EHR-based screenings under-identify housing instability, financial strain

Electronic health record-based questionnaires were less accurate in identifying patients with housing and financial struggles compared with single-domain questionnaires, a research letter found.

“Public health, care delivery organizations, federal agencies, and researchers have advocated for better collection of patient social risk factor information,” Christopher A. Harle, PhD, a professor in the department of health outcomes and biomedical informatics at the University of Florida, and colleagues wrote in JAMA. “Social factors can determine referrals to community partners, increase awareness of patients’ needs, measure population health, or improve risk prediction models.”

While electronic health record systems have multidomain questionnaires that can evaluate such socioeconomic risks, “their accuracy has not been established,” the researchers wrote.

Harle and colleagues compared the accuracy of EHR-based questionnaires with a single-domain questionnaire that assessed the prevalence of three social risk factors: food insecurity, housing instability and financial strain.

The questionnaires were completed concurrently across 11 primary care clinics by 826 patients with a mean age of 49 years. Among them, 63.9% were women, 39.2% were non-Hispanic Black individuals and 44.8% were non-Hispanic white individuals.

Overall, the prevalence of positive screens for housing instability from the EHR-based questionnaire (27.1%; 95% CI, 24-30.1) was significantly lower compared with the prevalence reported by the single-domain questionnaire (38.4%; 95% CI, 35.1-41.8). The EHR-based questionnaire also demonstrated a lower prevalence of financial strain (10.4%; 95% CI, 8.3-12.5 vs. 33.2%; 95% CI, 29.9-36.4).

Harle and colleagues found that the EHR screening demonstrated high specificity (93%; 95% CI, 90.4-95.1) but low sensitivity (59%; 95% CI, 53.4-64.5) and a low area under the curve (0.76; 95% CI, 0.73-0.79) for housing instability. Similarly, for financial strain, it demonstrated high specificity (98.1%; 95% CI, 96.6-99.1), low sensitivity (27.1%; 95% CI, 21.8-32.8) and a low area under the curve (0.626; 95% CI, 0.6-0.65).

“This suggests that the primary goal of screening may not be achieved for these two risks,” the researchers wrote.

In contrast, neither of the questionnaires showed a significant difference in food insecurity prevalence, according to the researchers. They reported that the EHR-based questionnaire was accurate, with a high area under the curve (0.938; 95% CI, 0.92-0.96), high specificity (93.1%; 95% CI, 90.6-95.2) and high sensitivity (94.5%; 95% CI, 91.2-96.8) for the risk factor.

Harle and colleagues noted that each risk factor yielded a positive predictive value greater than 80%, “meaning that a positive EHR-based screen would reflect the presence of the respective social factor in more than eight of 10 patients.”

The study limitations included results potentially not being generalizable to other multidomain screening questionnaires, social factors or care settings. Still, “many other screening questionnaires use the same items as the questionnaire in this study,” the researchers wrote.

“Assessing performance is the first step in understanding the utility and implications of screening results,” they concluded. “Further research is needed on which and how such screening tools should be used.”