The Impact of the North Carolina Statewide Telepsychiatry Program (NC-STeP) on Patients’ Dispositions From Emergency Departments

In 2014, there were 137.8 million emergency department (ED) visits in the United States, a number that has continued to increase (1). Between 2006 and 2014, ED visits increased by 14.8% (1). During the same period, mental health– and substance abuse–related ED visits increased by 44.1% (1). In hospitals without psychiatrists or ED physicians (EPs) who feel comfortable treating mental health crises, patients with psychiatric symptoms are often boarded until they can be adequately assessed or treated by a psychiatrist. A boarded patient is “a patient who remains in the ED after they have been admitted or placed into observation status at the facility but has not yet been transferred to an inpatient or observation unit” (2). According to the American College of Emergency Physicians, boarding is a main cause of ED overcrowding, and patients with psychiatric symptoms are a key contributor to patient boarding (2, 3).

One proposed reason for boarding of patients with psychiatric symptoms is that EPs often lack the experience or training to manage mental health crises, and they are therefore unable to accurately evaluate whether patients are suitable for discharge (4). A survey by the American College of Emergency Physicians found that nearly 99% of EPs admit patients with psychiatric symptoms daily (5), despite most EPs having only limited training in psychiatry; only 2% of the questions on the American Board of Emergency Medicine certification exam are related to psychobehavioral disorders (6). In fact, evidence has suggested that when patients present with the same condition to either an EP or consulting psychiatrist, the two medical professionals often choose different courses of treatment for these patients, resulting in differences in patient disposition from the ED. One study found that EPs and psychiatrists agreed only 67% of the time on the need for involuntary holds and 76% of the time on final patient disposition (7). The agreement was greater when EPs thought a patient required admission rather than when they thought a patient could be discharged (7). A more recent study corroborated these results, finding that EP disposition decisions had an 87.3% positive predictive value compared with a psychiatrist’s disposition decision, suggesting that a significant number of patients were being admitted without a need for admission as determined by a psychiatrist (8).

These substantial disagreements about patient disposition suggest that some patients may end up being discharged, boarded, or admitted unnecessarily on the basis of whether they were assessed by an EP or a psychiatrist. EPs may err on the side of caution and board patients with psychiatric symptoms, whereas psychiatrists may discharge these same patients with outpatient treatment. It is imperative to streamline the disposition process for patients with psychiatric symptoms to ensure that these patients are getting appropriate care in a suitable setting.

Allowing patients with psychiatric symptoms who are appropriate for outpatient management to be discharged home can benefit both the patient and the hospital. One study found that when boarded, patients with psychiatric symptoms may develop anxiety and psychological stress because of the often chaotic environment in EDs (4). Appropriately discharging patients with psychiatric symptoms home allows them to recover in the comfort of their own homes and prevents a costly hospital stay. Unnecessary admission of patients with psychiatric symptoms also has a significant financial impact on ED reimbursement because of the lack of efficient bed turnover (4). In one study, psychiatric boarding caused a direct loss of $1,198 per patient when compared with nonpsychiatric admissions (9). When the inability to fill that bed with another patient was factored in, the ED lost more than $2,000 per patient (9). Another study found that providing services to other patients in ED beds occupied by boarded patients could generate almost $4 million of revenue for a hospital (10). With such a large population of behavioral health patients seeking treatment in the ED, hospitals are facing several challenges to control overcrowding, allocate resources effectively, and deliver high-quality care to all patients.

The North Carolina Statewide Telepsychiatry Program (NC-STeP) was developed in 2013 by the East Carolina University Center for Telepsychiatry with support from the North Carolina legislature and the Office of Rural Health of the North Carolina Department of Health and Human Services. The program connects remote hospital EDs with a psychiatrist via secure audio and video technologies to assess and treat patients with behavioral health symptoms. The program began in 2013 with 13 hospitals and one psychiatric provider hub site, and it has since grown to include 53 hospitals and seven clinical provider hubs (11). In this study, we examined differences in discharge disposition for ED patients with behavioral health issues between North Carolina hospitals where telepsychiatry services were available and those where these services were not available. We were particularly interested in outcomes for patients who had an extended length of stay (LOS) because of the challenges these patients pose for doctors, EDs, and themselves. We expected that the availability of telepsychiatry services in the ED would be associated with better discharge outcomes for patients and with fewer patients being transferred to a psychiatric facility.

