Maternal Deaths Using Coroner's Data: A Latent Class Analysis

OBJECTIVE

Knowledge regarding the antecedent clinical and social factors associated with maternal death around the time of pregnancy is limited. This study identified distinct subgroups of maternal deaths using population-based coroner's data, and that may inform ongoing preventative initiatives.

METHODS

A detailed review of coroner's death files was performed for all of Ontario, Canada, where there is a single reporting mechanism for maternal deaths. Deaths in pregnancy, or within 365 days thereafter, were identified within the Office of the Chief Coroner for Ontario database, 2004-2020. Variables related to the social and clinical circumstances surrounding the deaths were abstracted in a standardized manner from each death file, including demographics, forensic information, nature and cause of death, and antecedent health and health care factors. These variables were then entered into a latent class analysis (LCA) to identify distinct types of deaths.

RESULTS

Among 273 deaths identified in the study period, LCA optimally identified three distinct subgroups, namely, (1) in-hospital deaths arising during birth or soon thereafter (52.7% of the sample); (2) accidents and unforeseen obstetric complications also resulting in infant demise (26.3%); and (3) out-of-hospital suicides occurring postpartum (21.0%). Physical injury (22.0%) was the leading cause of death, followed by hemorrhage (16.8%) and overdose (13.3%).

CONCLUSION

Peri-pregnancy maternal deaths can be classified into three distinct sub-types, with somewhat differing causes. These findings may enhance clinical and policy development aimed at reducing pregnancy mortality.

Retraction notice to "Part II: a step-by-step guide to latent class analysis" [Journal of Clinical Epidemiology 148 (2022) 170-173]

This article has been retracted: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/policies/article-withdrawal). This article has been retracted at the request of the Editors-in-Chief, and with the full cooperation of the authors of the article. The article used notable portions of text from papers previously published by Pratik Sinha, Carolyn S. Calfee and Kevin L. Delucchi in Crit Care Med 49(2021) e63-e79 (https://doi.org/10.1097/CCM.0000000000004710) and by Bridget E. Weller, Natasha K. Bowen and Sarah J. Faubert in J Black Psychol 46(2020) 287-311 (https://doi.org/10.1177/0095798420930932). While these two papers were cited in the original article, a reader brought to the Editors' attention areas of verbatim text overlap without clear attribution through the use of quotation marks. One of the conditions of submission of a paper for publication is that authors declare explicitly that their work is original and has not appeared in a publication elsewhere. Re-use of any text must be appropriately cited and made visible/transparent to the reader. The scientific community takes a very strong view on this matter. We apologize to readers of the journal that this was not detected ahead of publication. The article has been republished with an updated version with clear attribution of all text where appropriate.

Part II: A step-by-step guide to latent class analysis

Latent class analysis (LCA) is an analytical approach for the identification of more homogeneous subgroups within an otherwise dissimilar patient population. In the current paper, Part II, we present a practical step-by-step guide for LCA of clinical data, including when LCA might be applied, selecting indicator variables, and choosing a final class solution. We also identify common pitfalls of LCA, and related solutions.

Part II: a step-by-step guide to latent class analysis

Latent Class Analysis (LCA) is an analytical approach for the identification of more homogeneous subgroups within an otherwise dissimilar patient population. In the current paper, Part II, we present a practical step-by-step guide for LCA of clinical data, including when LCA might be applied, selecting indicator variables, and choosing a final class model. We also identify some common pitfalls of LCA, and some related solutions.

Part I: A Friendly Introduction to Latent Class Analysis

Latent class analysis (LCA) offers a powerful analytical approach for categorizing groups (or "classes") within a heterogenous population. LCA identifies these hidden classes by a set of predefined features, known as "indicators". Unlike many other grouping analytical approaches, LCA derives classes using a probabilistic approach. In this first paper, we describe the common applications of LCA, and outline its advantages over other analytical subgrouping methods.

Comparison of Self-harm or Overdose Among Adolescents and Young Adults Before vs During the COVID-19 Pandemic in Ontario

IMPORTANCE

Self-harm and deaths among adolescents and young adults are notably related to drug poisonings and suicide. With the emergence of the COVID-19 pandemic, there are projections about a greater likelihood of such events arising among adolescents and young adults.

