Predictors of High-Intensity Psychiatric Service Utilization by Veterans Health Administration Patients With Early Psychosis

Data from the VHA National Psychosis Registry (NPR) (11) and the Corporate Data Warehouse were used to assess the characteristics of VHA patients diagnosed as having EEP or LEP. The NPR is an administrative data set that contains care information related for all VHA patients diagnosed as having a serious mental illness (schizophrenia spectrum disorders, bipolar spectrum disorders, other psychotic disorders). The EEP group represented a subset of VHA NPR patients who met the following inclusion criteria: received a qualifying psychotic diagnosis (schizophrenia spectrum disorder, bipolar disorder with specified psychosis, other disorder specifying psychotic symptoms; ICD-10 codes F06.0, F06.2, F20.0–F20.3, F20.5, F20.81, F20.89, F20.9, F22–F25, F28, F29, F30.2, F31.2, F31.5, F31.64, and F53) during a 1-year period from July 1, 2018, through June 30, 2019; no previous qualifying psychosis diagnosis within VHA administrative medical records before July 1, 2015; and age ≤30 years as of July 1, 2018. A total of 4,595 VHA patients were identified for inclusion in the EEP group.

We developed this definition for the EEP group by using an iterative approach that involved the creation of multiple potential definitions based on different cutoffs for patient age and history of qualifying diagnoses. Each potential EEP definition was checked via a review of patient medical records by study staff to assess the accuracy of the EEP designation. The definition of EEP included in this article (and in formal VHA efforts to identify patients for EEP specialty services) was selected after the predictive validity of the identification algorithm reached 75% (149 of 199 patients flagged were confirmed as having EEP on the basis of review of patient charts) across six VHA care sites of differing complexity and at different geographic regions.

The LEP group comprised all individuals in the NPR who received qualifying psychosis diagnoses during the 1-year period of interest but did not meet the other criteria for the EEP group. This process identified 108,713 VHA patients who qualified for inclusion in the LEP group. Multiple measures of demographic and clinical characteristics as well as treatment participation were abstracted for all patients in the evaluation cohort. Because of the size of the cohorts being compared in these analyses, decisions related to group equivalence were based on statistical significance as well as a relative probability (RP) difference of ±0.2 between the groups across measures of interest. Similarly, in multivariate analyses, only odds ratios (ORs) that fell outside of the 0.8–1.20 range were considered meaningful. All analyses were adjusted to account for site of care receipt.

Data abstraction and analysis for this article were conducted from January through September 2020. This work was conducted as a part of national operational efforts to monitor the development and rollout of a new VHA clinical program; therefore, institutional review board appraisal was not required. Data management and analysis were conducted with SAS, version 9.2.

Patients with EEP were, on average, younger (mean±SD age 26.6±2.8 years) than patients with LEP (58.7±13.1 years, t=165.9, df=113,306, p<0.001), but both groups were similar in terms of sex, race, and level of service-connected disability. Patients with EEP were more likely to have a documented high-risk suicide flag (EEP, 13.9%, and LEP, 4.5%, χ²=833.6, df=1, p<0.001, RP=3.09), more likely to be flagged as homeless (EEP, 19.6%, and LEP, 14.1%, χ²= 108.5, df=1, p<0.001, RP=1.39), and more likely to receive a diagnosis of posttraumatic stress disorder (PTSD) (EEP, 41.3%, and LEP, 27.2%, χ²=437, df=1, p<0.001, RP=1.52) or a substance use disorder (EEP, 48.1%, and LEP, 32.1%, χ²=518.7, df=1, p<0.001, RP=1.50). Conversely, the EEP group had lower levels of medical morbidity (mean Elixhauser score of 2.7±1.5 for EEP and 4.1±2.5 for LEP, t=38.3, df=113,306, p<0.001). The two groups were equivalent in terms of the proportion with a documented high-risk flag for behavioral disruption.

