A recent article¹ described increases in MMRs from 1999 to 2019, racial and ethnic disparities, and differences by US state. The use of bayesian models to provide more precise estimates of MMRs for small racial and ethnic subgroups is a strength, but not a panacea for the problem of small numbers. The utility of state-level estimates by subgroup is questionable when driven mainly by the priors and associated with wide uncertainty intervals, a concern for subgroups with fewer than 20 events each year. Additionally, failure to account for differential adoption of the pregnancy checkbox by US states over time biased the trend estimates and resulting conclusions of this study.¹

The study authors acknowledged in the Discussion section that “it is possible that some of the increases in maternal mortality over time are due to an increasing number of states incorporating the pregnancy checkbox….”¹ However, it is insufficient to simply note the effect of the pregnancy checkbox as a limitation, given numerous prior studies describing trends in maternal mortality in the context of changes in ascertainment. These studies have quantified the effect of the incremental adoption of the pregnancy checkbox on the standard certificate of death (which occurred from 2003 to 2017), showing that observed increases in MMRs from the early 2000s to 2017 are entirely or nearly entirely an artifact of changes in measurement over time, with no significant trends in MMRs once the checkbox was accounted for.^2–4 It is highly likely that the increases reported in this study¹ were an artifact of changes in ascertainment.

Estimated racial and ethnic disparities in MMR trends are also subject to bias due to the incremental adoption of the checkbox by states over time, given the geographic concentration of specific subpopulations such as non-Hispanic American Indian or Alaska Native persons. The effect of checkbox implementation varies by age, race and ethnicity, state, and cause of death, with larger effects seen among people aged 35 years or older, among non-Hispanic Black individuals, and for nonspecific maternal causes of death.^2–4 Consequently, it is important to explicitly account for the effect of the pregnancy checkbox and other changes in measurement⁵ over time when estimating trends and racial and ethnic disparities in MMRs as well as state-level patterns. Without accurate and comparable measurement of MMR trends and disparities, conclusions cannot be drawn about the effect of prevention efforts.