Methods

NC-STeP

Hospitals participating in NC-STeP receive computer equipment along with training, program protocol, and support. This support system links participating hospitals to an NC-STeP psychiatric provider hub that provides consultations to patients at the remote hospital via an interactive audio-video link. NC-STeP was initiated in October 2013, but several hospitals already participated in a pilot program that began in 2012. As the program officially began, additional hospitals joined throughout 2014 and 2015, each with their own individual start date. Some hospitals also left the program. During a brief period in 2016, NC-STeP service was not available for some hospitals, and service was restored in the second half of 2016. The study period included the last quarter of 2012; all of 2013, 2014, 2015, and 2016; and the first quarter of 2017 (a total of 18 calendar quarters). We compared dispositional outcomes for ED patients with behavioral health issues when NC-STeP was available versus those when it was not available by assessing the proportions of patients who were discharged home or were transferred to a psychiatric facility.

Data Collection and Analysis

We used a deidentified, limited data set of patient claims records from the North Carolina Healthcare Association’s Patient Data System. The data set contained records for behavioral health patient visits to EDs for 30 North Carolina hospitals from October 2012 to March 2017. Each of the hospitals in the data set had participants in NC-STeP for at least one-quarter of the time during the 4.5-year study period. The elements in the patient claims data set included the hospital; the calendar quarter when the visit occurred; the discharge disposition; the patient’s age, gender, and race; and the LOS in whole days.

Because hospitals entered the program at different times, the particular period when NC-STeP was available differed for each hospital. For this study, we coded each individual hospital quarter as “program active” or “program inactive” on the basis of whether the hospital was an NC-STeP participant at that time. Quarters before a hospital joined were coded as “inactive.” Quarters when a hospital was participating were coded as “active.” Quarters when the service was not available or quarters after a hospital had left the program were also coded as “inactive.” We did not differentiate inactive quarters by whether they occurred before the program started, after it had been stopped, or during a gap in service, because we were primarily interested in whether patient outcomes differed when NC-STeP was available or not. Patient visits in the data set were then coded on the basis of whether they occurred during a program-active hospital quarter or not. For all hospitals, the patient visits that occurred during active quarters were combined into one program-active data subset, and the patient visits that occurred during any hospital’s inactive quarters were combined into one program-inactive data subset. We compared characteristics for patient visits between the program-active data subset and the program-inactive subset.

An “intervention” was defined as a completed assessment by a psychiatrist. The assessment included a diagnosis and treatment recommendations while the patient waited in the ED and dispositional recommendations whether to discharge to outpatient follow-up or pursue psychiatric hospitalization. Recommendations also included whether the psychiatrist should file for or overturn an involuntary commitment. When a patient stayed in the ED for >24 hours, the patient was seen for a follow-up assessment.

To assess the impact of NC-STeP on patient boarding, we specifically analyzed data for patients with an extended LOS, defined as a LOS of >2 days, which we used as a proxy for boarding. Operational definitions of boarding and extended LOS vary in the literature (12–18). When detailed LOS time-segment data are available, researchers generally calculate boarding as the time from when a patient’s examination in the ED has been completed to when the patient is discharged from the ED. When detailed data are not available, an extended LOS is generally used as a proxy for patient boarding, typically a LOS of >24–48 hours. In the North Carolina Healthcare Association’s data set, LOS was given only in whole days. LOSs of <24 hours were rounded up to 1 full day, and all other LOSs were also rounded up to the next full day. For our analysis, LOS was divided into two categories: 1–2 days and more than 2 days.

