The Swedish National Board of Health and Welfare (SNBHW)

Department of Registers and Statistics

Rålambsvägen 3

Socialstyrelsen

106 30 Stockholm, Sweden

Data on causes of death (CoD) provide information on mortality patterns and form a major element of public health information.

CoD data refer to the underlying cause which - according to the World Health Organisation (WHO) - is "the disease or injury which initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury".

CoD data are derived from death certificates. The information provided in the medical certificate of cause of death is mapped to the International Statistical Classification of Diseases and Related Health Problems (ICD).

Eurostat's CoD statistics build on standards set out by the World Health Organisation (WHO) in the International Statistical Classification of Diseases and Related Health Problems (ICD).

The regional breakdown is based on the Nomenclature of Territorial Units for Statistics (NUTS 2).

Classification and updates applied by years

Public Health

Concepts and definitions are described in the Commission regulation (EU) No 328/2011 in articles 2 and 3.

The place where a person normally spends the daily period of rest or municipality where the person is registered.

Up to and including June 2008: Death after 28 weeks of gestation. From July 2008: death after 22 weeks of gestation.

The statistical units are the deceased persons and the stillborns, respectively.

Residents and non-residents who have died in the country and stillbirths.

No

Non-residents are included in the CoD register since 2012 and the statistics are reported to Eurostat since 2011.  

Yes, coded and classified as for deaths in Sweden. A separate variable identifies the death as taking place abroad. 

Sweden or death among Swedish citizens, regardless of where the death occurs

Time series available from 2011 onwards, because 2011 data is the first data collection with a legal basis.

Not applicable.

The unit is number.

Data refer to the calendar year (i.e. all deaths occurring during the year).

CoD data was submitted to Eurostat on the basis of a gentleman's agreement established in the framework Eurostat's Working Group on "Public Health Statistics" until data with reference year 2010.

A Regulation on Community statistics on public health and health and safety at work (EC) No 1338/2008 was signed by the European Parliament and the Council on 16 December 2008. This Regulation is the framework of the data collection on the domain.

Within the context of this framework Regulation, a Regulation on Community statistics on public health and health and safety at work, as regards statistics on causes of death (EU) No 328/2011 was signed by the European Parliament and the Council on 5 April 2011. 

CoD data according to this regulation is submitted to Eurostat since reference year 2011.

CoD-data are shared with researchers, health authorities, WHO, Eurostat and media.

The Swedish Health Registries Act "Lag (1998:543) om hälsodataregister", SFS 2018:439 "Lag om ändring i lagen (1998:543) om hälsodataregister" and the Public Access to Information and Secrecy Act "Offentlighets- och sekretesslag (2009:400)" provides the legal basis for sharing data from the Cause of Death registry.

The Swedish Public Access to Information and Secrecy Act "Offentlighets- och sekretesslag (2009:400)" provides the legal basis for confidentiality of data in the Cause of Death registry.

All confidential microdata are treated according to the Swedish Health Registry Act and the Swedish Data Protection Act (2018:218), which incorporates the EU General Data Protection Regulation (GDPR).

Personally identifiable data is encrypted, only anonymous aggregated data is sent unencrypted.

We do not provide health data along with personal identity number or name. If directly identifiable data is provided, serial number and health data are provided on one data set and serial number and ID on another. The datasets can be connected by using serial numbers.

The Swedish cause of death statistics is released annually, usually within 8 months after the end of the reference year.

Release dates for official and other statistics published by the Swedish National board of Health and Welfare can be found in the publishing calendar on https://www.socialstyrelsen.se/statistik-och-data/statistik/publiceringskalender-for-statistik/.

Cause of Death statistics is released annually. Dissemination includes a press release, a short report, supplementary tables, and possibility of downloading statistics from our statistical database. Release of Cause of Death statistics is regulated in the Official Statistics Act (2001:99), in Swedish "Lag (2001:99) om den officiella statistiken") and the Official Statistics Ordinance (2001:100), in Swedish "Förordning (2001:100) om den officiella statistiken".

Annual

News releases are usually published online at the same time as the most recent reference year is made available.

News releases about cause of death statistics are published on the Swedish National Board of Helth and Welfare´s website, https://www.socialstyrelsen.se/.

Until 2015 pdf publications were published and available for free while paper publications were charged for.

