Women’s Health
Trends in risk of pregnancy loss among US women by urban-rural residence, 2000-2018: a new tool for small domain estimation Sarah Forrest* Sarah Forrest Lauren Rossen Katherine Ahrens
Approximately 20% of pregnancies end in spontaneous loss, an outcome associated with adverse physical and psychological consequences. A previous study described increasing trends in the risk of self-reported pregnancy loss from 1990 to 2011, but it is unclear if those trends have continued or if they vary by factors such as urban-rural residence. Sample sizes of rural populations are typically small, limiting the availability of stable estimates for this population.
Data from the National Survey of Family Growth (2006–2019) were used to estimate rates of self-reported pregnancy loss (miscarriage, stillbirth, ectopic pregnancy) among US women (15–44 years) who reported at least one pregnancy conceived during 2000–2018 that did not result in induced termination (n = 17,314 women; n = 35,988 pregnancies). Trends in self-reported pregnancy loss were estimated by age group and urban-rural residence. A new small domain estimation tool, the Enhanced Modified Kalman Filter, was used to smooth estimates over groups and time. We compared relative 95% confidence interval (95% CI) widths ([upper bound-lower bound]/estimate) of model-based estimates to direct estimates.
Relative 95% CIs for model-based estimates were 35% and 53% smaller for urban and rural groups, respectively, than for direct estimates. The risk of self-reported pregnancy loss increased by a relative 3% annually for both urban and rural women 15-44 years (rate ratios [RRs]: 1.03, 95% CIs: 1.01, 1.04) from 2000-2018. The risk of loss increased among urban women aged 20-24 (RR: 1.01, 95% CI: 1.00, 1.01), 25-29 (RR: 1.01, 95% CI: 1.01, 1.02), and 30-34 (RR: 1.04, 95% CI: 1.03, 1.05) and among rural women aged 30-34 (RR: 1.02, 95% CI: 1.01, 1.03).
From 2000 to 2018, risk of self-reported pregnancy loss increased by a relative 1-4% annually among several age and urban-rural groups. A new small domain estimation tool provided substantial improvements in estimate precision relative to direct estimates.