Discharge disposition was coded into four categories: discharged home, transferred to a psychiatric hospital, transferred to a health care facility other than a psychiatric hospital (i.e., skilled nursing, inpatient rehabilitation, or long-term care), or other disposition (i.e., left against medical advice, died, or transferred to law enforcement). We compared the proportion of visits with a discharge to home and the proportion of transfers to a psychiatric hospital for the program-active and program-inactive data subsets. We included only adult patients (age 18 and older) in our analyses.

The study used a limited data set with no identifiable private patient information. Although we did have patient information, patient visits were only identified by the calendar quarter when the visit occurred, aggregated at the level of the hospital and quarter, with no link to the patient. Therefore, this study did not constitute human subjects research, and institutional review board approval was not obtained.

Results

During the study period, 86,931 adults with behavioral health symptoms visited EDs at the 30 study hospitals. Of these, 44,857 (51.6%) visits occurred during periods when NC-STeP was available (active), and 42,074 (48.4%) occurred during periods when the program was unavailable (inactive). Of the visits during the program-active period, 76.0% (N=34,072) of patients were discharged home, compared with 72.2% (N=30,376) when the program was inactive (χ²=160.1, df=1, p<0.001) (Table 1). Of patients in the active period, 16.4% (N=7,363) were transferred to a psychiatric hospital, compared with 16.0% (N=6,725) during the inactive period (χ²=2.96, df=1, p=0.085).

Of all patients, 93.3% (N=81,125) had a LOS of 1–2 days, and 6.7% (N=5,806) had a LOS of >2 days. Among patients with a LOS of 1–2 days during the program-active period, 77.0% (N=32,131) were discharged home compared with 74.2% (N=29,237) when the program was inactive (χ²=89.08, df=1, p<0.001) (Table 2). When the program was active, 15.4% (N=6,441) of patients with a LOS of 1–2 days were transferred to a psychiatric hospital versus 13.9% (N=5,495) when the program was inactive (χ²=36.27, df=1, p<0.001).

Among patients with an extended LOS (>2 days), 62% (N=1,941) were discharged home when the program was active compared with 43% (N=1,139) when the program was inactive (χ²=207.81, df=1, p<0.001) (Table 3). When the program was active, 29% (N=922) of the extended LOS patients were transferred to a psychiatric hospital versus 46% (N=1,230) when the program was inactive (χ²=176.08, df=1, p<0.001).

Discussion

The findings of this study suggest that telepsychiatry effectively increases the rate of patients with behavioral health issues being discharged from EDs to their home and decreases the rates of hospital admissions and transfers of these patients. The proportion of patients with discharges home increased overall after NC-STeP implementation, and this increase was more pronounced among patients with a LOS of >2 days. NC-STeP implementation increased the proportion of patients with a LOS of >2 days who were discharged home by 19 percentage points. Among patients with a LOS of 1–2 days, the percentage of those discharged home also increased, but this increase was considerably lower (i.e., 2.8 percentage points). The conditions of patients with a longer LOS may be more complex, so the benefit of having a psychiatric consultation may be more pronounced for these patients. For example, a patient seeking treatment for drug intoxication may be discharged by an EP within 48 hours, whereas it may take longer to thoroughly assess a patient with symptoms of suicidal ideation with intent; the latter patient would likely be boarded until the EP feels comfortable discharging or until the patient can be assessed by a psychiatrist. These results show that telepsychiatry clearly has utility for patients with psychiatric symptoms in the ED for longer periods and whose conditions are presumably more complex. However, the results also call for more study of the utility of telepsychiatry in the ED for patients with less complex chief symptoms or with a shorter LOS.

Although the results of this study indicated an increase in discharges home when NC-STeP was active, it does not speak specifically to changes in boarding times of patients with psychiatric symptoms. Because our data set enumerated LOS only in whole days, we could not assess detailed boarding times before and after NC-STeP implementation. If telepsychiatry does in fact decrease boarding times, EPs could utilize this service to more effectively treat patients with complex psychiatric symptoms, better utilize ED resources, and decrease overcrowding in their ED.

