Analysing Multiple Causes of Death: Which Methods For Which Data? An Application to the Cancer-Related Mortality in France and Italy

Cardiovascular mortality in Brazil during the COVID-19 pandemic: a comparison between underlying and multiple causes of death

OBJECTIVES

The COVID-19 pandemic has differentially impacted cardiovascular disease (CVD) mortality worldwide. Causes of death misclassification may be one of the reasons. We evaluated the impact of the pandemic on CVD mortality in Brazil, comparing underlying causes (UCs) and multiple causes (MCs) of death.

STUDY DESIGN

Ecological time-series study.

METHODS

An ecological, time-series study was conducted analysing age-standardised death rates for CVD, from epidemiological week (EW) 10/2020 to 39/2021, using data from the Mortality Information System, Brazil. CVD was defined using the International Classification of Diseases (ICD-10) coding, if reported as UC or MC of death. Observed and expected data (mean for the same EW, 2017-2019) were compared. Risk ratios (RiRs) were analysed, and 95% confidence intervals (CIs) were calculated.

RESULTS

Age-standardised mortality rate for CVD as UC of death was 165.8 (95%CI: 165.4-166.3) per 100,000 inhabitants, similar to what was expected (165.6/100,000, 95%CI: 165.2-166.1, RiR = 1.00). There was increased out-of-hospital mortality (RiR = 1.18; 95%CI: 1.17-1.19) and deaths of ill-defined causes (RiR = 1.43; 95%CI: 1.42-1.44). The increase in out-of-hospital deaths was more pronounced in the North (RiR = 1.33; 95%CI 1.30-1.36) region, with a less resilient health system. Conversely, as MCs of death, there was a 10% increase in CVD mortality (observed: 243.2 [95%CI: 242.7-243.7], expected: 221.6 [95%CI: 221.1-222.1] per 100,000). An increase also occurred in the North and Central West regions (RiR = 1.16; 95%CI: 1.15-1.18), among men (RiR = 1.11; 95%CI: 1.11-1.12) and individuals aged ≥60 years (RiR = 1.11; 95%CI: 1.10-1.11).

CONCLUSIONS

During the pandemic, mortality rates for CVD as MCs of death increased in Brazil, whereas as UC mortality rates did not change. Higher out-of-hospital mortality, misclassification, and competing causes of death may explain this pattern.

Analysis of Multiple Causes of Death: A Review of Methods and Practices

BACKGROUND

Research and reporting of mortality indicators typically focus on a single underlying cause of death selected from multiple causes recorded on a death certificate. The need to incorporate the multiple causes in mortality statistics-reflecting increasing multimorbidity and complex causation patterns-is recognized internationally. This review aims to identify and appraise relevant analytical methods and practices related to multiple causes.

METHODS

We searched Medline, PubMed, Scopus, and Web of Science from their incept ion to December 2020 without language restrictions, supplemented by consultation with international experts. Eligible articles analyzed multiple causes of death from death certificates. The process identified 4,080 items of which we reviewed 434 full-text articles.

RESULTS

Most articles we reviewed (76%, n = 332) were published since 2001. The majority of articles examined mortality by "any- mention" of the cause of death (87%, n = 377) and assessed pairwise combinations of causes (57%, n = 245). Since 2001, applications of methods emerged to group deaths based on common cause patterns using, for example, cluster analysis (2%, n = 9), and application of multiple-cause weights to re-evaluate mortality burden (1%, n = 5). We describe multiple-cause methods applied to specific research objectives for approaches emerging recently.

CONCLUSION

This review confirms rapidly increasing international interest in the analysis of multiple causes of death and provides the most comprehensive overview, to our knowledge, of methods and practices to date. Available multiple-cause methods are diverse but suit a range of research objectives. With greater availability of data and technology, these could be further developed and applied across a range of settings.