OBJECTIVE

To evaluate the risk of self-harm, overdose, and all-cause mortality among adolescents and young adults during the COVID-19 pandemic.

DESIGN, SETTING, AND PARTICIPANTS

This population-based cohort study took place in Ontario, Canada, where a universal health care system captures all emergency department (ED) visits, hospitalizations, and deaths. The participants included all adolescents and young adults born in Ontario between 1990 and 2006, who were aged 14 to 24 years between March 1, 2018, and June 30, 2021.

EXPOSURES

The COVID-19 pandemic era (April 1, 2020 to June 30, 2021), relative to the 2 years preceding the pandemic (March 1, 2018 to February 28, 2020).

MAIN OUTCOMES AND MEASURES

ED encounters or hospitalizations for self-harm or overdose. A secondary outcome was self-harm, overdose, or all-cause mortality. Cause-specific hazard models to estimate hazard ratios (HR) and 95% CIs were used for the primary outcome. Follow-up started at March 1, 2018, or the individual's 14th birthday, whichever was later, and age was used as the time scale.

RESULTS

In this study, 1 690 733 adolescents and young adults (823 904 [51.3%] female participants) were included with a median (IQR) age of 17.7 (14.1-21.4) years at start of follow-up. After 4 110 903 person-years of follow-up, 6224 adolescents and young adults experienced the primary outcome of self-harm or overdose during the pandemic (39.7 per 10 000 person-years) vs 12 970 (51.0 per 10 000 person-years) prepandemic, with an HR of 0.78 (95% CI, 0.75-0.80). The risk of self-harm, overdose, or death was also lower during than before the pandemic (HR, 0.78; 95% CI, 0.76-0.81), but not all-cause mortality (HR, 0.95; 95% CI, 0.86-1.05).

CONCLUSIONS AND RELEVANCE

Among adolescents and young adults, the initial 15-month period of the COVID-19 pandemic was associated with a relative decline in hospital care for self-harm or overdose.

Identifying maternal deaths with the use of hospital data versus death certificates: a retrospective population-based study

BACKGROUND

Accurate identification of maternal deaths is paramount for audit and policy purposes. Our aim was to determine the accuracy and completeness of data on maternal deaths in hospital and those recorded on a death certificate, and the level of agreement between the 2 data sources.

METHODS

We conducted a retrospective population-based study using data for Ontario, Canada, from Apr. 1, 2002, to Dec. 31, 2015. We used Canadian Institute for Health Information (CIHI) databases to identify deaths during inpatient, emergency department and same-day surgery encounters. We captured Vital Statistics deaths in the Office of the Registrar General, Deaths (ORGD) data set. Deaths were considered within 42 days and within 365 days after a pregnancy outcome (live birth, miscarriage, ectopic pregnancy or induced abortion) for all multiple and singleton pregnancies. We calculated agreement statistics and 95% confidence intervals (CIs).

RESULTS

Among 1 679 455 live births and stillbirths, 398 pregnancy-related deaths in the ORGD data set were mapped to a birth in CIHI databases, and 77 (16.2%) were not. Among 2 039 849 recognized pregnancies, 534 pregnancy-related deaths in the ORGD data set were linked to CIHI records, and 68 (11.3%) were not. Among live births and stillbirths, after pregnancy-related deaths in the ORGD data set not matched to a maternal death in the CIHI databases were removed, concordance measures between CIHI and ORGD records for maternal death within 42 days after delivery included a κ value of 0.87 (95% CI 0.82-0.91) and positive percent agreement of 0.88 (95% CI 0.83-0.94). The corresponding measures were similar for maternal death within 42 days after the end of a recognized pregnancy. When unlinked pregnancy-related deaths in the ORGD data set were retained, agreement measures declined for death within 42 days after a live birth or stillbirth (κ = 0.68, 95% CI 0.62-0.74). For maternal death within 365 days after a live birth or stillbirth, or after the end of a recognized pregnancy, the concordance statistics were generally favourable when unlinked pregnancy-related deaths in the ORGD data set were removed but were substantially declined when they were retained.

INTERPRETATION

Maternal mortality cannot be ascertained solely with the use of hospital data, including beyond 42 days after the end of pregnancy. To improve linkage, we propose including health insurance numbers on provincial and territorial medical death certificates.