Since 2016 the pdf-publications are replaced with a spreadsheet including tables and short fact sheet (maximum 4 pages). Both are available for free on the Swedish National Board of Helth and Welfare´s website, https://www.socialstyrelsen.se/.

An online database is available to the public with causes of death on a three character level of detail (ICD 10) and possibility of time series starting 1997 onwards, https://sdb.socialstyrelsen.se/if_dor. Numbers can be broken in geographic regions, sex, year and age in group of five years. The online database include causes of death among Swedish residents.

Micro data is only disseminated by charge for research purposes, https://www.socialstyrelsen.se/en/statistics-and-data/statistics/. For other cases only aggregated data is disseminated.

Not applicable.

Not applicable

Documentation for CoD statistics and registry is available on the website of the National Board of Health and Welfare.

99%

The Swedish Cause of Death Registry has fully implemented the IRIS system, with its associated rules and guidelines for selection of underlying cause of death, for routine coding.

If the medical data is incomplete additional information is requested from the institution where the certificate was issued. 

The causes of death data are based on a regulation, which defines scope, definitions of variables and characteristics of the data.

Overall assessment of data quality in the cause of death registry is considered good. Validation studies have been performed for specific causes of death in the past, and while limited in scope, overall result is that the cause of death in the registry is conform the information in the cause of death certificate.

The Swedish Cause of Death Registry is needed for the following purposes:

1. Monitoring causes of death over time
2. Preparation of national, regional and local statistics related to causes of death
3. Research
4. Planning, quality assurance and quality impovement in the Health and Care Services

Not available.

The Swedish Cause of Death registry has a completeness rate of 98-99% for the underlying cause of death. Physicians have up to three weeks to send the cause of death certificate to the National Board of Health and Welfare according to the Burial Ordinance (Begravningsförordning, 1990:1147). The National Board of Health and Welfare uses the death certificates to followup physicians that did not send within a three week period. Exception is made in this process for external cause of death, because the forensic medical examination takes more time. 

1. For mandatory variables: Sweden is providing all the mandatory variables, 100%

2. For voluntary variables: Not available

3. For additional variables: See "13.3.3.2. Item non-response - rate"

There is no recent assessment of the overall accuracy. Comparisons between hospital discharge records and the cause of death indicates that the overall accuracy of the cause of death register is good, see Unexplained differences between hospital and mortality data indicated mistakes in death certification: an investigation of 1094 deaths in Sweden during 1995.

Medical advances give doctors new and better opportunities to decide the cause of death, which contributes to increased accuracy in diagnosis. New methods and changes in diagnostic policy can also lead to ill-judgments which previously was often undiagnosed now more clearly seen in the statistics or vice versa. For example, new diagnostic methods and criteria to be part of the explanation for the large increase in dementia diagnoses seen in recent decades, although other causes are also possible. Changes in medical terminology is also reflected in the statistics. Examples include heart attack unlikely that diagnostic terminology in the early 1900s. The diagnosis came into general use after the Second World War, and myocardial infarction and other ischemic heart disease is now the leading cause of death.
Changes in administrative conditions can also affect the statistics. Since 1991 the majority of deaths that occurred outside the hospital are no longer forensically examined. New regulations also gave families a greater opportunity to refuse clinical autopsy. This may have contributed to an increase in the proportion of deaths with incompletely specified cause of death.

Not applicable. Data collection is from administrative sources.

Not applicable.

Not applicable.

Not applicable.

Not applicable.

Not applicable. Data collection is from administrative sources.

Not applicable.

Not applicable.

Not available. However, seeing how there is only around 1%-2% missing certificates and the missing certificates show no different distribution regarding age, sex and geographic region, the non-reponse error is believed to be very small.

The item non-response-rates are presented separately for residents and non-residents since the rate in many cases is sifnificantly higher among non-residents. 

Not applicable.

Not applicable.

Since 2019 causes of death get published around 6 months after the last day of the reference period. Before that time, the first publication was around 8 months after the last day of the reference period. This publication only include Swedisch residents. Even though some minor corrections due to delayed certificates are done afterwards this is also considered to be the final and complete result. 

Not applicable.

From data collection with reference year 2011 onwards, Eurostat asks for the submission of final data at national and regional level and related metadata for the year N at N+24 months, according to the Implementing Regulation (EC) No. 328/2011, Article 4.