Cause-of-Death Diversity From a Multiple-Cause Perspective in the United States

Cause-of-death diversity captures the variability of deaths across causes and is an important marker of heterogeneity in a population's health. We contribute to the debate of cause-of-death diversity dynamics by following a novel multiple causes of death (MCOD) approach and applying it to the U.S. context between 2003 and 2018 and across education groups. Results show that cause-of-death diversity increased over this period, especially up to 2012. These trends were mainly driven by increases in the groups aged 65 years or more. The inclusion of MCOD resulted in higher increases in cause-of-death diversity over time compared with merely using underlying causes of death, except for the 85 or more age group, where no difference was observed for males and a reverted gradient was observed for females. Results by educational attainment reveal lower diversity among the highest educated groups and widening differences across groups from around 2012 onward. The clear educational gradient observed at ages 30-64 diminished at older ages. The observed increases in cause-of-death diversity should be monitored to better understand mortality dynamics in aging populations. Our new MCOD diversity measures suggest that traditional approaches relying on single causes of death might be underestimating cause-of-death diversity dynamics, particularly for males.

Patterns in age and cause of death contribution to the sex gap in life expectancy: a comparison among ten countries

Women live longer than men and the absolute difference between male and female mortality risk reaches its maximum at old ages. We decomposed the sex gap in life expectancy and investigated the changes over time of the profile of the age–cause specific contributions with indicators of location, magnitude and dispersion in ten countries. Data were retrieved from the Human Cause-of-Death Database. The decomposition analyses revealed that neoplasm, heart diseases and external causes were the main drivers of the gender gap. We also find two main patterns in the development of age-specific contributions. With mortality delay, regarding neoplasm-related mortality and heart disease-related mortality, the shift (i.e., movement of the modal age at contribution towards older ages) and compression (i.e., dispersion concentrated on a shorter age interval) of the survival advantage of women over a narrower age range reveal that men are gradually improving their survival. This might be linked to improvements in survival, diagnosis and access to treatment, at least to those ages no longer affected by the most significant differences.

Quantifying cause-related mortality in Australia, incorporating multiple causes: observed patterns, trends and practical considerations

BACKGROUND

Mortality statistics using a single underlying cause of death (UC) are key health indicators. Rising multimorbidity and chronic disease mean that deaths increasingly involve multiple conditions. However, additional causes reported on death certificates are rarely integrated into mortality indicators, partly due to complexities in data and methods. This study aimed to assess trends and patterns in cause-related mortality in Australia, integrating multiple causes (MC) of death.

METHODS

Deaths (n = 1 773 399) in Australia (2006-17) were mapped to 136 ICD-10-based groups and MC indicators applied. Age-standardized cause-related rates (deaths/100 000) based on the UC (ASRUC) were compared with rates based on any mention of the cause (ASRAM) using rate ratios (RR = ASRAM/ASRUC) and to rates based on weighting multiple contributing causes (ASRW).

RESULTS

Deaths involved on average 3.4 causes in 2017; the percentage with >4 causes increased from 20.9 (2006) to 24.4 (2017). Ischaemic heart disease (ASRUC = 73.3, ASRAM = 135.8, ASRW = 63.5), dementia (ASRUC = 51.1, ASRAM = 98.1, ASRW = 52.1) and cerebrovascular diseases (ASRUC = 39.9, ASRAM = 76.7, ASRW = 33.5) ranked as leading causes by all methods. Causes with high RR included hypertension (ASRUC = 2.2, RR = 35.5), atrial fibrillation (ASRUC = 8.0, RR = 6.5) and diabetes (ASRUC = 18.5, RR = 3.5); the corresponding ASRW were 12.5, 12.6 and 24.0, respectively. Renal failure, atrial fibrillation and hypertension ranked among the 10 leading causes by ASRAM and ASRW but not by ASRUC. Practical considerations in working with MC data are discussed.

CONCLUSIONS

Despite the similarities in leading causes under the three methods, with integration of MC several preventable diseases emerged as leading causes. MC analyses offer a richer additional perspective for population health monitoring and policy development.

