EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Postpartum readmission for hypertension: Validating a predictive model with incorporation of neighborhood‐level social determinants of health

Celeste A Green iD, Brooklynn Earls, Michael Jochum iD, J Eric Jelovsek, Kartik Venkatesh iD, April D Adams, Dominique Williams

Pregnancy · 2026-05 · 2 (3)

DOI: 10.1002/pmf2.70288 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Pregnancy, online erschienen am 07.04.2026. Verfasst von Celeste A Green und 6 weiteren Autor:innen. Beteiligte Einrichtung: Division of Maternal Fetal Medicine Department of Obstetrics and Gynecology Baylor College of Medicine Houston Texas USA. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

To refine a pre‐existing predictive model (“the original calculator”) for postpartum readmission for hypertensive disorders of pregnancy (HDP), and to assess whether model performance is further enhanced by adding neighborhood‐level social determinants of health (SDOH).

Design

We conducted a retrospective cohort study of deliveries and readmissions from 2019 to 2022.

Setting

Two hospital systems in the southwestern United States. Site‐stratified analysis was planned due to differences in patient demographics at each hospital.

Participants

Participants included any person 18 years of age or older who delivered a live‐ or stillborn singleton neonate at 20 weeks or greater during the study period. Only the first pregnancy was included for those patients who had more than one delivery in the study period.

Methods

Postpartum readmission for HDP included inpatient stays with a diagnostic code of HDP occurring 1–42 days postdelivery discharge. Each hospital cohort's data were input to the original calculator to externally validate that model. Predictive model recalibration was then performed individually for each hospital's cohort. Model performance was measured using the concordance statistic (c‐statistic), Akaike Information Criterion (AIC), and calibration curves. The census tract‐level Social Vulnerability Index (SVI) score was added as a variable to the recalibrated calculators. Incremental prognostic information added by the SVI variable was measured in recalibrated models. The study received human subjects approval from our institutional review board.

Results

The frequency of readmission at Ben Taub Hospital (BTH) was 0.5% (48 readmissions/8646 deliveries) and at Texas Children's Hospital Pavilion for Women (PFW) was 0.7% (163 readmissions/22,966 deliveries). The median SVI score among readmitted patients was 0.84 (interquartile range [IQR] 0.71, 0.96) and 0.46 (IQR 0.18, 0.70) at BTH and PFW, respectively. Model discrimination for each of our cohorts using the original calculator was satisfactory, albeit with lower discriminative ability (c‐statistics were 0.71 and 0.61 for BTH and PFW, respectively), compared with the original calculator's cohort (c‐statistic 0.80). Model recalibration and re‐estimation were done for the existing variable coefficients. This recalibration demonstrated improved discriminative ability (c‐statistics for BTH and PFW were 0.84 and 0.74, respectively). Decision curve analyses showed a net benefit to using the recalibrated calculators when the average risk of readmission was between 0% and 3% (or three times the average rate). Addition of SVI to the model improved overall model fit for BTH (AIC 555 vs. 523) and for PFW (AIC 1850 vs. 1533). SVI added 2.0% to 3.4% additional prognostic information to BTH's model, and 36% to 50% additional prognostic information to PFW's model. Conclusions and relevance Refitting a pre‐existing risk prediction model on our hospital cohorts showed improved model performance compared with using the original calculator as published. Our models’ performances were slightly improved with the addition of the SVI variable. We highlight important risk factors to consider in calculators in development at other institutions, and emphasize that for postpartum HDP readmission, calculators should be context‐specific and require local recalibration for optimal use.

Schlagwörter

Stichwörter der Autor:innen

SVIpreeclampsiareadmissionrisk prediction

Autor:innen und Institutionen

  • Division of Maternal Fetal Medicine Department of Obstetrics and Gynecology Baylor College of Medicine Houston Texas USA.
  • Department of Obstetrics & Gynecology Baylor College of Medicine Houston Texas USA.
  • Division of Urogynecology & Pelvic Reconstructive Surgery Department of Obstetrics and Gynecology Duke University Durham North Carolina USA.
  • Division of Maternal Fetal Medicine Department of Obstetrics and Gynecology The Ohio State University Wexner Medical Center Columbus Ohio USA.
  • Department of Cardiology Ochsner Health System Slidell Louisiana USA.

ORCID-Profile: Celeste A Green, Michael Jochum, Kartik Venkatesh

Zitieren

APA

Green, C. A., Earls, B., Jochum, M., Jelovsek, J. E., Venkatesh, K., Adams, A. D., & Williams, D. (2026). Postpartum readmission for hypertension: Validating a predictive model with incorporation of neighborhood‐level social determinants of health. Pregnancy, 2(3). https://doi.org/10.1002/pmf2.70288

Vancouver

Green CA, Earls B, Jochum M, Jelovsek JE, Venkatesh K, Adams AD, et al. Postpartum readmission for hypertension: Validating a predictive model with incorporation of neighborhood‐level social determinants of health. Pregnancy. 2026;2(3). doi:10.1002/pmf2.70288

BibTeX
@article{Green2026,
  author = {Green, Celeste A and Earls, Brooklynn and Jochum, Michael and Jelovsek, J Eric and Venkatesh, Kartik and Adams, April D and Williams, Dominique},
  title = {Postpartum readmission for hypertension: Validating a predictive model with incorporation of neighborhood‐level social determinants of health},
  journal = {Pregnancy},
  year = {2026},
  volume = {2},
  number = {3},
  doi = {10.1002/pmf2.70288}
}
Kostenlos starten →