematic misallocation of certain social groups to particular diagnoses. Information on this topic was obtained by matching occupational data gathered at death registration with data on the accuracy of diagnosis of cause of death (measured by diagnostic agreement between clinician and pathologist) collected in a prospective study of 1152 hospital necropsies. Extrapolation from these data to national mortality rates should be cautious, but it appears that in the majority of the most common causes of death grouped by ICD chapter (neoplasms, cerebrovascular and digestive) social class gradients would be steeper if mortality data were based on pathologists' rather than clinicians' diagnoses. Only in the respiratory chapter would the gradient be reduced, with the gradient in cardiovascular deaths unaffected.

Certification and coding of two underlying causes of death in The Netherlands and other countries of the European Community

Differences in certification and coding of causes of death between countries of the European Community were studied by sending sets of case histories to samples of certifying physicians. Completed certificates were coded by national coding offices and by by a WHO reference centre. Detection fractions ranged from 60% to 92% in a first study (concerning cases of chronic obstructive pulmonary disease) and from 80% to 94% in a second study (concerning cases of cancer). A detailed analysis of the findings for the Netherlands, which performed very well in both studies, reveals a substantial frequency of errors in certification (as opposed to errors in diagnosis). Comparison of national and reference centre coding suggests that the Dutch coding process is to a certain extent adapted to the frequency of these certification errors, leading to deviations from WHO coding rules. It is concluded that certification and coding practices should be studied together and that further international standardisation of coding practices will not necessarily improve the validity of national cause of death statistics.

An appraisal of the epidemic rise of coronary heart disease and its decline

Epidemiologists have used mortality statistics to demonstrate a sharp rise in the incidence of coronary heart disease in several countries since the turn of the century and a decline in some countries since the late 1960s. However, increased longevity, changes in coding and diagnostic practices, and familiarity with the clinical and pathological features of the disease make the increase largely spurious. Diagnostic errors in certified causes of death in general, and coronary heart disease in particular, indicate that vital statistics are too unreliable for determining whether there has been an increase and a subsequent decline in the incidence of coronary heart disease.

The patterning of health by social position in contemporary Britain: directions for sociological research

Associations are described between health (variously measured) and occupational class, gender, marital status, age, ethnicity and area of residence, using British data. It is argued that when exploring the social patterning of health, illness and death, it would be profitable for sociologists to consider several or all of these social positions, and to develop models of general vulnerability to ill-health rather than of specific etiology. Three main types of explanation, artefact, health selection and social causation, are reviewed, and six general points are made about how sociological research on social patterning in health should proceed.

Death certification by British doctors: a demographic analysis

Regional differences in mortality from cardiovascular diseases, respiratory diseases, strokes, and other causes have been observed in England and Wales. To determine to what extent the death certification practices of doctors influence these variations, we surveyed 123 British general practitioners and housemen. Each doctor completed death certificates based on fictitious case histories. We found that diagnostic groupings did not vary significantly according to the doctors' region, type of practice, place of medical training, sex, or year of qualification. Doctors qualifying before 1955, however, appeared more inclined than their younger colleagues to list stomach cancer on the death certificate.

Comparison of cause of death coding on death certificates with coding in the Royal College of General Practitioners Oral Contraception Study

A comparison has been made between the coding of the cause of death by (a) the Royal College of General Practitioners (RCGP) during the Oral Contraception Study and (b) the Office of Population Censuses and Surveys (OPCS) or the General Register Office for Scotland (GRO) on death certificates for the same subjects. Broad grouping of the International Classification of Diseases (ICD) showed close agreement between RCGP and OPCS or GRO coding for all deaths which occurred from the start of the Oral Contraception Study in 1968 up to December 1978. Moreover, where discrepancies occurred there were no systematic differences between ever-users of oral contraceptive and non-users. Detailed examinations of discrepancies in the coding of the causes of those deaths included in the RCGP publication of October 1977 shows that our previous estimate of mortality risk associated with oral contraceptives would not be materially altered by the use of death certificate information.

Relation between clinical and autopsy diagnoses, especially as regards cancer

The clinical diagnoses of 377 patients were compared with those derived from autopsy; the incidence, confirmation rate and detection rate were correlated to age, sex and duration of stay in the hospital. The diagnoses were in complete agreement in 75% of the cases; as regards location they differed in 20% and malignancy being unsuspected in 5% of the cases, where vascular diseases were the most frequent clinical findings. Lymphomas and leukemias were overrepresented in the study and had a high confirmation and detection rate, 97% resp. 94%. Tumours of the liver, gallbladder and pancreas were underrepresented and had a low detection rate, 37%, and confirmation rate, 59%. For other tumours the confirmation rate was higher, 84-97%, and detection rate lower, 72-80%. Similar antemortem and postmortem diagnoses were seen most commonly in young patients, regardless of sex and who stayed in the hospital for an extended period of time.

Terminology in chronic obstructive lung diseases

Until the 1960's there was great confusion, both within and between countries, on the meaning of diagnostic terms such as emphysema, asthma, and chronic brochitis. Proposals made by a group of British doctors in 1959 gradually received widespread acceptance but in recent years some new problems have developed. These include difficulties in the definition of airflow obstruction, recognition that what used to be regarded as a single disease, chronic bronchitis, comprises at least two distinct pathological processes, and uncertainty about the degree of variability which distinguishes asthmatic from more persistent forms of airflow obstruction. These are all problems which could be solved by continuance of appropriate research and of riqorous attention to the principles which determine accurate and acceptable definitions of disease.

Regional variations in mortality from ischaemic heart and cerebrovascular disease in Britain

In middle-aged men and women, mortality from ischaemic heart disease and cerebrovascular disease is highest in the north and west of Britain. The worst region is West Central Scotland. Statistical analysis using a linear logistic model shows that the differences between the regions are significant and the yearly fluctuation in numbers of deaths contributes little to the overall variation.

Natural history of stroke in Rochester, Minnesota, 1945 through 1954

The records of the Mayo Clinic have been used as the primary basis for the study of stroke in the population of Rochester, Minnesota, from 1945 through 1954. The incidence rate for first stroke of all types was 1947/100,000/year. The rates increased with age, and at age 65 to 74 years, about 1% of the population was affected annually. No significant differences were noted between men and women. The rate for cerebral thrombosis was 146/l00,000/year for all ages. Cerebral hemorrhage represented less than 10% of all strokes and occurred in less than 15% of those who died; this is contrary to what is published in the U.S. Mortality Statistics . The prevalence rate was 547/100,000 on January 1, 1955. Twenty-one percent of these persons who had previously had a stroke were functioning with no incapacity on the date of prevalence determination and only 3% were bedridden. Death certificates were reviewed for all those who died after a stroke. Among those who died within a month after a stroke, some type of stroke was noted as the underlying or an associated cause of death in 91%. However, among those who died more than a month after a stroke, only 54% had such a diagnosis recorded. Probability of surviving after each type of stroke was determined and compared with survival in a normal population. The curve showing probability of surviving from cerebral thrombosis diverges throughout its course from that of the expected survival but diverges more sharply in the first several months. Among those patients with stroke who died, 41% died of causes related to the acute stroke and 10% died of a subsequent stroke. Heart disease was the cause of nearly twice as many deaths as was a subsequent stroke among those who survived the initial stroke. Autopsies were performed on approximately 50% of those persons who died following a stroke, and information from these studies confirmed the high degree of accuracy of the clinical appraisals.