Multimorbidity as assessed by reporting of multiple causes of death: variations by period, sociodemographic characteristics and place of death among older decedents in England and Wales, 2001-2017

BACKGROUND

Multimorbidity is common at older ages and is associated with disability, frailty and poor quality of life. Research using clinical databases and surveys has shown associations between multimorbidity and indicators of social disadvantage. Use of multiple coded death registration data has been proposed as an additional source which may also provide insights into quality of death certification.

METHODS

We investigate trends in reporting multiple causes of death during 2001-2017 among decedents aged 65 years and over included in a census-based sample of 1% of the England and Wales population (Office for National Statistics Longitudinal Study). Using Poisson regression analysis, we analyse variations in number of mentions of causes of death recorded by time period, place of death, age, sex and marital status at death and indicators of health status and individual and area socioeconomic disadvantage reported at the census prior to death.

RESULTS

Number of mentions of causes recorded at death registration increased 2001-2017, increased with age, peaking among decedents aged 85-9 years, and was positively associated with indicators of prior disadvantage and poor health, although effects were small. Number of mentions was highest for hospital decedents and similar for those dying in care homes or their own homes.

CONCLUSION

Socioeconomic disadvantage, prior poor health, dying in hospital and older age-although not extreme old age-are associated with dying with more recorded conditions. Results may reflect both differences in multimorbidity at death and variations in quality of medical certification of death. Quality of death certification for decedents in care homes needs further investigation.

Slowdown in Mortality Improvement in the Past Decade: A US/UK Comparison

OBJECTIVES

To investigate the slowdown in mortality improvement in the United States, United Kingdom, and comparator countries observed in the first decade of the twenty-first century and critically evaluate proposed explanations.

METHODS

Change-point analysis to identify the year of change in comparison of national mortality trends and linear spline models in the investigation of subnational differences using data from the Human Mortality Database, Global Burden of Disease cause-specific data, and, for the United Kingdom, national statistics data. Consideration of the impact of using different methods to estimate overall mortality is also concluded together with a review of methodological assumptions made in previous studies.

RESULTS

The results confirm the slowdown in mortality improvement observed in the early twenty-first century but indicate that proposed explanations for this are inadequate on a range of counts.

DISCUSSION

Mortality improvement slowed down in the early twenty-first century but the explanations advanced, such as opioid use in the United States or influenza epidemics and austerity programs in the United Kingdom, seem unlikely to account for this. Further research considering longer-term life course and cohort influences is needed.

Factors associated with under-reporting of head and neck squamous cell carcinoma in cause-of-death records: A comparative study of two national databases in France from 2008 to 2012

OBJECTIVE

To date, no study has evaluated the detection rate of head and neck squamous cell carcinoma (HNSCC) in cause-of-death records in Europe. Our objectives were to compare the number of deaths attributable to HNSCC from two national databases in France and to identify factors associated with under-reporting of HNSCC in cause-of-death records.

METHODS

The national hospital discharge database and the national underlying cause-of-death records were compared for all HNSCC-attributable deaths in adult patients from 2008 to 2012 in France. Factors associated with under-reporting of HNSCC in cause-of-death records were assessed using multivariate Poisson regression.

RESULTS

A total of 41,503 in-hospital deaths were attributable to HNSCC as compared to 25,647 deaths reported in national UCoD records (a detection rate of 62%). Demographics at death were similar in both databases with respect to gender (83% men), age (54% premature deaths at 25-64 years), and geographic distribution. In multivariate Poisson regression, under-reporting of HNSCC in cause-of-death records significantly increased in 2012 compared to 2010 (+7%) and was independently associated with a primary HNSCC site other than the larynx, a former primary or second synchronous cancer other than HNSCC, distant metastasis, palliative care, and death in hospitals other than comprehensive cancer care centers. The main study results were robust in a sensitivity analysis which also took into account deaths outside hospital (overall, 51,129 HNSCC-attributable deaths; a detection rate of 50%). For the year 2012, the age-standardized mortality rate for HNSCC derived from underlying cause-of-death records was less than half that derived from hospital discharge summaries (14.7 compared to 34.1 per 100,000 for men and 2.7 compared to 6.2 per 100,000 for women).