The results of this study reveal that the availability of telepsychiatry consultation services increases the proportion of patients discharged home. This outcome is beneficial for both patients and hospital systems. Disposition home is financially beneficial for patients because they do not need to pay for a lengthy, and sometimes unnecessary, hospital stay. It is also likely to improve patient satisfaction with the ED experience. Rural and underserved hospital EDs, such as the ones studied in this analysis, do not have inpatient psychiatric units; therefore, any patients requiring inpatient psychiatric treatment must be transferred to an outside facility, which can take several hours to days. The expedited discharge of patients enabled by NC-STeP quickly frees up an ED bed, a process known to generate additional revenue for hospitals (9, 10). By boarding and admitting only those patients who require and would benefit from inpatient hospitalization, these facilities can use their limited time and resources for treating patients who need the most help.

Many patients who are stabilized in the ED may benefit from outpatient or community-based services after discharge to prevent further ED visits. Some studies have found that discharging patients home without social support and physician follow-up can increase the risk for readmission or adverse patient outcomes (19, 20). Telepsychiatry consultants need to be familiar with the community resources to connect patients to these resources. Although not the subject of this study, NC-STeP provided this service by creating hubs for psychiatrists in closer geographical proximity to the EDs they served with the understanding that psychiatrists are likely to be familiar with resources available in those areas.

We propose that it may be beneficial to assess readmission rates and outpatient follow-up rates for patients who are discharged home. Such assessment could help to determine how this disposition affects the patients’ health and influences long-term outcomes. Furthermore, analyzing patient dispositions on the basis of a patient’s diagnosis may generate information about which patients benefit the most from being assessed by telepsychiatrists. For example, are patients who seek treatment for suicidality more likely to be discharged home by a telepsychiatrist than by an EP? Finally, and as also further discussed below, gathering data that provide a more granular view on the time patients spent in the ED would allow better determination of the true LOS when telepsychiatry was available versus when it was not.

An important limitation of this study was that we did not have access to information about the actual recommendations of the telepsychiatrist; therefore, we were unable to determine exactly why the use of telepsychiatry increased the number of discharges home. The increase may be attributable to several factors, including more effective treatment in the ED or decreased perception of liability by ED staff as a result of a telepsychiatry consultation.

We also did not have any follow-up information on the patients in this study, so we were unable to assess outcomes for patients that were relevant for their communities. It would be beneficial to gather information regarding patient experience, resolution of symptoms, number of ED visits for the same chief symptom, admission rates, and patient knowledge or utilization of community resources for patients who were discharged home.

Our analysis included patients who had symptoms of drug intoxication or withdrawal. EPs may feel more comfortable treating patients with substance abuse symptoms than patients with other behavioral health emergencies and therefore may not use telepsychiatry consultations for the latter patients.

As mentioned in the foregoing, the data set documented LOS in whole days, rather than hours, so we could not precisely analyze the outcomes associated with patient LOS. For example, a patient who was boarded for 2 hours and another boarded for 12 hours were both documented as having a LOS of 1 day. Although this difference is only a few hours, similar differences in boarding time are associated with a large difference in mortality (21). Furthermore, the documentation of LOS in whole days prevents an accurate analysis of the financial impact of boarded patients.

Another possible limitation is that the findings of this study may have limited generalizability to other states because we analyzed only hospitals in North Carolina. However, we note that we included hospitals differing in size, demographic characteristics, and region to make the findings of this study more widely applicable.

Conclusions

Telepsychiatry consultation services such as NC-STeP may decrease ED overcrowding by increasing the number of discharges home and decreasing the number of transfers to psychiatric facilities. These results are therefore significant for both patients and hospitals. They indicate that availability of NC-STeP facilitates quick psychiatric assessment and treatment and increases discharge home, all of which likely promote patient satisfaction, improve outcomes for patients, and decrease their costs for a hospital stay. Increased ED efficiency and timely disposition planning are also financially beneficial for hospitals.