Pediatric Asthma Risk Score Robust Across Diverse Backgrounds

The Pediatric Asthma Risk Score (PARS) remained an effective tool for estimating asthma risk when used for patients of varying backgrounds, a study of 10 cohorts found.

The area under the receiver operating curve (AUC) for the PARS assessment was 0.76 (95% CI 0.74-0.78), with AUC scores ranging from 0.61-0.83 across the cohorts, reported Gurjit K. Khurana-Hershey, MD, PhD, of the Cincinnati Children’s Hospital Medical Center, and co-authors.

Following analysis for sex, ethnicity, missing factors and more, PARS performed similarly across the cohorts (P>0.05 for all), according to the study published in NEJM Evidence.

PARS yielded similar performance between Hispanic and non-Hispanic ethnicities (0.77 vs 0.76, P=1.0), male and female patients (0.75 vs 0.73, P=0.73), and among high-risk and general populations of participants (0.69 vs 0.76, P=0.42).

PARS was developed for asthma prediction in “relatively homogenous populations” but has proven to be a robust tool, the researchers noted.

“Although PARS was developed in a single cohort, the meta-analysis supports that all six PARS factors contribute to predicting asthma risk. When performing multivariable regression in each cohort, the factors that predicted asthma varied. For almost all factors, the direction of effect was consistent, but the effect sizes varied across the individual cohorts,” they wrote.

The model is made up of six factors including wheezing without a cold, early wheeze, eczema, polysensitization, parental history of asthma, and whether a patient self-identifies as Black/African-American race.

The researchers noted that the initial reasoning behind including Black/African-American race as a factor in PARS was likely an attempt to address social determinants of health and the impacts of structural racism. However, they wrote, that is not a perfect method and may have negative effects.

“Notably, a prior CREW study showed that higher asthma risk in Black and Hispanic children remained even after controlling for neighborhood-level socioeconomic indicators defined by U.S. census tracts, indicating that race is not simply a proxy for socioeconomic factors,” they wrote.

They continued: “In this analysis, we cannot capture the overlapping determinants of health encompassed by race, which may include structural, institutional, cultural, socioeconomic, heritable, environmental, and social factors. Despite our inability to capture all determinants of health associated with race, it is retained in the PARS model as a predictor of asthma risk, but we acknowledge that utilizing race in clinical care may unintentionally perpetuate structural racism.”

The Asthma Predictive Index (API), another tool used to predict asthma in children, had a lower AUC of 0.70 (95% CI 0.68-0.72) across all cohorts. When utilizing the more strict definition, the API had an AUC of 0.70 (95% CI 0.67-0.73). While the API was still considered to be robust at predicting asthma, PARS consistently performed better.

Another advantage of PARS in asthma prediction “is that it can identify asthma across mild, moderate, and high risk compared with a dichotomous outcome, such as API,” the researchers noted.

Researchers also utilized more strict definitions for PARS, which featured lung function measurements and only utilized data from four of the 10 cohorts included in the study. When using the more strict definitions, the AUC for PARS was 0.78 (95% CI 0.75-0.81).

A total of 5,634 patients from 10 cohorts were included in the study. The proportion of Black patients varied across the cohorts, ranging from 1.6-72.2%, as did Hispanic ethnicity, ranging from 1.5-61.3%. Patients were categorized by race as either Black or non-Black (including Asian, American Indian, Native Hawaiian or other Pacific Islander, Alaska Native, or white). Approximately half the patients were male across all cohorts.

Asthma prevalence for all participants was 12.8%, eczema prevalence ranged from 12.1-56.7%, parental asthma was present in 13.5-66.3%, and wheezing without a cold was seen in 9.9-38.4% of patients.

The data used to test PARS was taken from the Children’s Respiratory and Environmental Workgroup (CREW), a nationwide set of 13 birth cohorts used for asthma development study, although three cohorts had to be excluded as the participants were too young to receive an asthma assessment.