Additional analyses were conducted to compare the treatment characteristics of patients with EEP and patients with LEP. The two groups were equivalent in terms of the proportion of patients who received prescriptions for an antipsychotic (EEP, 71.9%, and LEP, 71.5%) or mood stabilizer (EEP, 37.9%, and LEP, 40.3%); however, the EEP group was less likely to have received clozapine (EEP, 0.8%, and LEP, 2.5%, χ²=54.4, df=1, p<0.001, RP=0.32). The groups received equivalent outpatient mental health visits during the year before the evaluation period; however, the EEP group had fewer primary care visits (EEP, 2.7±2.6, and LEP, 4.0±4.3, t=14.7, df=113,306, p<0.001). The EEP group was more likely to have had an inpatient psychiatric admission during the assessment period (EEP, 32.0%, and LEP, 15.0%, χ²=964.7, df=1, p<0.001, RP=2.13) as well as an ED visit for the treatment of a mental health condition (EEP, 31.2%, and LEP, 16.3%; χ²=696.6, df=1, p<0.001, RP=1.91). These results are summarized in Table 1.

We then conducted analyses using the entire evaluation cohort (patients with EEP and LEP) to examine the impact of EEP status on the risk for using high-intensity mental health care services while controlling for other patient characteristics. Within a logistic regression model, increased risk for an inpatient psychiatric admission was associated with female gender (OR=1.29, 95% confidence interval [CI]=1.20–1.37); the presence of a behavioral risk flag (OR=3.12, 95% CI=2.75–3.55) or suicide risk flag (OR=4.67, 95% CI=4.25–5.14); homelessness (OR=3.24, 95% CI= 3.02–3.48); a diagnosis of PTSD (OR=1.39, 95% CI=1.31–1.47) or substance use disorder (OR=3.17, 95% CI=2.95–3.40); increased medical morbidity (OR=2.47, 95% CI=2.35–2.59); and receipt of a prescription for an antipsychotic medication (OR=1.76, 95% CI=1.66–1.87), a mood stabilizer (OR=1.54, 95% CI=1.48–1.61), or clozapine (OR=1.46, 95% CI= 1.27–1.67). Black race was associated with a reduced probability of inpatient psychiatric admission (OR=0.78, 95% CI= 0.74–0.82). After adjusting for these factors, we found that EEP status was significantly associated with risk for inpatient psychiatric admission, with those with EEP being significantly more likely to have an inpatient psychiatric admission than those with LEP (OR=1.47, 95% CI= 1.31–1.65). All findings were statistically significant at the p<0.001 level. These results are summarized in Table 2.

In another logistic regression, we focused on risk for an ED visit for the treatment of a mental health condition while controlling for other patient characteristics. Increased risk for mental health–related ED use was associated with female gender (OR=1.25, 95% CI= 1.17–1.34); greater number of prior-year outpatient mental health visits (OR=1.00, 95% CI=1.001–1.003); the presence of a behavioral risk (OR=2.87, 95% CI=2.61–3.14) or suicide risk (OR=2.23, 95% CI=2.01–2.48) flag; homelessness (OR=2.95, 95% CI=2.79–3.12); a diagnosis of PTSD (OR=1.16, 95% CI=1.10–1.23) or substance use disorder (OR=2.56, 95% CI=2.43–2.70); elevated medical morbidity (OR=2.11, 95% CI=2.02–2.21); and the receipt of a prescription for an antipsychotic medication (OR=1.21, 95% CI=1.15–1.27), clozapine (OR=1.32, 95% CI=1.15–1.32), or a mood stabilizer (OR=1.41, 95% CI=1.36–1.46). After adjusting the model for these factors, we found that patients with EEP were significantly more likely to have an ED visit for the management of a mental health condition than were those with LEP (OR=1.45, 95% CI=1.30–1.62). All findings were significant at the p<0.001 level and are summarized in Table 3.

The early phases of psychotic disorders are associated with increased risk for a variety of negative outcomes. An understanding of the unique characteristics and treatment needs of VHA patients with EEP can help inform care for this group via programs such as VHA EPIC. Patients with EEP or LEP received equivalent levels of outpatient mental health care and psychopharmacology; however, patients with EEP had more than double the rate of use of inpatient psychiatric care and nearly double the rate of use of EDs for the management of mental health conditions. Although the EEP group had greater psychiatric morbidity relative to the LEP group, including elevated rates of suicide risk, homelessness, and diagnosed PTSD and substance use disorders, these characteristics did not explain the association between EEP status and increased use of high-intensity mental health services. These results suggest that unique aspects of the EEP condition may contribute to the need for high-intensity mental health services to manage symptoms within an outpatient psychosis treatment-as-usual care setting. Furthermore, our findings suggest that standard outpatient mental health services, which may be adequate to address the mental health care needs of patients with LEP, may not adequately address the specific mental health care needs of patients with EEP.