Since June 2015 region of occurence is collected and included in the Swedish cause of death register. In the data for 2021, approximately 7 percent of the deaths lack information about the region of occurence. The region of the event is calculated based on the region of occurence, but where no information is available the the regional residence is used, unless the the death took place outside Sweden. In the latter region of occurence is stated as unknown. The process is nationally coordinated by the National Board of Health and Welfare, inkluding data collection and coding.

Not applicable.

Note that due to the fact that 2011 data is the first data collection with a legal basis (and few changes in the requested variables and breakdowns), the data between 1994-2010 and starting from 2011 are not always comparable (In part due to the different groupings of causes of deaths). Moreover time series for data on stillbirths starts in 2011 and no information on previous data is available.

As the reliability of mortality data comparability is affected over time by several factors. The most important are changes in diagnostic methods and medical concepts, changed the routine collection of cause of death data, new revisions of cause of death classification and amended rules for the classification of the underlying cause of death.

Revisions to the cause of death classification

Changes in medical theory and epidemiological interests means that cause of death classifications must be revised at intervals. Comparability over time is hampered not only by the diagnostic groups and diagnostic concepts are changing in connection with audits and other updates cause of death classifications, but also by changes in the rules for the classification of the underlying cause, see Table F.

Table F. Cause of death classification in Sweden

Each new classification is more detailed than before. New medical discoveries and theories can lead to diseases transferred from one division to another. One example is the HIV / AIDS in the ICD-9 were classified in Chapter IV (diseases of the blood and blood-forming organs) and ICD-10 Chapter I (certain infectious and parasitic diseases). Alzheimer's disease is another example of ICD-9 were classified in Chapter V (Mental Disorders) and ICD-10 Chapter VI (diseases of the nervous system).
In order to assess the effects of change classifications - both that diagnoses moved and that the classification rules changed - a selection of about 25,000 certificates for 1996 were coded according to both ICD-9 and ICD-10. For 4% of deaths were moved on the underlying cause of death from one chapter to another. According to the grouping in the abbreviated European Cause of Death List (65 groups) 7% ​​of deaths got a different cause of death in ICD-10 than in ICD-9. Some groups were affected more strongly than others. A description of the causes of death whose frequency is particularly affected by classification changes is at 2001 Cause of Death publication entitled Guidance on individual tables (table 1).

Updates to ICD-10
Unlike previous revisions of the ICD updated ICD-10 continuously. Decisions of updates are made by the directors of the international classification-center which WHO collaborates with. Proposal of updates prepared by the Update Reference Committee, which consists of experts nominated by the different classification-centres. A special group of experts, Mortality Reference Group, in charge of proposals concerning the classification of causes of death.
Updates apply to both specific codes in ICD-10 and the classification rules. Codes can be added or removed or and content of individual code groups can change. Changes in the rules for the classification of causes of death can lead to a death that were recorded in a certain group of ICD after the update are classified elsewhere. A full account of the changes agreed to be found on the WHO website for ICD-10: http://www.who.int/classifications/.

Analysis of shifts in statistics

Shifts in the statistics due to new diagnostic methods and concepts are difficult to identify, especially as changes usually comes gradually and are difficult to distinguish from such changes in pattern of causes of death that depends on an amended illness or injury patterns. However, the displacements due to new revisions of the ICD or change of classification practice can usually be recognized in that they occur suddenly. If a specific cause of death suddenly switch level, the explanation is thus probably a change in cause of death classification or classification routines.
The multiple causes of death may give some indication of a change for a particular underlying cause of death due to revised classification rules or have other causes. If the number of occurrences of the cause of death as multiple cause of death is stable but the number of cases in which the cause of death recorded as underlying cause of death changed the explanation is likely that the classification rules changed. If, however, both the number of multiple and underlying causes of death has changed is the explanation of another.

ICD 10 adoption from 1997.

Small changes in cause of death certificate in from june 2015. Added possibility to mark disease as chronic or acute. The result of this change was fewer unspecified causes of death thereafter.

Comparisons between hospital discharge records, demography statistics and the cause of death statistics indicate that the statistics are reconciliable with other data sources and statistical domains.

Not applicable.

Not applicable.

There is a small difference in the total number of deaths between the CoD and the number of deaths according to Statistics Sweden. The reason for this is that Statistics Sweden close their collection a little bit earlier than the National Board of Health and Welfare and therefore get a lower number of deaths.

The total cost and burden for the production of the Swedish cause of deah register is unknown. The cost and burden of the data collection is reduced by using validation and dissemination IT tools.

