Abstract :
Background: I examined patient demographic, clinical, payment, and geographic factors associated with admission to low-safety inpatient psychiatric facilities.
Methods: Massachusetts all-payer 2017 discharge data (N = 39,128) were linked to facility-level indicators of safety (N = 38). A composite of safety performance was constructed by averaging standardized measures of restraint and seclusion, as well as five-year (2014–2018) averages of overall, substantiated, and abuse-related (verbal, physical, sexual abuse) complaints per 1,000 discharges (• = 0.73). This composite informed the grouping of high (top 20%), middle, and low-safety (bottom 20%) performers. I first examined unadjusted differences across safety groups, as well as differences in bypass patterns across racial and ethnic groups. I then fit a series of multinomial regression models, adding payment and geography separately.
Results: Outstanding factors independently associated with admission to low-safety facilities were being a racial or ethnic minority compared to White patient (relative risk ratio [RRR] for non-Hispanic Black = 1.7, 95% CI = 1.5–2.0; non-Hispanic Asian = 5.6, 95% CI = 3.6–8.7; non-Hispanic “other” race = 2.2, 95% CI = 1.7–2.7; Hispanic/Latinx = 1.3, CI = 1.1–1.5), and not having private insurance (RRR for uninsured/self-pay = 2.4, CI = 1.6–3.6, Medicaid = 1.8, CI = 1.6–2.0, Medicare = 1.3, CI = 1.2–1.5). Several other factors were independently associated with admission to low-safety facilities, such as substance use disorder other than alcohol, proximity, severity, schizophrenia/psychosis, homelessness, and younger age.
Conclusion: There were considerable racial and ethnic inequities in admission to low-safety inpatient psychiatric facilities even after accounting for clinical, geographic, and payment characteristics. Future research should further examine quality variation and outcomes, as well as how community-based referrals, mode of transport (e.g., police, self), and deliberate steering and selection affect admissions and outcomes.
The original paper has been updated.
