Risk of delivery complications among pregnant people experiencing housing insecurity

Bacterial Vaginosis Treatment Patterns, Associated Complications, and Health Care Economic Burden of Women With Medicaid Coverage in the United States

BACKGROUND

Bacterial vaginosis (BV) is a highly prevalent vaginal infection.

OBJECTIVES

Primary objectives of this study were to examine treatment patterns among female patients with Medicaid coverage who were diagnosed with incident BV, the frequency of BV-associated complications, and health care resource utilization (HCRU) and associated costs of incident BV and its recurrence. Secondary objectives were to identify predictors of total all-cause health care costs and number of treatment courses.

METHODS

Female patients aged 12-49 years with an incident vaginitis diagnosis and ≥1 pharmacy claim for a BV medication were selected from the Merative MarketScan Medicaid database (2017-2020). Additional treatment courses were evaluated during a ≥12-month follow-up period, in which new cases of BV-associated complications and HCRU and the associated costs were also measured. Generalized linear models were used to identify baseline predictors of total all-cause health care costs and number of treatment courses.

RESULTS

An incident vaginitis diagnosis and ≥1 BV medication claim were present in 114 313 patients (mean age: 28.4 years; 48.6% black). During the follow-up, 56.6% had 1 treatment course, 24.9% had 2, 10.2% had 3, and 8.3% had ≥4; 43.4% had BV recurrence. Oral metronidazole (88.5%) was the most frequently prescribed medication. Nearly 1 in 5 had a new occurrence of a BV-associated complication; most (76.6%) were sexually transmitted infections (STIs). Total all-cause and BV-related costs averaged $5794 and $300, respectively, per patient; both increased among those with more treatment courses. Older age, pregnancy, comorbidity, any STIs, postprocedural gynecological infection (PGI), and infertility were predictive of higher total all-cause health care costs, while race/ethnicity other than white was predictive of lower costs. Older age, black race, any STIs, pelvic inflammatory disease, and PGI were predictive of >1 treatment courses.

CONCLUSION AND RELEVANCE

The high recurrence of BV represents an unmet need in women's health care and better treatments are necessary.

Using Large Language Models to Annotate Complex Cases of Social Determinants of Health in Longitudinal Clinical Records

Social Determinants of Health (SDoH) are an important part of the exposome and are known to have a large impact on variation in health outcomes. In particular, housing stability is known to be intricately linked to a patient's health status, and pregnant women experiencing housing instability (HI) are known to have worse health outcomes. Most SDoH information is stored in electronic health records (EHRs) as free text (unstructured) clinical notes, which traditionally required natural language processing (NLP) for automatic identification of relevant text or keywords. A patient's housing status can be ambiguous or subjective, and can change from note to note or within the same note, making it difficult to use existing NLP solutions. New developments in NLP allow researchers to prompt LLMs to perform complex, subjective annotation tasks that require reasoning that previously could only be attempted by human annotators. For example, large language models (LLMs) such as GPT (Generative Pre-trained Transformer) enable researchers to analyze complex, unstructured data using simple prompts. We used a secure platform within a large healthcare system to compare the ability of GPT-3.5 and GPT-4 to identify instances of both current and past housing instability, as well as general housing status, from 25,217 notes from 795 pregnant women. Results from these LLMs were compared with results from manual annotation, a named entity recognition (NER) model, and regular expressions (RegEx). We developed a chain-of-thought prompt requiring evidence and justification for each note from the LLMs, to help maximize the chances of finding relevant text related to HI while minimizing hallucinations and false positives. Compared with GPT-3.5 and the NER model, GPT-4 had the highest performance and had a much higher recall (0.924) than human annotators (0.702) in identifying patients experiencing current or past housing instability, although precision was lower (0.850) compared with human annotators (0.971). In most cases, the evidence output by GPT-4 was similar or identical to that of human annotators, and there was no evidence of hallucinations in any of the outputs from GPT-4. Most cases where the annotators and GPT-4 differed were ambiguous or subjective, such as "living in an apartment with too many people". We also looked at GPT-4 performance on de-identified versions of the same notes and found that precision improved slightly (0.936 original, 0.939 de-identified), while recall dropped (0.781 original, 0.704 de-identified). This work demonstrates that, while manual annotation is likely to yield slightly more accurate results overall, LLMs, when compared with manual annotation, provide a scalable, cost-effective solution with the advantage of greater recall. At the same time, further evaluation is needed to address the risk of missed cases and bias in the initial selection of housing-related notes. Additionally, while it was possible to reduce confabulation, signs of unusual justifications remained. Given these factors, together with changes in both LLMs and charting over time, this approach is not yet appropriate for use as a fully-automated process. However, these results demonstrate the potential for using LLMs for computer-assisted annotation with human review, reducing cost and increasing recall. More efficient methods for obtaining structured SDoH data can help accelerate inclusion of exposome variables in biomedical research, and support healthcare systems in identifying patients who could benefit from proactive outreach.

