Embedded Psychological Services and Lower 60-Day Acute-Care Utilization in High-Need, High-Cost Adults: A Target Trial Emulation

Abstract In safety-net care management, high-need, high-cost (HNHC) adults often receive bundled services. The independent contribution of embedded psychological-services engagement is unclear. This retrospective cohort study used target trial emulation under the ICH E9(R1) estimand framework to evaluate an opt-out embedded psychological-services pathway nested within a manualized social-determinants-of-health–focused care-management program at four Phoenix Federally Qualified Health Centers. Eligible HNHC adults had ≥ 1 inpatient admission or ≥ 2 emergency department visits in the prior 12 months. Each received a scripted opt-out engagement protocol with structured warm-handoff referral where operationally available. Accepted/engaged status was classified as attendance at ≥ 1 behavioral health session within 60 days; the comparator was declined/no-show after the standardized offer. The 60-day composite outcome combined hospital admissions and emergency department visits ascertained from claims linked to a shared electronic health record. Stabilized inverse-probability-of-engagement weighting addressed measured confounders. Sensitivity analyses spanned engagement timing or classification, missing data, unmeasured confounding, propensity-score specification, and clinic-level dependence. Among 1038 adults, 485 (46.7%) were classified as accepted/engaged. Accepted/engaged status was associated with a 33% lower 60-day composite acute-care rate (adjusted incidence-rate ratio [aIRR] 0.665; 95% CI, 0.585–0.757; P < .001). The lower composite rate was driven by ED visits (aIRR 0.619; 95% CI, 0.536–0.715; P < .001). The 60-day admissions component alone was not statistically significant (aIRR 0.891; 95% CI, 0.669–1.188; P = .43). All five Holm–Bonferroni-corrected endpoints survived correction: the 60-day composite outcome, 60-day EQ-5D-5L, 30-day WHO-5 wellbeing, 30-day PHQ-9, and 30-day GAD-7. The E -value (point 2.37; CI-bound 1.97) implies that an unmeasured confounder would need strong association with both engagement and outcome to fully explain the result. Observed accepted/engaged status with embedded psychological services was associated with lower short-term acute-care utilization in safety-net HNHC adults already enrolled in multicomponent care management.