Introduction

Defined as the delivery of behavioral health services through video teleconferencing technology, telepsychiatry existed in the pre-COVID-19 era and was utilized by major healthcare organizations including the Department of Veterans Affairs [1]. Studies before the pandemic reflected that behavioral healthcare delivered through telepsychiatry is equal to, if not better, than face-to-face appointments [2]. Telepsychiatry is effective across many populations, varying psychiatric illnesses, and when utilized in different settings [2–3]. These promising results led many to believe telepsychiatry to be a valid alternative when face-to-face appointments are not possible [4].

The telepsychiatry landscape today is different than in the pre-COVID era. Many clinicians and organizations were forced to rapidly adopt telepsychiatry [5]. The sheer number of virtual appointments and patients served from distances afar, including different states than the clinician, have drastically changed. Within one large health system of the United States, during the COVID-19 pandemic, a patient was 6.68 times more likely to complete a telepsychiatry visit than a face-to-face appointment [6]. The Department of Veterans Affairs saw dramatic rises in telehealth appointments [7]. The transition to telepsychiatry has been largely positive for patients and clinicians alike, despite some frustrations and limitations [8–9]. A recent analysis estimates that telehealth will continue to play a critical role in patient care delivery even after the current public health emergency [10]. With this shift to increasing telepsychiatry utilization, organizations must continue to plan for future telehealth delivery [11].

Despite the wide adoption of telepsychiatry services in the face of the COVID-19 pandemic, little is known regarding how clinicians have engaged in safety planning to handle behavioral health emergencies from afar. The American Psychiatric Association (APA) offers a toolkit for telepsychiatry which includes information on “Patient Safety and Emergency Management” [12]. This toolkit outlines unique features of telepsychiatry that clinicians must consider while assessing patient safety. For instance, recommendations include being mindful of such issues as the patient’s level of agitation, potential for harm, safety hazards, and physical location that will help the clinician determine if a higher level of care, including civil commitment, is required [12]. Further, the APA Committee on Telepsychiatry and the APA College Mental Health Caucus submitted best practices in August 2020, outlining specific considerations in caring for college students. However, there was no guidance enumerated in these best practices for managing acute safety issues [13]. In April 2018, the APA and the American Telemedicine Association published “Best Practices in Videoconferencing-Based Telemental Health” [14–15]. These best practices included recommendations during emergencies, including that “professionals shall maintain both technical and clinical competence in the management of mental health emergencies.” These recommendations discuss having a procedure or protocol for mental health emergencies with clear roles and responsibilities as well as awareness of resources including police, emergency rooms, and crisis teams. Further, it is recommended that clinicians are familiar with civil commitment regulations and have arrangements to initiate civil commitment [14]. Issues around civil commitment, in particular, pose challenges for clinicians who may not be familiar with telehealth practice. Further, due to the COVID-19 pandemic, some states initiated changes to civil commitment statutes to accommodate evaluations conducted using virtual technology.

Despite the importance of assessing and managing mental health emergencies via telehealth, there has been no published account of clinicians’ experience in providing emergency services through virtual care, including petitioning patients for civil commitment. Given the rapid expansion of tele-behavioral health during the COVID-19 pandemic, an understanding of the prevalence of the need for activation of emergency planning via telehealth and use of civil commitment would be beneficial for further development of guidance and best practices in this space. In order to understand this need, we conducted a survey of behavioral health clinicians practicing within a large, academic medical center, where over 94,000 real-time audio-visual virtual visits, along with over 16,000 audio-only (telephone) virtual visits were conducted between March 2020 and January 2022.

Methods

Survey

We conducted an anonymous, electronic survey within a large academic psychiatry department to assess clinicians’ experience with assessing and managing emergent safety concerns while delivering outpatient tele-behavioral health services during the COVID-19 pandemic (Supplemental Materials). For the purposes of this study, “emergent safety planning” is defined as “recommending a higher level of care such as inpatient hospitalization, recommending presentation to an emergency department, utilizing a police welfare check, activating a mobile crisis team, petitioning or considering petition for involuntary commitment.” Qualtrics ™ was used for survey development, distribution, and collection. The survey was distributed to all clinical providers within the academic department, even though some clinicians do not practice in the outpatient setting. Ambulatory clinical leadership provided a list of clinicians who perform at least some ambulatory care in the department. Surveys that were started but not fully completed were excluded from the analysis. This study received Institutional Review Board (IRB) exemption by the Office of Human Research Ethics at University of North Carolina at Chapel Hill (UNC).

Analysis

All analyses were done using the FREQ procedure in SAS 9.4. P-values for tests of no association between each binary variable (proactive, IVC, aware) and each nominal variable (for example, “Credentials” and “Are you a trainee”) were obtained using. Pearson’s Chi-Square test. For ordered variables (for example, “How many years of experience do you have providing direct patient care after completing your terminal degree?“) the Cochran-Mantel-Haenszel test of equal means was used.

