Data quality, concepts and methodology

Vital Statistics – Stillbirth Database

Survey description

This is an administrative survey that collects demographic information annually from all provincial and territorial vital statistics registries on all stillbirths (fetal deaths) in Canada. Some data are also collected on stillbirths to Canadian resident women in selected American states.

The data are used to calculate basic indicators (such as counts and rates) on stillbirths to Canadian resident women. Information from this database is also used in the calculation of statistics, such as the late fetal death rate and the perinatal death rate. Information from the Stillbirth database is found in both the birth and death annual publications.

For Canada as a whole, it was impossible to compile a satisfactory series of vital statistics prior to 1921. Eight provinces initially joined the cooperative Canadian vital statistics system, leading to the publication of the first annual report for Canada in 1921; that report included Prince Edward Island, Nova Scotia, New Brunswick, Ontario, Manitoba, Saskatchewan, Alberta and British Columbia. Quebec began to participate in 1926 and Newfoundland in 1949 (after joining Confederation) and their data were included in the tabulations from those years onward. Basic data from the Yukon and Northwest Territories were published as appendices to the national tables from 1924 to 1955; their data were first included in the regular tabulations in 1956. Nunavut came into being officially as a Territory of Canada on April 1, 1999. The name Northwest Territories applies to a Territory with different geographic boundaries before and after April 1, 1999.

Prior to 1944 all vital events were classified by place of occurrence. Since 1944, births, stillbirths, and deaths have been classified by area of reported residence, with births and stillbirths according to the residence of the mother.

Stillbirth is currently defined as the complete expulsion or extraction from its mother of a product of conception, which did not at any time after birth breathe or show other sign of life. In 2004, most provinces and all three territories required a stillbirth with a gestational age of at least 20 weeks or a birth weight of at least 500 grams to be registered. In Quebec (as well as in Saskatchewan prior to 2001 and in New Brunswick prior to November 1996), only stillbirths weighing at least 500 grams were required to be registered, regardless of the gestational age. Until 1997, a gestational age of at least 20 weeks was required for stillbirths to be registered in Prince Edward Island, regardless of the birth weight.

History

Starting in 1959, the definition of a stillbirth was revised to conform, in substance, to the definition of “fetal death” recommended by the World Health Organization. At the same time, the compulsory registration of stillbirths was extended to 20 weeks’ gestation – from 28 weeks – and the new period of gestation incorporated into the definition: “Stillbirth means the complete expulsion or extraction from its mother, after at least 20 weeks’ pregnancy, of a product of conception in which, after such expulsion or extraction, there is no breathing, beating of the heart, pulsation of the umbilical cord, or unmistakable movement of voluntary muscle.”

Provinces implemented the new definition at different times, as shown in the table below:

Prior to 1959 the following definition of stillbirth was incorporated in the vital statistics legislation of the provinces: "Stillbirth means the birth of a fetus, after at least 28 weeks' pregnancy, which, after complete separation from the mother, does not show any sign of life."

The underlying cause of stillbirth variable in the stillbirth database is classified according to the World Health Organization "International Statistical Classification of Diseases and Related Health Problems" (ICD). The following table shows the data years for which each revision of this classification was used. Data users must note that underlying cause of stillbirth data coded to different revisions of the classification are not comparable and they should contact Statistics Canada for assistance with the use of this variable across classification revisions.

An online version of ICD-10, second edition, (in English) became accessible as of October 2004 on the World Health Organization (WHO) website (www.who.int/classifications/en). The second edition incorporates the updates to ICD-10 that came into effect up to January 1, 2003.

Data sources and methodology

Survey population

The conceptual universe of the Stillbirth database is stillbirths to Canadian resident women anywhere in the world. The target population of the Stillbirth database is stillbirths to Canadian resident women in Canada and to Canadian resident women in American states. The actual (survey) population of the Stillbirth database is stillbirths to Canadian resident women and non-resident women in Canada, and stillbirths to Canadian resident women in some American states.

