Classifying Veterans' pre-pregnancy health risks using latent class analysis.

BACKGROUND

Veterans using Department of Veterans Affairs (VA) healthcare have a high burden of pre-pregnancy chronic disease that likely contributes to the observed high rate of pregnancy-related morbidity. Many common diseases frequently co-occur; understanding patterns of multimorbidity may inform the design and delivery of pre-pregnancy interventions to lower pregnancy morbidity risk.

OBJECTIVE

The current study sought to identify patterns of co-occurrence of pre-pregnancy chronic disease among Veterans.

DESIGN

We conducted a retrospective cohort study using VA administrative data.

METHODS

Our population included Veterans ages 18-45 with ⩾1 pregnancy outcome (ectopic, spontaneous abortion, stillbirth, and/or live birth) during fiscal years 2010-2019. Presence of common chronic diseases with implications for pregnancy was detected using encounter International Classification of Diseases, 9th and 10th Revision (ICD-9 and ICD-10) codes in the 2 years prior to pregnancy. Patients were grouped based on latent class models of diagnosis patterns; two to seven latent groups were examined for model fit and clinical interpretability.

RESULTS

We identified 56,853 pregnancies from 41,034 Veterans. More than half of pregnancies were complicated by an array of pre-pregnancy medical and mental health conditions that may negatively impact pregnancy health and contribute to adverse pregnancy outcomes. The most frequently occurring conditions included chronic pain (51.2% of pregnancies), depression (31.4%), anxiety (25.9%), and post-traumatic stress disorder (22.8%). A five-group model demonstrated the best balance between model fit and clinical interpretability. Groups included: "Pain and Mental Health" (28%), with high prevalence of chronic pain, depression, and anxiety; "Pain and Metabolic" (17%), high prevalence of chronic pain, obesity, and migraines; "Substance Use and Mental Health" (7%), high prevalence of alcohol use disorder, depression, and post-traumatic stress disorder; "Low Diagnosis" (43%), lower than average prevalence of diagnoses; and "High Complexity" (5%), high prevalence of conditions across multiple physiologic systems.

CONCLUSIONS

We identified five distinct, clinically meaningful groups of Veterans based on co-occurring pre-pregnancy diseases. Tailoring interventions to these groups may address Veterans' complex pre-pregnancy health risks effectively and efficiently.