CONCLUSION

HNSCC is largely under-reported in cause-of-death records. This study documents the value of national hospital discharge databases as a complement to death certificates for ascertaining cancer deaths.

Multi-morbidity and frailty at death: A new classification of death records for an ageing world

Mortality statistics based on underlying cause of death are challenged by increased life expectancy and the growing share of population reaching ages associated with frequent multi-morbidity (with death likely resulting from interactions between multiple diseases). We provide a novel way of analysing causes of death: accounting for all causes mentioned on death certificates and summarizing this information along two dimensions emblematic of ageing populations-multi-morbidity and frailty. We implement this classification for all deaths at ages 50+ in Italy in 2014. Multi-morbid processes represent the majority of deaths, rising from 43 per cent at ages 50-54 to 63 per cent at ages 85-89. Multi-morbidity at death is more frequent among males, although age patterns are identical for both sexes. About one in four deaths involves frailty symptoms, rising to 45 per cent at ages 95+. Mortality rates involving frailty are very similar for both sexes. Supplementary material is available for this article at: https://doi.org/10.1080/00324728.2020.1820558.

Comparative Disease Assessment: a multi-causal approach for estimating the burden of mortality

Background

The Comparative Risk Assessment (CRA) framework comprehensively evaluates the impact of exposure to risk factors on health populations using the counterfactual causal approach.

Methods

We propose a framework, Comparative Disease Assessment (CDA), for assessing the impact of exposure to morbidity from some diseases on health outcomes, particularly death from other (relevant) diseases. This framework has been developed following the ideas of the CRA framework and using the widely accepted concept that exposure to morbidity is usually a risk factor for health outcomes (morbidity/mortality) related to other diseases. Our framework uses a counterfactual and not a categorical approach when attributing the burden of health outcomes to potential causes.

Results

This paper describes the different steps and assumptions required to implement the CDA framework, and an illustrative example is used considering diabetes mellitus morbidity as a risk factor for death from heart diseases.

Conclusions

One advantage of the CDA framework is that it can be applied using multi-causal death registries. Some assumptions are needed to implement it in order to avoid biases, but at least it can provide preliminary estimations of the impact of exposure to diseases as risk factors for deaths from other diseases. Another main advantage is that the burden of deaths is no longer attributed to a single cause, the underlying cause, as it is almost always done. Finally, this framework provides information on the pattern of comorbidity in a (sub)population of subjects who is about to die. These patterns can be used as a reference for alternative patterns of the general population or patterns of other specific subpopulations.

Mortality, morbidity and health in developed societies: a review of data sources

The purpose of this paper is to review the major sources of data on mortality, morbidity and health in Europe and in other developed regions in order to examine their potential for analysing mortality and morbidity levels and trends. The review is primarily focused on routinely collected information covering a whole country. No attempt is made to draw up an inventory of sources by country; the paper deals instead with the pros and cons of each source for mortality and morbidity studies in demography. While each source considered separately can already yield useful, though partial, results, record linkage among data sources can significantly improve the analysis. Record linkage can also lead to the detection of possible causal associations that could eventually be confirmed. More generally, Big Data can reveal changing mortality and morbidity trends and patterns that could lead to preventive measures being taken rather than more costly curative ones.

Obesity-related mortality in France, Italy, and the United States: a comparison using multiple cause-of-death analysis

OBJECTIVES

We investigate the reporting of obesity on death certificates in three countries (France, Italy, and the United States) with different levels of prevalence, and we examine which causes are frequently associated with obesity.

METHODS

We use cause-of-death data for all deaths at ages 50-89 in 2010-2011. Since obesity may not be the underlying cause (UC) of death, we compute age- and sex-standardized death rates considering all mentions of obesity (multiple causes or MC). We use cluster analyses to identify patterns of cause-of-death combinations.