The Impact of Social Vulnerability on Substance Use Detection Practices in Pregnancy

This study aimed to synthesize current literature regarding the impact of social vulnerability on pregnancy-related substance use detection in the United States and highlight disparities in substance use detection practices. Clinicaltrials.gov, Google Scholar, PubMed (includes MEDLINE), and Cochrane Library databases were searched using the following Medical Subject Headings (MeSH): (["pregnancy" or "prenatal"] AND ["substance use screening" or "urine toxicology testing" or "toxicology testing" or "urine drug screening" or "CRAFFT" or "4P's" or "4P's Plus" or "NIDA Quick Screen" or "DAST-10" or "SURP-P" or "WIDUS"], AND ("bias" or "disparities" or "social vulnerability"]). The search included systematic reviews, prospective and retrospective studies, randomized controlled trials, case studies, and qualitative and quantitative research from January 2014 through November 2023. Selected literature was limited to studies published in English, which included a study population of either pregnant individuals or pregnancy health care providers in the United States, and that were focused on inequities in pregnancy substance use detection. Using Covidence, three authors screened abstracts, and two screened full articles for inclusion. The included studies were evaluated for quality of evidence using the mixed methods appraisal tool. The search yielded 4,188 manuscripts; 37 were eligible for full review. A total of 18 manuscripts were included based on the relevancy of the topic. The most common social vulnerability domain identified was minority status (17/18), followed by socioeconomic status (11/18), household characteristics (8/18), and housing type (1/18). Social vulnerability plays a role in substance use detection among pregnant individuals. Most notably, race and ethnicity, age, and public insurance lead to increased rates of detection, though most individual factors need to be studied in greater depth. This study was registered with PROSPERO (PROSPERO ID CRD42022352598), the International Prospective Register of Systematic Reviews. KEY POINTS: · Socially vulnerable pregnant individuals are more likely to receive substance use screening or urine toxicology testing.. · Race, ethnicity, age, and insurance influence substance use detection disparities.. · More research is needed to understand how other characteristics influence disparities in substance use detection..

Abortion Restrictiveness and Infant Mortality: An Ecologic Study, 2014-2018

INTRODUCTION

The U.S. has the highest infant mortality rate among peer countries. Restrictive abortion laws may contribute to poor infant health outcomes. This ecological study investigated the association between county-level infant mortality and state-level abortion access legislation in the U.S. from 2014 to 2018.

METHODS

A multivariable regression analysis with the outcome of county-level infant mortality rates, controlling for the primary exposure of state-level abortion laws, and county-level factors, county-level distance to an abortion facility, and state Medicaid expansion status was performed. Incidence rate ratios and 95% CIs were reported. Analyses were conducted in 2022-2023.

RESULTS

There were 113,397 infant deaths among 19,559,660 live births (infant mortality rate=5.79 deaths/1,000 live births; 95% CI=5.75, 5.82). Black infant mortality rate (10.69/1,000) was more than twice the White infant mortality rate (4.87/1,000). In the multivariable model, increased infant mortality rates were seen in states with ≥8 restrictive laws, with the most restrictive (11-12 laws) having a 16% increased infant mortality level (adjusted incidence rate ratios=1.162; 95% CI=1.103, 1.224). Increased infant mortality rates were associated with increased county-level Black race individuals (adjusted incidence rate ratios=1.031; 95% CI=1.026, 1.037), high school education (adjusted incidence rate ratios=1.018; 95% CI=1.008, 1.029), maternal smoking (adjusted incidence rate ratios=1.025; 95% CI=1.018, 1.033), and inadequate prenatal care (adjusted incidence rate ratios=1.045; 95% CI=1.036, 1.055).

CONCLUSIONS

State-level abortion law restrictiveness is associated with higher county-level infant mortality rates. The Supreme Court decision on Dobbs versus Jackson and changes in state laws limiting abortion may affect future infant mortality.

Trends, Characteristics, and Maternal Morbidity Associated With Unhoused Status in Pregnancy

Importance

Unhoused status is a substantial problem in the US. Pregnancy characteristics and maternal outcomes of individuals experiencing homelessness are currently under active investigation to optimize health outcomes for this population.

