Calendar | For Your Patients | PHA Main Site | Contact Us | About Us | Not a registered user? Sign up here.

Resource Library

Intersecting Identities and Patient Outcomes in Pulmonary Arterial Hypertension: Early Results in Derivation of a Model Using the Pulmonary Hypertension Association Registry (PHAR)

Dan Grinnan

Le Kang

Chris DeWilde

D Johnson

Jeffrey Sager

David Badesch

Murali Chakinala

Teresa De Marco

Jeremy Feldman

Jimmy Ford


  Sign in to add a review

Leave a Comment

Release Date: 06.29.2018

Presentation Type: Abstracts

File Download: 1024 Abstract

Download Adobe Acrobat

Authors Continued

Klinger J8, McConnell J9, Berman Rosenzweig E10, Shlobin O11, Zamanian R12, Bartolome S13, Elwing J14, Franz R15 and 9 more

8Brown University, Riverside, RI

9Kentuckiana Pulmonary Associates, Louisville, KY

10Columbia University, New York, NY

11INOVA Fairfax, Fairfax, VA

12Stanford University, Stanford, CA

13UT Southwestern Medical Center, Dallas, TX

14University of Cincinnati, Cincinnati, OH

15Mayo Clinic, Rochester, MN



Existing guidelines for pulmonary arterial hypertension (PAH) yield vastly differing results between patients, even when meticulously implemented. This may be explained through intersectionality. Intersectionality refers to the overlapping of identities, experiences, and characteristics including race, ethnicity, gender, gender-identity, and class, which combine to make a person uniquely who they are and place the individual along a continuum of disadvantage to advantage. We believe intersecting identities may have an additive effect on health outcomes such that disadvantaged patients are at risk for poor outcomes. Using data from the Pulmonary Hypertension Association Registry (PHAR), we will derive a model (the PHAR Evaluation or PHARE) to assess the impact of intersecting identities on PAH patients for future validation.  As a preliminary to deriving this model, we provide data from three medically important outcomes to serve as a framework for assessing intersectionality in PAH.


PHAR data collected at initial and follow-up visits (every 6 months) from participating patients is used.  Independent variables include age, race, sex born, level of education, income, household number, marital status, drug use, and type of insurance.  The dependent variables include the number of Emergency Department (ED) visits, number of hospitalizations, and days of hospitalizations. Mortality was excluded in this initial dataset due to the short duration of follow-up and resultant events.  Analysis was performed using repeated measures Poisson regression models for the outcome count data. For inclusion in the regression model, a p-value of 0.2 was used.


Data from the initial 362 patients is being used in the Poisson regression analyses.  172 patients had data from one follow-up visit and 68 patients had data from two or more follow-up visits  (Table 1) . Lower income, lower education, and divorced or ‘never married’  were positively associated with more ED visits while having private health insurance was negatively associated with fewer ED visits. Lower income, lower education, drug use, and divorced or never married were positively associated with more hospitalizations while being uninsured or underinsured was negatively associated with fewer hospitalizations.


Early analyses of the PHAR data shows positive associations between income and education level on ED visits and a negative association between private insurance and ED visits.  That under/no insurance is associated with fewer hospitalizations may indicate these patients have difficulty being admitted to a hospital. As further longitudinal data from this cohort strengthens these associations, we plan to derive and validate a model that will explore their intersectionality.

Early Outcomes from the PHARE

Figure 1: Association of patient characteristics with number of ED visits, number of nights in the hospital, and number of hospitalizations. Level of education, household income, number in household, drug use, and age are continuous variables. All categorical variables have a reference value (eg. “white” for race and “female” for gender.