Latent Class Analysis of Maternal Comorbid Conditions among Hospitalized Pregnant Women Diagnosed with Opioid Use Disorders in North Carolina.

BACKGROUND

Pregnant women are among the groups most affected by the United States opioid epidemic.

OBJECTIVES

To determine latent classes of maternal comorbidities, examine their relationship to opioid use disorder (OUD), and how they can predict hospital discharge status among hospitalized pregnant women with and without OUD.

STUDY DESIGN

This is a cross-sectional study.

SETTING

Hospitals in North Carolina.

METHODS

A latent class analysis (LCA) was performed using 929,085 hospital discharge records from the 2000-2014 State Inpatient Databases for North Carolina. We defined OUD status and 24 maternal comorbid conditions based on International Classification of Diseases, Ninth Revision, Clinical Modification diagnostic codes and Clinical Classification Software codes, respectively. Discharge status was categorized as home, institution, or died. Binary and multinomial logistic regression models were constructed adjusting for demographic and hospital characteristics.

RESULTS

LCA of maternal comorbid conditions resulted in 591,745 records belonging to Class 1 (birth complications) and 337,340 records belonging to Class 2 (pre-existing and pregnancy-related morbidities). Class 2 records less frequently belonged to patients with OUD than those without OUD, and more frequently to younger, Black/Hispanic/other race or ethnicity, and patients with a higher socioeconomic status who resided in large metropolitan areas. Non-Medicare primary payers were more likely among Class 2 records. Irrespective of OUD status, patients belonging to Class 2 were less likely to be discharged to an institution or be deceased, controlling for confounders.

LIMITATIONS

Administrative database; data clustering; misclassification bias; confounding bias; temporality; data-driven approach; generalizability.

CONCLUSIONS

Hospitalized pregnant women may be classified based on comorbid conditions into 2 latent classes ("birth complications" and "pre-existing and pregnancy-related morbidities"), with the former exhibiting greater OUD frequency than the latter. These findings can inform health care needs of populations with a high-risk for OUD.