Reference Guide for Pregnancy-Associated Death Identification

Overview

Accurately identifying deaths during and within 1 year of pregnancy (pregnancy-associated deaths) is a crucial first step to better understand:

• Leading causes of death.
• Contributing factors.
• Opportunities for preventing pregnancy-related deaths.

Relying only on death record information can underestimate and overestimate the number of identified deaths that occurred during or within 1 year of the end of pregnancy.1 Using additional data sources and linkage methods are key to overcoming these limitations.

This reference guide was developed to describe best practices for linking jurisdictional vital records data to identify pregnancy-associated deaths. Vital records include data from:

• Registration of births.
• Fetal deaths.
• Death reporting systems.

Improved identification of pregnancy-associated deaths may occur by linking death records of females ages 10–60 to pregnancy outcome information (birth or fetal death).

This guide and the following methods were developed by the Pregnancy-Associated Death Identification Workgroup. This workgroup included members from state health departments and the Centers for Disease Control and Prevention (CDC). The methods are considered best practices for pregnancy-associated death identification by the workgroup. The exact order for completing this algorithm may differ depending on the resources and protocols within individual jurisdictions.

Data Sources

Death Records

Death records should be selected for:

• Females ages 10–60.
• Residents of your jurisdiction, regardless of where the death occurred.

Provisional death files are preferred, and are sufficient, for more timely identification of deaths.

Birth Records and Fetal Death Records

Linking birth records and fetal death records occurring during the year prior to a female's death requires:

• Two calendar years of birth records data.
• Two calendar years of fetal death records data.

For example, if linking 2017 death records data, birth and fetal death records from both 2016 and 2017 should be included to determine pregnancy outcomes in the year prior to death.

Identifying Pregnancy-Associated Deaths by Vital Records Linkages

The application of both deterministic and probabilistic linkage, as suggested by the workgroup, as described below.

Deterministic Linkage Using Social Security Number

Death records of females ages 10–60 are linked to birth and fetal death records using deterministic record linkage.

• Files are matched on the female's social security number (SSN).

• Only matches where the death occurred less than or equal to 1 year from the date of birth and/or date of fetal death are retained.
  • For example, a birth or fetal death that occurred on June 30, 2017, requires exploring pregnancy-associated deaths through June 29, 2018.
• Matches identified should be reviewed for accuracy.

Probabilistic Record Linkage

Death records of females ages 10–60 not linked using deterministic linkage may be matched to birth and fetal death records using probabilistic linkage methods. 23Probabilistic record linkage methods require using linkage software such as Link Plus, Match*Pro, or statistical software such as SAS.23

• Blocking variables may include female's first name, female's last name, and female's date of birth.

Matching variables may include:

• Female's first name, last name, date of birth, zip code, SSN (if available), and maiden name.
  • Baby's last name.
  • Father's last name.

• Only matches where the death occurred less than or equal to 1 year from the date of birth and/or date of fetal death are retained.
• Matches identified should be reviewed for accuracy.

Identification of Pregnancy-Associated Deaths based on Cause(s) of Death Information

Some pregnancy-associated deaths, such as those that occur early during pregnancy, will not have birth or fetal death records to link. Approaches suggested by the workgroup for identifying pregnancy-associated deaths among those death records not using the deterministic and probabilistic linkage are described below.

Literal Cause of Death Fields in Death Record Data

Select death records where the literal cause of death fields contain any of the following pregnancy-related terms:

amniotic, chorioamnionitis, eclampsia, ectopic, intrauterine fetal demise, peripartum, peripartum cardiomyopathy, placental, postpartum, pregnancy, pregnant, uterine hemorrhage, and ulterine rupture

This is not an exhaustive list of potential pregnancy-related keywords. Additional terms may be added based on review of death record data. If using software to identify keywords, adding common misspellings may improve identification of pregnancy-associated deaths. For example, "eclmpsia" or "eclampsa" for eclampsia.

ICD-10 Code

An ICD-10 code, if available, may be used to identify pregnancy-associated death using the ICD-10 codes related to pregnancy (A34 and O00–O99.9). To be comprehensive, all the cause of death fields should be examined for the ICD-10 codes related to pregnancy.

O-codes are assigned based in part on the pregnancy checkbox on the death record. Therefore, these records should be confirmed with an additional information source to avoid misclassification. Examples of confirmatory sources are provided in the section on Additional Data Sources. Deaths identified only through ICD-10 codes need confirmation of pregnancy from the death certifier.

Identification of Pregnancy-Associated Deaths based on Pregnancy Checkboxes on the Death Records

Select death records where the pregnancy checkbox field indicates the female was:

• Pregnant at the time of death.
• Not pregnant, but pregnant within 42 days of death.
• Not pregnant, but pregnant 43 days to 1 year before death.

Selected records need to be confirmed with the death certifier to identify potential pregnancy checkbox errors.

Records are considered "false positives" when the following conditions are both met:

• The pregnancy checkbox on the death record indicates pregnancy.
• There is no documentation of pregnancy in the year preceding death documented in the confirmation process.

Examples of additional confirmatory sources identified by the workgroup are provided below.

Additional Data Sources

Additional sources can help confirm pregnancy for deaths which do not link to birth or fetal death records, but have pregnancy indicated by other information on the death record, such as the pregnancy checkbox.

Hospital and Emergency Department Records: Pregnancy status may be confirmed in labs, physician narratives, surgical histories, ultrasound results, medication lists, etc.

Obituaries: Can be found on obituary and funeral home websites.

Social Media: Facebook pages for both the deceased and family members, GoFundMe pages, etc.

Media and News Reports: Especially useful for deaths due to homicide, suicide, or motor-vehicle crashes. Helpful search terms include deceased's name, date of birth, date of death, cause of death, city of death, and/or county of death.

Certifier Confirmation: Vital Records staff may be able to confirm pregnancy checkbox information with the certifier.

Autopsy Reports: When available, these reports may provide information on pregnancy status.