Sampling

This survey is a census with a cross-sectional design.

Collection

Responding to this survey is mandatory. Data are extracted from administrative files.

Provincial and territorial Vital Statistics Acts (or equivalent legislation) render compulsory the registration of all live births, stillbirths, deaths and marriages within their jurisdictions. These Acts follow, as closely as possible, a Model Vital Statistics Act that was developed to promote uniformity of legislation and reporting practices among the provinces and territories.

The Canadian Vital Statistics system operates under an agreement between the Government of Canada and governments of the provinces and territories. The Vital Statistics Council for Canada, an advisory committee set up by an Order-in-Council, oversees policy and operational matters. All provincial and territorial jurisdictions and Statistics Canada are represented on the Vital Statistics Council. Under the agreement, all registrars collect a specified set of data elements, although any of them may decide to collect additional information.

The form for the registration of a stillbirth is usually completed by the parents, who are responsible for filing it with the local, provincial or territorial registrar. Stillbirth registration requires a medical certificate of the cause of stillbirth to be completed by a physician or coroner.

The central Vital Statistics Registry in each province and territory provides data from stillbirth registrations to Statistics Canada. The following statistical data items are reported for each stillbirth by all provinces and territories for inclusion in the Canadian Vital Statistics system:

• Date and place of stillbirth
• Sex, birth weight and gestational age of fetus (See "History" for further information on definitions of stillbirth related to birth weight and gestational age.)
• Underlying cause of stillbirth classified to the World Health Organization "International Statistical Classification of Diseases and Related Health Problems" (ICD) (See "History" for further information about this variable).
• Parents' age, marital status and birthplace
• Mother's place of residence
• Type of birth (single or multiple)
• Parity

All provinces and territories supply microfilm copies or optical images of registration forms to Statistics Canada. In addition, Prince Edward Island, Nova Scotia, New Brunswick, Quebec, Ontario and the Western provinces supply machine-readable abstracts of registrations, which contain the required standard information. For Newfoundland and Labrador, and the territories, the required standard information on microfilm is converted to machine-readable format at Statistics Canada. Subsequent changes to registrations due to errors or omissions are transmitted to Statistics Canada as the information becomes available. However, changes received after a cut-off date are not reflected in published tabulations.

Error detection

Provinces and territories that supply machine-readable data carry out edits (presence of the data, validation of code ranges, and data consistency) before transmitting their data, based on standard edit specifications prepared by Statistics Canada. Health Statistics Division has actively promoted the use of a standard data dictionary and standard correlation edits for provincial/territorial data entry. More extensive edit routines are applied to the data by Statistics Canada to ascertain the completeness and quality of the data. For example, additional edits for multiple births identify possible errors and inconsistencies between the stillbirth and birth databases. If the characteristics of the mother of triplets (1 live born and 2 stillborn) are different on the separate registration forms, manual updates make these data consistent on both the birth and stillbirth databases. For 2004, about 45% of the records were assessed for follow-up action either by referring to the microfilmed registrations or optical images or by consulting with the registries. After the preparation of a preliminary data file, verification tables are prepared for data review by the registries and Statistics Canada (for example, distributions, large changes, percentage and number of unknowns, outliers, changes in the relative composition).

The last comprehensive study of the quality of data capture and data coding was done in 1981, when error rates for most variables were found to be quite low. Most provinces do their own data capture, but because of the small size of the Stillbirth database, it is often faster for Statistics Canada to re-capture the records from the microfilms rather than wait for electronic files which usually require reformatting. For 2004 data, Operations and Integration Division (OID) data captured about 43% (approximately 900) of the stillbirth records. OID maintains data capture quality controls such as 100% verification for new clerks, and sample batch re-capture for experienced clerks to maintain an error rate of less than 3%. For 2003 data, Statistics Canada captured the records of unscreened data from Ontario, which resulted in a higher than usual number of stillbirths. Normal Statistics Canada procedures such as duplicate detection, were carried out on the data.