Several factors were found to be associated with increased utilization of high-intensity mental health services by people with psychotic disorders, independently of EEP status. Increased risk for the use of inpatient and ED psychiatric care was associated with diagnoses of PTSD and substance use disorders as well as an elevated risk for suicide and disruptive behavior. These results are consistent with those of previous studies that found increased health care costs and utilization of inpatient and emergency psychiatric services among individuals with schizophrenia and comorbid substance use disorders compared with those with schizophrenia and no substance use disorder (12). Comorbid PTSD among individuals with schizophrenia has also been associated with higher rates of psychiatric hospitalization and outpatient services use (13–15), more severe symptom levels (16), and more health problems (15). Similarly, in one retrospective study, individuals with schizophrenia and a history of suicide attempts had a higher total number of hospitalizations than those without suicide attempts (17).

Of note, we found that Black race was associated with reduced likelihood of admission to inpatient psychiatric care in our study, a finding that contradicts previous research that had focused on the relationship between race and psychiatric inpatient care for patients with serious mental illness (18). The lack of agreement between our findings and previous work would benefit from additional exploration to understand whether this pattern may be unique to the U.S. Department of Veterans Affairs system of care or due to other important treatment factors.

Our findings suggest condition-specific areas of vulnerability that may influence a reliance on high-intensity care for the EEP group. Because of specific vulnerabilities, patients with EEP may be “invisible contributors” to chronic use of high-intensity mental health services; therefore, the provision of EEP-focused care could help increase health care efficiency by reducing the number of patients chronically using higher levels of psychiatric services or patients with serious mental illness that is treatment resistant. This idea is consistent with research that found that EEP-targeted interventions can be highly cost-efficient (8).

Of note, interventions designed to provide care for people with EEP have been specifically shown to effectively reduce the need for high-intensity mental health services (7), suggesting a potential path forward to address the excess use of high-intensity care services by patients with EEP within the VHA care system. Although systematic implementation of programs with complexities similar to that of CSC could be a significant undertaking, the VHA system has many characteristics that suggest it could be an ideal setting for such efforts, such as diverse outpatient psychosocial rehabilitation services, vocational support, and an integrated medical record to support care coordination.

Our study had several limitations. It was observational; therefore, no causal inferences should be drawn between patient EEP status and the use of high-intensity psychiatric services. Similarly, our focus on administrative data did not allow us to assess the treatment needs or levels of functioning of patients, characteristics that were considered important in previous work focusing on care for people with EEP and which are often the targets of EEP care programs. Our use of administrative data, dichotomous coding of some measures, and an analytic approach that focused on the use of multiple patient characteristics within the same prediction models may have increased the likelihood of false discoveries. Although we took steps to improve the accuracy of our EEP assignment algorithm, some patients were likely misidentified as having EEP because of the limitations of our approach of using administrative data; moreover, our identification process was limited to treatment information available within the VHA system, and some patients likely experienced psychotic symptoms, diagnosis, and treatment before their intake into VHA.

The diagnoses included in our EEP cohort creation included conditions with affective psychosis, such as bipolar disorder with psychotic features, and affective psychosis is often excluded from CSC EEP research. The time line to transitioning from early psychosis to later psychosis likely occurred on a continuum across patients, which may have resulted in misclassification of some patients in our approach, which focused on the 4 years after receipt of a first psychotic diagnosis. We also note potential issues with an overlap between some predictors and high-intensity care settings such as suicide risk flags and medication initiation, which may have been disproportionately assigned within ED and inpatient care settings. We also did not evaluate whether patients with EEP and patients with LEP received specific interventions consistent with CSC or the potential impact of these services on our outcomes of interest. These areas are important for future work to better understand the unique care needs of patients with EEP and the extent to which these needs are met by currently available outpatient treatment-as-usual services within VHA.

This work represents the first evaluation of potential unmet treatment needs among patients with EEP within a national coordinated health care system serving thousands of patients with psychotic disorders. Patients with EEP had higher levels of psychiatric morbidity and use of high-intensity mental health services than did patients with LEP. Because these differences persisted after analyses controlled for group characteristics, our results suggest that EEP presents a unique risk for high-intensity mental health service use; moreover, currently available outpatient mental health services may not be adequate to fully address the needs of patients with EEP. Future work could clarify which aspects of outpatient psychiatric services most effectively prevent high utilization of inpatient or high-intensity care services. Effective treatment has the potential to reduce the risk for psychiatric morbidity and mortality among individuals with EEP.