Case Management Time in SCD Adults Is Correlated with ED and Ambulatory Utilization and with Behavioral Health Diagnoses, but Not with Hospitalization and Costs

Background

Since 2018, The VCU sickle cell disease (SCD) Adult Medical Home case management program has assigned community health workers (CHWs) assigned as patient navigators (PNs) to manage its highest SCD utilizers, adjusting assignment each year based on prior-year costs and utilization patterns. We previously showed lower utilization among our enrolled SCD adults after participation in our SCD Adult Medical home[i], and we therefore advocated CHWs and/or PNs as an essential component of the SCD Adult Medical Home[ii]. However Medical Home managers have little guidance for assigning caseloads to SCD PNs. We showed that the weight of biological comorbidities more influences hospitalizations, ambulatory visits, and total costs, whereas the weight of behavioral comorbidities more influences ED visits[iii].

We therefore compared the impact of biological and behavioral comorbidities as well as healthcare utilization on the time of care spent by assigned PNs on care for SCD patients. We hypothesized that the behavioral comorbidities would more influence PN time than biological ones, and that the category of behavioral diagnoses would influence PN time. We also hypothesized that emergency department (ED) and hospital utilization rates might influence PN time more than ambulatory utilization.

Methods

PNs also partially manage some lower utilizers. PNs documented each patient interaction on case report forms, which included patient names and details about the contact, such as the method, reason for contact, services provided, and duration.

Additionally, we extracted ICD-10 codes for all medical diagnoses from patient electronic medical records and collected behavioral health ICD-10 mentions by patient each year. Behavioral diagnoses were categorized into clinically actionable intervention (CAI) groups commonly used by clinicians: mood disorders (e.g., prescribe or refer for mood stabilizers and/or talk therapy), psychotic disorders (e.g., referral to a psychiatrist for major antipsychotics), and substance use disorders (e.g., referral for substance use treatment). This analysis utilizes data from 2022 and 2023.

Partial correlation coefficient approaches were used to analyze the relationship between behavioral health and various measures of utilization. Point biserial correlation was used to quantify the impact of the presence of various behavioral health diagnoses on PN time as well as on ED and hospital utilization.

Results

During the years of these analyses, PN time was statistically significantly correlated with SCD patients’ ED visits, hospitalization, ambulatory costs, and total costs (all p < 0.05). None of the correlation coefficients met typically clinically significant levels (³0.3). Nonetheless, ED visits and ambulatory costs were more highly correlated (r= 0.2 and 0.14, respectively) than inpatient utilization and total costs (r=.07 and .06, respectively).

PN time was also statistically significantly correlated with the presence of behavioral diagnoses including mood disorder (r=.22, P<.001), psychotic disorder (r=.19, p<.001), and substance use (r=.14, p<.001). Mood disorder was statistically (p < 0.05) but not clinically significantly correlated with inpatient discharges (r=0.18), inpatient days(r=0.18) ED use (r=0.11), and outpatient use (r=0.10) as well as total costs (r=0.18). Similarly, psychotic disorder was statistically but not clinically significantly correlated with total costs (r=0.12), ambulatory use (r=0.11), and inpatient days (r=0.11), and substance use was statistically significantly (p<0.05) correlated with inpatient discharges(r=0.11), ED use (r=.05), and inpatient days (r=.08).

Conclusion

Patients who frequent EDs and clinics utilize more PN time than those who frequent hospitals or are high cost. Patients with behavioral health comorbidities utilize more PN time. Those assigning PNs based on patient utilization patterns should consider improving the assignment process by incorporating behavioral comorbidities into the assignment algorithm.