RESULTS

Obesity is selected as UC in no more than 20% of the deaths with a mention of obesity. Mortality levels, whether measured from the UC or the MC, are weakly related to levels of prevalence. Patterns of cause-of-death combinations are similar across the countries. In addition to strong links with cardiovascular diseases and diabetes, we identify several less familiar associations.

CONCLUSIONS

Considering all mentions on the deaths certificates reduces the underestimation of obesity-related mortality based on the UC only. It also enables us to describe the various mortality patterns involving obesity.

Quantifying cause-related mortality by weighting multiple causes of death

Objective

To investigate a new approach to calculating cause-related standardized mortality rates that involves assigning weights to each cause of death reported on death certificates.

Methods

We derived cause-related standardized mortality rates from death certificate data for France in 2010 using: (i) the classic method, which considered only the underlying cause of death; and (ii) three novel multiple-cause-of-death weighting methods, which assigned weights to multiple causes of death mentioned on death certificates: the first two multiple-cause-of-death methods assigned non-zero weights to all causes mentioned and the third assigned non-zero weights to only the underlying cause and other contributing causes that were not part of the main morbid process. As the sum of the weights for each death certificate was 1, each death had an equal influence on mortality estimates and the total number of deaths was unchanged. Mortality rates derived using the different methods were compared.

Findings

On average, 3.4 causes per death were listed on each certificate. The standardized mortality rate calculated using the third multiple-cause-of-death weighting method was more than 20% higher than that calculated using the classic method for five disease categories: skin diseases, mental disorders, endocrine and nutritional diseases, blood diseases and genitourinary diseases. Moreover, this method highlighted the mortality burden associated with certain diseases in specific age groups.

Conclusion

A multiple-cause-of-death weighting approach to calculating cause-related standardized mortality rates from death certificate data identified conditions that contributed more to mortality than indicated by the classic method. This new approach holds promise for identifying underrecognized contributors to mortality.

Measuring causes of death in populations: a new metric that corrects cause-specific mortality fractions for chance

Background

Verbal autopsy is gaining increasing acceptance as a method for determining the underlying cause of death when the cause of death given on death certificates is unavailable or unreliable, and there are now a number of alternative approaches for mapping from verbal autopsy interviews to the underlying cause of death. For public health applications, the population-level aggregates of the underlying causes are of primary interest, expressed as the cause-specific mortality fractions (CSMFs) for a mutually exclusive, collectively exhaustive cause list. Until now, CSMF Accuracy is the primary metric that has been used for measuring the quality of CSMF estimation methods. Although it allows for relative comparisons of alternative methods, CSMF Accuracy provides misleading numbers in absolute terms, because even random allocation of underlying causes yields relatively high CSMF accuracy. Therefore, the objective of this study was to develop and test a measure of CSMF that corrects this problem.

Methods

We developed a baseline approach of random allocation and measured its performance analytically and through Monte Carlo simulation. We used this to develop a new metric of population-level estimation accuracy, the Chance Corrected CSMF Accuracy (CCCSMF Accuracy), which has value near zero for random guessing, and negative quality values for estimation methods that are worse than random at the population level.

Results

The CCCSMF Accuracy formula was found to be CCSMF Accuracy = (CSMF Accuracy - 0.632) / (1 - 0.632), which indicates that, at the population-level, some existing and commonly used VA methods perform worse than random guessing.

Conclusions

CCCSMF Accuracy should be used instead of CSMF Accuracy when assessing VA estimation methods because it provides a more easily interpreted measure of the quality of population-level estimates.

Electronic supplementary material

The online version of this article (doi:10.1186/s12963-015-0061-1) contains supplementary material, which is available to authorized users.

What do people die from? The challenges of measuring disease burden posed by multi-morbidity

Determining the precise cause of death of an individual has become increasingly difficult as life prolonging treatments mean that many people may die with multiple disorders, any of which, individually or in concert, might have killed them. Yet this problem has been recognised from the very earliest attempts to develop classifications of diseases. This paper traces the evolution of thinking on death certification and disease classification, describing what has been achieved but also identifying what more needs to be done to advance research and inform policy.

