Introduction

The prevalence of mental illness has increased in the U.S. since the early 1990s [1]. In 2016, over 55 million adults were living with a mental illness or substance use disorder [2]. Emergency departments (EDs) are increasingly bearing the brunt of these patients, with the rate of ED visits related to mental illness or substance use disorder increasing by 44.1% between 2006 and 2014 [3]. With many EDs lacking in the ability to efficiently care for psychiatric patients [4], these visits are more likely than non-psychiatric visits to result in inpatient admission [2]. Yet, even with increased need for resources, the number of inpatient psychiatric beds has not kept up with demand, leaving psychiatric patients spending hours in the ED waiting for transfer to an open bed and creating crowding in the ED [5]. Previous efforts to solve ED crowding has focused on diverting psychiatric patients out of the ED. In recent years, many have argued instead for the creation of an effective environment within the ED for the emergency management and treatment of psychiatric patients, with an increasing interest in the ability of telepsychiatry to alleviate the rising demand [4].

Telepsychiatry, or the receipt of telehealth services for the evaluation of ED patients who require mental or behavioral health care [6], is one of the fastest growing segments of telehealth [7]. In 2016, 20% of US EDs received telepsychiatry services [6]. The use of telepsychiatry in EDs allows remote access to psychiatric providers with expertise in emergency psychiatry, helps improve patient care, facilitate patient discharge, and prevent ED crowding [4]. Preliminary studies suggest that, in comparison to traditional face-to-face delivery of mental health services, telepsychiatry is comparable in both effectiveness and patient satisfaction [8, 9]. Patients who did not receive ED telepsychiatry services appeared to spend more time waiting for treatment or an inpatient bed, were more likely to have their visit result in inpatient admission, and were less likely to attend outpatient follow-ups [10,11,12]. Despite the promise of telepsychiatry, much of the current research on ED telepsychiatry use focuses either on specific patient populations based on age [9, 12] or only includes hospital-based EDs enrolled into a specific telepsychiatry intervention program [10, 11]. We sought to investigate the role of telepsychiatry across a wider and more representative range of EDs. Thus, we examined all ED visits across New York State to investigate the difference between psychiatric visits in EDs that did not receive telepsychiatry services versus visits in EDs that did.

Because disposition of patients with psychiatric complaints in the ED continues to be an area of national focus, we focused our investigation on the use of observation services during the ED visit [13,14,15] and the final ED visit disposition [13, 14]. While many studies have observed the effect of telepsychiatry on ED length of stay and admission/transfer outcomes in psychiatric patients [10,11,12], we were also interested in the observation services variable. Observation services are used when patients require a period of treatment or monitoring before an admission or discharge decision can be made, with most cases of observation stays lasting between 8 to 48 hours [15]. While under observation, a psychiatric patient would not only still be taking up space in the ED, but also be subjected to chaotic ED environments, which have been shown to worsen symptoms in psychiatric patients [4]. Thus we hoped to explore the ability of telepsychiatry to reduce need for observation services and help facilitate more efficient discharges. We hypothesized that relative to visits in EDs with telepsychiatry services, visits in EDs that did not receive telepsychiatry services would be more likely to use observation services and have higher rates of admission.

Methods

Study Design

This observational study was based on data gathered from an institutional-level survey of U.S. EDs, which was then linked to a state database with patient outcomes. New York was chosen due to the fact that the state had rural and urban EDs as well as teaching hospitals and Critical Access Hospitals (CAH). The study was approved by the Human Research Committee of Partners Healthcare.

National Emergency Department Inventory

Using the National Emergency Department Inventory (NEDI)-USA database [16], we identified 194 New York EDs open in 2016. EDs were included in NEDI-USA if they were open 24 hours per day, 7 days per week, year-round and were available for use by the general public. This included freestanding EDs (EDs that are not physically connected to a hospital) [17]. We mailed a one-page survey to all ED directors up to three times and then contacted nonresponding EDs to complete the survey through telephone interview. The 2016 NEDI-USA survey asked ED directors to report basic characteristics of their ED, including annual visit volumes and telemedicine use (see Online Supplement).

Exposure Measurement

Receipt of ED telepsychiatry was assessed with the question: “Does your ED receive telemedicine for patient evaluation?” [18]. Those who responded yes were then asked to report the clinical applications for which they used telemedicine. EDs whose directors reported use of telemedicine for psychiatry or wrote-in that they used telemedicine for “behavioral health” or “mental health” were classified as receiving telepsychiatry. We also collected: rural location, defined as ED location outside of a core-based statistical area (CBSA) [19]; CAH [20] designation; and Council of Teaching Hospitals (COTH) [21] designation.