Objective

To assess the trends, characteristics, and maternal outcomes associated with unhoused status in pregnancy.

Design, Setting, and Participants

This cross-sectional study analyzed data from the Healthcare Cost and Utilization Project National (Nationwide) Inpatient Sample. The study population included hospitalizations for vaginal and cesarean deliveries from January 1, 2016, to December 31, 2020. Unhoused status of these patients was identified from use of International Statistical Classification of Diseases, Tenth Revision, Clinical Modification code Z59.0. Statistical analysis was conducted from December 2022 to June 2023.

Main Outcomes and Measures

Primary outcomes were (1) temporal trends; (2) patient and pregnancy characteristics associated with unhoused status, which were assessed with a multivariable logistic regression model; (3) delivery outcomes, including severe maternal morbidity (SMM) and mortality at delivery, which used the Centers for Disease Control and Prevention definition for SMM indicators and were assessed with a propensity score-adjusted model; and (4) choice of long-acting reversible contraception method and surgical sterilization at delivery.

Results

A total of 18 076 440 hospital deliveries were included, of which 18 970 involved pregnant patients who were experiencing homelessness at the time of delivery, for a prevalence rate of 104.9 per 100 000 hospital deliveries. These patients had a median (IQR) age of 29 (25-33) years. The prevalence of unhoused patients increased by 72.1% over a 5-year period from 76.1 in 2016 to 131.0 in 2020 per 100 000 deliveries (P for trend < .001). This association remained independent in multivariable analysis. In addition, (1) substance use disorder (tobacco, illicit drugs, and alcohol use disorder), (2) mental health conditions (schizophrenia, bipolar, depressive, and anxiety disorders, including suicidal ideation and past suicide attempt), (3) infectious diseases (hepatitis, gonorrhea, syphilis, herpes, and COVID-19), (4) patient characteristics (Black and Native American race and ethnicity, younger and older age, low or unknown household income, obesity, pregestational hypertension, pregestational diabetes, and asthma), and (5) pregnancy characteristics (prior uterine scar, excess weight gain during pregnancy, and preeclampsia) were associated with unhoused status in pregnancy. Unhoused status was associated with extreme preterm delivery (<28-week gestation: 34.3 vs 10.8 per 1000 deliveries; adjusted odds ratio [AOR], 2.76 [95% CI, 2.55-2.99]); SMM at in-hospital delivery (any morbidity: 53.8 vs 17.7 per 1000 deliveries; AOR, 2.30 [95% CI, 2.15-2.45]); and in-hospital mortality (0.8 vs <0.1 per 1000 deliveries; AOR, 10.17 [95% CI, 6.10-16.94]), including case fatality risk after SMM (1.5% vs 0.3%; AOR, 4.46 [95% CI, 2.67-7.45]). Individual morbidity indicators associated with unhoused status included cardiac arrest (AOR, 12.43; 95% CI, 8.66-17.85), cardiac rhythm conversion (AOR, 6.62; 95% CI, 3.98-11.01), ventilation (AOR, 6.24; 95% CI, 5.03-7.74), and sepsis (AOR, 5.37; 95% CI, 4.53-6.36).

Conclusions and Relevance

Results of this national cross-sectional study suggest that unhoused status in pregnancy gradually increased in the US during the 5-year study period and that pregnant patients with unhoused status were a high-risk pregnancy group.

Impact of Adverse Gestational Milieu on Maternal Cardiovascular Health

Cardiovascular disease affects 1% to 4% of the nearly 4 million pregnancies in the United States each year and is the primary cause of pregnancy-related mortality. Adverse pregnancy outcomes are associated with cardiovascular complications during pregnancy persisting into the postpartum period. Recently, investigations have identified an altered sex hormone milieu, such as in the case of hyperandrogenism, as a causative factor in the development of gestational cardiovascular dysfunction. The mechanisms involved in the development of cardiovascular disease in postpartum women are largely unknown. Animal studies have attempted to recapitulate adverse pregnancy outcomes to investigate causal relationships and molecular underpinnings of adverse gestational cardiac events and progression to the development of cardiovascular disease postpartum. This review will focus on summarizing clinical and animal studies detailing the impact of adverse pregnancy outcomes, including preeclampsia, gestational diabetes mellitus, and maternal obesity, on gestational cardiometabolic dysfunction and postpartum cardiovascular disease. Specifically, we will highlight the adverse impact of gestational hyperandrogenism and its potential to serve as a biomarker for maternal gestational and postpartum cardiovascular dysfunctions.