Multiple causes of death analysis of chronic diseases: the example of diabetes

BACKGROUND

Identifying a single disease as the underlying cause of death (UCOD) is an oversimplification of the clinical-pathological process leading to death. The multiple causes of death (MCOD) approach examines any mention of a disease in death certificates. Taking diabetes as an example, the study investigates: patterns of death certification, differences in mortality figures based on the UCOD and on MCOD, factors associated to the mention of diabetes in death certificates, and potential of MCOD in the analysis of the association between chronic diseases.

METHODS

The whole mortality archive of the Veneto Region-Italy was extracted from 2008 to 2010. Mortality rates and proportional mortality were computed for diabetes as the UCOD and as MCOD. The position of the death certificate where diabetes was mentioned was analyzed. Conditional logistic regression was applied with chronic liver diseases (CLD) as the outcome and diabetes as the exposure variable. A subset of 19,605 death certificates of known diabetic patients (identified from the archive of exemptions from medical charges) was analyzed, with mention of diabetes as the outcome and characteristics of subjects as well as other diseases reported in the certificate as predictors.

RESULTS

In the whole mortality archive, diabetes was mentioned in 12.3 % of death certificates, and selected as the UCOD in 2.9 %. The death rate for diabetes as the UCOD was 26.8 × 10(5) against 112.6 × 10(5) for MCOD; the UCOD/MCOD ratio was higher in males. The major inconsistencies of certification were entering multiple diseases per line and reporting diabetes as a consequence of circulatory diseases. At logistic regression the mention of diabetes was associated with the mention of CLD (mainly non-alcohol non-viral CLD). In the subset of known diabetic subjects, diabetes was reported in 52.1 %, and selected as the UCOD in 13.4 %. The probability of reporting diabetes was higher with coexisting circulatory diseases and renal failure and with long duration of diabetes, whereas it was lower in the presence of a neoplasm.

CONCLUSIONS

The use of MCOD makes the analysis of mortality data more complex, but conveys more information than usual UCOD analyses.

After the epidemiologic transition: a reassessment of mortality from infectious diseases among over-65s in France and Italy

OBJECTIVES

To assess more accurately the contribution of infectious diseases (IDs) to mortality at age 65+.

METHODS

We use cause-of-death data for France and Italy in 2009. In addition to chapter I of the 10th International Classification of Diseases (ICD-10), our list of IDs includes numerous diseases classified in other chapters. We compute mortality rates considering all death certificate entries (underlying and contributing causes).

RESULTS

Mortality rates at age 65+ based on our extended list are more than three times higher than rates based solely on ICD-10 chapter I. IDs are frequently contributing causes of death. In France, the share of deaths at age 65+ involving an ID as underlying cause increases from 2.1 to 7.3 % with the extended list, and to 20.8 % when contributing causes are also considered. For Italy, these percentages are 1.4, 4.2 and 18.7 %, respectively.

CONCLUSIONS

Publicly available statistics underestimate the contribution of IDs to the over-65s' mortality. Old age is a risk factor for IDs, and these diseases are more difficult to treat at advanced ages. Health policies should develop targeted actions for that population.

Mortality From Alzheimer’s Disease, Parkinson’s Disease, and Dementias in France and Italy

Objective: We perform an in-depth analysis of all death certificates collected in France and Italy with an entry of Parkinson’s disease (PD), Alzheimer’s disease (AD), or another dementia. Method: Data are for 2008. We measure how frequently these conditions are the underlying cause of death. We then examine what other causes are reported on the certificates. Results: In both countries, AD is the underlying cause for about 6 in 10 certificates with an AD entry. The proportion is lower for PD and dementia, but higher in France than in Italy. Many contributing causes reflect the circumstances surrounding the end of life in AD, PD, and dementia, often characterized by bed confinement and frailty. Discussion: Our research highlights several consequences of the conditions under study that could be targeted by public health policy. It also speaks to the existence of differences in diagnosis/certification practices that may explain differences in mortality levels.