Quality evaluation

Upon completion of the annual national stillbirth data base (produced as described in the section Error Detection above), Statistics Canada carries out a series of quality checks that include:

1. producing a set of verification tables which consist of basic tabulations for the majority of variables in the data base by province or territory of occurrence;
2. sending the verification tables to each provincial/territorial registrar of vital statistics for their review and approval that Statistics Canada and the registry obtain the same results;
3. checking for internal consistencies, for example, running frequencies and looking for outliers on certain data elements; and
4. comparing the most recent data year with past data years to detect any unusual or unexpected changes.

Comparisons of tabulated data are made with vital statistics data published by the provinces and territories, where available. After Statistics Canada creates the publication data file, the availability of stillbirth statistics is announced in The Daily.

Disclosure control

Statistics Canada is prohibited by law from releasing any data which would divulge information obtained under the Statistics Act that relates to any identifiable person, business or organization without the prior knowledge or the consent in writing of that person, business or organization. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.

Data accuracy

Coverage

Since the registration of stillbirths is a legal requirement in each Canadian province and territory, reporting is virtually complete, with some important exceptions. Québec uses slightly more limited reporting criteria (500 or more grams birth weight, compared with 500 or more grams birth weight or a gestational age of at least 20 weeks used in other provinces). New Brunswick does not require the registration of stillbirths meeting the reporting criteria when the fetus is removed during a therapeutic abortion. Stillbirth rates may be slightly underestimated in these provinces as a result. Under-coverage may also occur because of late registration. Some stillbirths are registered by local authorities, but the paperwork is not forwarded to provincial or territorial registrars before a cut-off date. These cases for 1999 represent approximately 26 stillbirths, 7 years after the year of stillbirth (accumulated late records), or one percent of the total records.

Other missing registrations may occur with Canadian women who have a stillbirth outside of Canada. Only stillbirths in the United States are regularly reported to Statistics Canada, and of these, Statistics Canada has received only three records over the past decade.

Over-coverage is minimal. Stillbirths to non-resident women in Canada are registered but are excluded from most tabulations. Duplicate stillbirth registrations are identified as part of the regular processing operations on each provincial and territorial subset, as well as by additional inter-provincial checks. Additional edits for multiple births identify possible duplicates or missing records between the stillbirth and birth databases. Possible duplicate registrations are checked against microfilmed registrations or optical images, or by consulting with the provinces and territories.

Response rates

Item response

For 1997 to 2004, the response rates were 98% to 100% for most of the demographic variables on the stillbirth database (age of mother, sex of fetus, province of residence of mother). The response rate for last name of mother and date of birth were both at 100%. The underlying cause of stillbirth is unknown in about 20% of the records, in most cases because the certifier was unsure of the cause, and not because of a data capture or processing error or omission. Postal codes are not well reported and/or captured, with only 60% of the records having postal codes.

Other accuracy issues

Underlying cause of stillbirth certification

There are two stages in the determination of an underlying cause of stillbirth: certification done by the certifier, and classification (coding) done by a medical coder. When a stillbirth occurs, the medical doctor in attendance, or the coroner, medical examiner or other certifier completes the medical certificate of stillbirth. The certificate consists of several sections eliciting the direct cause of stillbirth, antecedent causes, and other significant conditions of the mother and fetus. No quality studies specifically focusing on stillbirth certification have been done in Canada, but quality studies done on the certification of deaths have shown that approximately one-third of certificates contain major errors. Most of the errors (about 85%) involve the use of non-specific conditions such as “stroke” or “heart failure”. Approximately 10% of the errors involve illogical sequences of conditions entered in on the line items. Some of these can be resolved during coding, where the classification rules pinpoint illogical sequences. The remaining 5% involve competing causes. Proper training of certifiers, however, can greatly reduce these errors. Statistics Canada is working with the provincial and territorial registrars on two projects to improve quality: a half-day workshop suitable for continuing medical education credit, and an on-line tutorial.