Outcome Variables

EDs identified in the 2016 NEDI-USA database were linked to the 2016 American Hospital Association (AHA) database [22] using methods previously described [23]. Using AHA ID numbers, the NEDI-USA EDs were then linked to the New York 2016 State Emergency Department Databases (SEDD) and New York 2016 State Inpatient Database (SID). Visits in EDs that did not respond to the 2016 NEDI-USA survey or to EDs unable to be linked to the 2016 AHA database were excluded. We first compiled an overview of psychiatric services available to New York hospitals in 2016, comparing hospitals with EDs that did not report receipt of telepsychiatry services versus hospitals with EDs that did. Through the use of SEDD/SID ICD-10-CM diagnosis codes [24], we then identified visits that received a psychiatric-related primary, secondary, or tertiary diagnosis in SEDD and psychiatric visits with subsequent admissions by admitting diagnosis in SID. We obtained patient characteristics, including age, sex, race/ethnicity, insurance status, and median household income (based on patient’s ZIP code) [13, 14]. We then identified the use of observation services during the visit and the final disposition of the ED visit. The visit dispositions of interest were transfer to a short-term hospital, transfer to another type of facility (including psychiatric hospitals), or admission through the ED.

Statistical Analysis

Descriptive statistics are presented as both frequencies and percentages. We used chi-square test to evaluate bivariate associations between an ED’s telepsychiatry service status and other characteristics at visit and hospital levels. A two-sided P<0.05 was considered statistically significant. We then used both unadjusted and multivariable logistic regression models to examine the associations between telepsychiatry status and two outcomes: use of observation services and final ED visit disposition, with a random intercept for hospital to account for variation in dependent variables across hospitals. In the disposition model, outcome was defined as visits admitted through ED or transferred to a short-term hospital or other type of facility. Non-case was defined as visits with routine discharge. Other types of facility include a skilled nursing facility, intermediate care facility (ICF), and a psychiatric hospital or psychiatric unit. All multivariable models were adjusted for several visit and hospital characteristics. In sensitivity analyses, we re-ran our multivariable models in subpopulations restricted to adults only and patients with a primary diagnosis in SEDD. Patient data from SID had to be excluded from the primary diagnosis population, as it was not possible to tell whether the psychiatric diagnosis in the ED was primary or secondary. Results are reported as odds ratios (ORs) with 95% confidence intervals (95% CIs). All analyses were performed with SAS software, version 9.4 (SAS; Cary, NC).

Results

In 2016, there were 194 EDs open in New York State. The data from these EDs were consolidated into 160 observations [23] (each observation representing between 1 and 3 New York EDs) linked across the NEDI-USA, AHA, and SEDD/SID databases, with 18 (11%) of those being EDs that received telepsychiatry. After excluding visits in EDs that did not respond to the 2016 NEDI-USA survey, visits in the remaining 133 EDs then formed our analytic sample (Fig. 1).

Fig. 1
figure 1

Inclusion flow diagram. Abbreviations: NEDI, National Emergency Department Inventory; ED, Emergency Department; AHA, American Hospital Association; SEDD, State Emergency Department Databases; SID, State Inpatient Database

Full size image

Psychiatric Services in New York State

The EDs within our analytic sample were first examined for hospital psychiatric services based on data from the AHA database (Table 1). Eighteen hospitals reported having no psychiatric services. Of the 18 EDs that reported receiving telepsychiatry services in 2016, 8 (44%) also had psychiatric beds within their hospital, 13 (72%) had psychiatric emergency services, and 8 (44%) had psychiatric outpatient services. Only 1 was a CAH, with 3 others identified as teaching hospitals. EDs without telepsychiatry saw a median of 37,000 annual total ED visits, with an interquartile range of 16,000 – 75,000 visits. ED’s with telepsychiatry services saw a median of 50,000 total ED visits per year, with an interquartile range of 34,500-87,500 visits.

Table 1 Facility characteristics of New York State Emergency Departments without and with telepsychiatry services, 2016
Full size table

Psychiatric ED Visits and Telepsychiatry

Across all EDs consolidated into our analytic sample, there were 7,821,081 total ED visits. Among those visits, 957,654 (12%) received a psychiatric diagnosis. Among these visits, 712,236 (74%) were in an ED without telepsychiatry and 101,025 (11%) were in an ED with telepsychiatry; 144,393 (15%) visits were excluded due to their unknown telepsychiatry status. The receipt of telepsychiatry services was associated with several patient and hospital characteristics (Table 2). In bivariate comparisons, visits in EDs with telepsychiatry services were more likely to occur in an urban ED (99.8% vs 98%), EDs with >40,000 annual visits (83% vs 77%) , and among patients with public (63% vs 62%) or private insurance (24% vs 21%).