Published official statistics on causes of death is not changed unless errors in the production are detected. The CoD register is however updated with death certificates submitted late. This update happens within n+24 months after the reference year. The public statistical database is then updated in the same way. 

There is no systematic revision of previous years. Data are occasionally revised, e.g. if the "NUTS" changes or if a country notifies Eurostat about any changes in the data.The first publication of causes of death is done around 8 months after the last day of the reference period. This publication only include Swedisch residents. Even though some minor corrections due to delayed certificates are done afterwards this is also considered to be the final and complete result.Regarding the reporting of stillborn children, their place of death has previously been reported where the child's mother was registered. This has now been changed and the place where the stillborn child was actually born is now reported (since CoD 2020).

Not applicable.

Information about the death, cause of death and previous surgeries etc, is collected from death certificates. Other information like region of residence and country of birth is collected from the civil register. Information on stillbirths is collected from the medical birth registry.

Death certificates are sent to the Swedish National Board of Helath and Welfare continuously during the year as people are dying.  

Cause of death data were collected in a similar way since the population registration was transferred from the Swedish Church to the Tax Agency in 1991. The forms for the death certificate and death certificate has been revised on several occasions but the changes are not radical.

Since 1996, the death certificate from the forensic examinations submitted in electronic format. Since June 2015 electronic death certificates are available for other death certificates as well. However s smaller part is still recieved in paper format and scanned and recorded by the National Board of Health and Welfare. 

Table on certification (Percentage)

Stillbirths are not included in the national cause of death statistics. Stillbirths are recorded in our Medical Birth Register but the cause of death is not coded. 

Cause of death for neonates is registered on the regular cause of death certificate.

External source (civil register).

At the statistical process ICD codes are first allocated to the diagnostic expressions on the certificates using the software Iris. By the encoding it is also apparent where on the death certificate the physician wrote respective states. Underlying cause of death is classified using the ACME (Automated Classification of Medical Entities), a program made available by the National Center for Health Statistics in the United States.
If the medical data is incomplete additional information is requested from the institution where the certificate was issued. The processing also includes reviewing and correcting the individual variables, including that the certificate is completed and that the above values are reasonable considering, among other things, the deceased's age and gender. All coding is carried out at the National Board of Healt and Welfare by trained coders and not by the certifier.
Around 10 percent are autopsied.
Double coding is not performed.

In connection with classification system is often a revision of the second machining routines. With the transition to ICD-10 approached Sweden international coding practices and accepting now usually interpreted practices that are programmed into the acme, the software for the automatic selection of the underlying cause of death, which is increasingly being regarded as an international standard. For ICD-9 Sweden used a version of ACME adapted to Swedish coding practices.
Cause of death statistics were long severely delayed. In order to deal with the backlog, a number of simplifications in processing routines at work on the 2010 material a number of changes.
• Fewer requests to the doctor for further medical data
• Major loss was accepted
• Tentative causes of death were not verified manually
• No independent verification coding of deaths by violence and poisoning
• Traffic Category not manually coded, reintroduced from 2011 year material
• Deaths abroad was not classified but were coded as R99.8 (cause of death not defined), was removed from the 2012 materials
A consequence of this is that less well specified diagnoses slightly higher degree accepted as the underlying cause of death. An increase in both missing death certificates and insufficient specific causes of death could be seen for the 2010 statistics. Both of these have fortunately gone down somewhat since then. 

ICD codes for the underlying cause (% of the Total)

Unknown country/region (%) for residents and non-residents who died in the country

All deaths from the civil register are included in the cause of death registry. The civil register is maintained by the Swedish Tax Agency. The Swedish Tax Agency also informs Statistics Sweden on all deaths, which get published in Swedens officicial population statistics. The Cause of death statistics, however, show slightly more deaths than the official population statistics. This excess in number of deaths relative to population statistics consists largely of deaths reported to the Swedish Tax Agency later than January 31 of the year following the death. Such deaths are not included in the population statistics but taken into death statistics.

The physician fills in the cause of death certificate and send it to the National Board of Health and Welfare where the cause of death is coded according to the WHO regulation by cause of death coders.

Not applicable.

Not applicable.

Not applicable.

When reporting the causes of death for 2020, there was an error when Swedish citizens died abroad. There, the County of occurrence would be reported as UN99, but they have been given the NUTS2 code where they were registered. This has now been corrected for the reporting of the 2021 causes of death.