Table 2 Visit characteristics among psychiatric visits in Emergency Department without and with telepsychiatry services, 2016
Full size table

Telepsychiatry Receipt Status and Visit Disposition

Observation services were used in 2.3% of visits presenting to EDs with telepsychiatry, versus 2.7% of patients presenting to EDs without. We found that visits in EDs without telepsychiatry were less likely to result in admission or transfer than visits in EDs with telepsychiatry. When comparing psychiatric visits in EDs without telepsychiatry versus those with telepsychiatry, 12% versus 14% resulted in admission or transfer to a short-term hospital, 83% versus 78% resulted in routine discharge, and 2% versus 5% resulted in transfer to another type of facility.

In multivariable analyses (Table 3), psychiatric visits in EDs with telepsychiatry were three times less likely to use observation services (P = 0.03), after adjusting for patient demographic characteristics and hospital characteristics. In a separate multivariable analysis for final ED visit disposition (Table 4), psychiatric visits in EDs without telepsychiatry had similar odds of admission through the ED, transfer to a short-term hospital, or transfer to another type of facility, compared to psychiatric visits in EDs with telepsychiatry. In a sensitivity analysis restricting our analytical population to those with documentation of a primary psychiatric diagnosis in SEDD, visits in EDs with telepsychiatry remained less likely to have used observation services (adjusted OR 0.35, 95%CI 0.12-0.99, P = 0.048), and ED telepsychiatry status remained unassociated with ED disposition. In another sensitivity analysis restricting our analytical population to adults only, observation service also occurred less often in visits in EDs with telepsychiatry service (adjusted OR 0.30, 95% CI 0.10-0.85, P = 0.02). Adult patients had similar ED dispositions regardless of ED telepsychiatry status.

Table 3 Logistic regression of using observation services by Emergency Department telepsychiatry service status, 2016
Full size table
Table 4 Logistic regression of transfer or admission through ED by Emergency Department telepsychiatry service status, 2016
Full size table

Discussion

Through linkage of multiple 2016 datasets, we were able to compare visit- and hospital-level characteristics across 813,261 visits in New York EDs, with an analytic focus on the ED’s receipt of telepsychiatry services. We examined the association of telepsychiatry in the ED with two specific outcomes of an ED visit; the use of observation services and final ED visit disposition. Psychiatric visits in EDs without telepsychiatry, compared to those with, were three times more likely to have used observation services, but still had similar odds of being admitted or transferred as their final ED disposition.

To our knowledge, this study is the first to investigate the role of telepsychiatry across an entire state’s EDs, with a focus on patient disposition. Prior studies on the potential of telepsychiatry for use within ED settings were conducted within very specific settings, with some only focusing on pediatric populations or only among certain hospital systems [9,10,11,12]. While one prior study investigated the impact of telepsychiatry across South Carolina, only visits in hospitals enrolled in the state telepsychiatry program were eligible to be matched to visits in hospitals without telepsychiatry, limiting the generalizability to hospitals not receiving telepsychiatry through this dedicated state program [10]. By examining all EDs across New York State, we found that visits in EDs with telepsychiatry had lower odds of observation unit use while maintaining similar admission and transfer outcomes as visits in EDs without. Though we conducted a separate analysis excluding children from the population, we did not find that it materially affected results. This finding is consistent with previous research, in which, telepsychiatry has been suggested as a potential solution for patients in EDs without in-house psychiatric services, with evidence suggesting that telepsychiatry is comparable on several measures [8,9,10,11,12].

With many ED’s lacking the ability to care for psychiatric patients, these patients are often left in the ED waiting for transfer to an inpatient bed, exacerbating the problem of crowding in EDs [4]. Research suggests that the use of specialized observation services can reduce inpatient admission rates for psychiatric patients. However, it must be noted that this requires investment into the specialized resources that would allow psychiatric patients under observation to receive efficient and safe care [25], something most EDs lack [4]. Even with an efficient environment, the use of observation services, by definition, increases the amount of time spent in the ED [15] and does not resolve the problem of crowding in the ED, especially when other ED patients may require observation. In our findings, visits in EDs with telepsychiatry were less likely to use observation services while maintaining similar disposition outcomes. Though a low percentage of all psychiatric ED visits used observation services (Table 2), it is possible that what is considered use of observation services in one ED may be considered part of ED length of stay in another ED with no official observation unit [15]. Our findings still lend support to telepsychiatry’s ability to reduce overall time spent in EDs [11], whether the visit involves official use of observation services or non-classified time spent waiting for treatment.

With the increase in psychiatric-related visits in the ED in recent years, rural EDs have struggled to accommodate these patients as an increase in urban migration has led to decreased access to specialists and consultants in rural areas [26]. In New York, 40 out of its 62 counties were designated as Mental Health Shortage areas in 2014 [27]. Telepsychiatry has been touted before as a solution for rural EDs to access necessary mental health resources [28]. Despite this, we found that among our analytic sample, 98% of visits were in EDs in urban hospitals, with 99.8% of visits in EDs with telepsychiatry services occurring in urban hospitals. Only one hospital with telepsychiatry services reported having a CAH designation. These findings differ from results of a nationally representative sample, where EDs in rural areas or carrying CAH designations made up a higher proportion of EDs that received telepsychiatry [6]. It is unclear why a state with a documented need for mental health resources in its rural counties [29] sees the highest rate of telepsychiatry use in its urban hospitals or in hospitals with in-house psychiatric services. However, based on recent research on barriers to use of telemedicine by rural hospitals [26], we can hypothesize that cost and technologic concerns have played a role in the lack of telepsychiatry services in rural EDs despite research supporting its use.

The financial sustainability of telepsychiatry programs is not a new concern, as many programs are grant-funded and risk dissolution upon the disappearance of the grant [30]. In regards to technology, hospitals often use different electronic medical records systems, making it difficult to transfer information gathered by the ED provider to the off-site clinician [31]. To address these concerns, we encourage more research into alternative telepsychiatry delivery methods that help to reduce cost and technological difficulties. For example, a recent study examined the effect of psychiatric assessment officers in rural New York EDs [32]. A group of officers, including licensed social workers, mental health counselors, and psychiatric nurses, were integrated into EDs for 8 hours daily to help perform psychiatric evaluations and coordinate further care via telepsychiatry. This method cut costs by reducing the amount of time a licensed psychiatrist was needed and making use of lower-cost behavioral health providers.

In addition to the cost of telepsychiatry itself, hospitals must also obtain reimbursement for these patients through their insurance companies, an issue that telehealth experts have identified as a major barrier for incorporating telemedicine into everyday healthcare practice [33]. Within the US, Medicare covers telepsychiatry services only under certain specifications (e.g., historically for patients living in rural areas) [34, 35]. However, while it was found that there was an increase in rural Medicare beneficiaries who were able to access telepsychiatry services from 2004 to 2014, an overwhelming amount were using telepsychiatry to supplement in-person psychiatric care. Only 15% of beneficiaries used telepsychiatry as their sole source of mental health care [36]. Within the psychiatric ED visits that we examined, most (63%) were paid for by public insurance (including Medicare) and 98% occurred within urban settings. These findings parallel those found by Mehrotra et al. [36] and further support the benefit of Medicare coverage pivoting from its historic targeted population of rural beneficiaries. By adding the ability to cover telepsychiatry use in urban EDs, this could ensure the program is truly helping to increase access to care for those who do not have other sources of emergency psychiatric care. Currently, due to the COVID-19 pandemic, restrictions have been relaxed to allow Medicare coverage for use of telehealth in urban areas [37]. An extension of the current waiver could help motivate hospitals previously unwilling to invest in telepsychiatry services due to their location in a non-rural area.

Limitations

Our study has potential limitations. First, our decision to focus on one state limits the study’s statistical power as compared to analyzing a national sample. However, we chose New York not only due to its many Mental Health Shortage areas [27] but also because New York has both rural and urban EDs (Table 1), in addition to teaching hospitals and CAHs [19, 20]. The state had 7.8 million ED visits in 2016, which gave us a relatively large sample to analyze. Additionally, psychiatric patients also often struggle with physical health issues, which can complicate care and cause situations where their psychiatric illness is not the primary diagnosis for their ED visit but is still an underlying issue that requires treatment [38]. These could have excluded them from being in our visit population, however, we mitigated this issue by including visits where the secondary and tertiary diagnosis in SEDD were also psychiatric related. Our sensitivity analysis also showed the same difference in observation service usage and similar patient’s disposition among primarily diagnosed visits in EDs with and without telepsychiatry. In addition, the data we used was primarily administrative data for billing purposes and does not offer the in-depth detail of individual chart reviews. However, the SEDD/SID databases are often used to study patient-level data and outcomes across an entire state [39, 40]. Lastly, our study identified telepsychiatry service status as an ED-level variable rather than a visit-level variable, thus it is unknown which individual patients received telepsychiatry services versus in-person exams. Given that this is the first study to examine ED telepsychiatry state wide, review of service status at the facility level is appropriate. Given our finding that visits in EDs with telepsychiatry are less likely to use observation services, we encourage future studies to look into the possibility of identifying telepsychiatry use on a visit-level.

Conclusion

We investigated the association of ED-based telepsychiatry and ED visit outcomes across all psychiatric-related ED visits in New York State during 2016. Compared to EDs without telepsychiatry, the presence of telepsychiatry was associated with decreased odds of an observation stay, with similar final ED disposition. Factors affecting the delivery and effectiveness of telepsychiatry services to ED patients merit further investigation, especially in EDs with few resources or with a lack of any emergency psychiatric services whatsoever.