Results

We received 111 complete responses out of a possible 393 clinical faculty (28.2% response rate). Based on discussions with ambulatory clinical leadership, only 204 of those 393 clinicians were actively involved in ambulatory care in the year prior to survey administration (adjusted response rate = 54.4%). Of the respondents, 51 identified their primary credentials as MD (44.1%), 25 respondents were PhD or PsyD (25.2%), and 21 individuals were LCSW (18.1%) (Table 1). The vast majority, 82 respondents, identified that over 75% of their outpatient clinical encounters were conducted virtually since the start of the COVID-19 pandemic.

Table 1 Descriptive data of tele-behavioral health survey respondents
Full size table

23.3% of respondents reported engaging in proactive safety planning for over 50% of their virtual visits (Table 2). 65.8% of participants felt like they engaged in proactive safety planning during virtual visits at the same rate as in person visits, while 7.4% felt like they did more proactive safety planning for virtual visits. There were no statistically significant associations between clinical credentials, years of experience, or type of psychiatry practiced and engaging in proactive safety planning. A slight majority of respondents (53.2%) indicated they implemented emergent safety planning during one to five virtual visits. 27.0% of participants reported never utilizing emergency interventions for patient safety. Of the clinicians who emergently activated a safety plan, 58.3% felt its utilization was successful over 95% of the time to effectively reduce the risk of harm to self or others. A majority of respondents (54.1%) indicated that virtual visits necessitated emergent safety planning at about the same frequency as in-person visits, while 9.9% felt they enacted emergent safety planning more frequently using virtual visits. Enacting emergent safety planning was not associated with clinical credentials, years of experience, or type of psychiatry practiced. However, clinicians who more frequently utilized emergent protocols were more likely to engage in proactive safety planning with their patients prior to an emergency (p = 0.0115).

Table 2 Proactive and emergent safety planning by tele-behavioral health clinicians
Full size table

When participants were asked for specific emergency interventions utilized while providing virtual care, 60.3% of respondents recommended a higher level of care (such as inpatient hospitalization) and 57.7% of participants recommended that a patient present to the nearest emergency department. Just 13.5% of respondents reported calling a police welfare check. While 33.3% of respondents considered petitioning a patient for involuntary commitment, just 10.8% of participants completed a petition for involuntary commitment. Of those who petitioned for civil commitment, 22.2% indicated they had done so after a telephone (audio only) encounter, while the remaining were performed using synchronous audio and visual technology. Respondents identified a multitude of problems while attempting to petition for civil commitment, with common responses including “identifying local magistrate contact information” (14.8%), “identifying ambulatory staff members to assist” (14.8%), “accessing electronic petition paperwork” (11.1%), “accessing a fax machine/secure communication method” (11.1%), and “identifying who to send the petition to” (11.1%). Some clinicians noted that when having to petition a patient for involuntary commitment, it interrupted patient care and resulted in canceling the remainder of the clinic day. Narrative feedback illustrated three key themes: (1) Fear of virtual emergent safety planning, (2) Difficulties/limitation with risk assessment using virtual care, and (3) Issues with the civil commitment process (Table 3).

Table 3 Narrative comments regarding civil commitment (IVC) sorted by key themes
Full size table

Discussion

Despite the extensive utilization of telepsychiatry since the start of the COVID-19 pandemic in March 2020, little is known regarding the clinician experience in response to emergent situations. This work surveyed mental health clinicians at a large academic psychiatry department and revealed that only one quarter of clinicians conducted proactive safety planning for most of their virtual visits. In addition, a small majority of respondents implemented emergent safety planning while providing virtual care, while only a small portion of these individuals had to petition patients for involuntary commitment. The number of individuals who engaged in proactive safety planning is concerning, given that taking such steps are generally recommended by experts in virtual care. While it is possible that many clinicians are working with patients who are felt to be at “low-risk” of needing emergent steps to maintain safety, it may also represent a lack of knowledge and training in best practices while providing virtual care. Some organizations conduct or recommend specific telepsychiatry training or certification programs for clinicians. Our data suggest that these may be important to raise awareness of the need for proactive safety planning as a standard part of tele-behavioral health delivery.

While a small majority of respondents felt like emergent safety planning was necessary at the same rate as in-person visits, a substantial portion (17.1%) felt like it was less necessary than for in-person visits. This was surprising, given that it is generally accepted that behavioral health symptoms and crises have risen during the pandemic, especially for certain populations [16,17,18]. While unable to compare this self-reported measure with actual rates of emergent safety planning, it is possible that use of tele-behavioral health led to less robust safety assessments in some cases, thus leading to a lower need for emergency planning. This result suggests that a more rigorous study of safety assessments conducted by virtual means compared to in-person evaluations may be warranted. Surprisingly, only 58% of respondents felt like enactment of their emergent safety planning was successful at reducing risk in most cases. It is difficult to compare this value to in-person visits, as we could not find published data in the literature. It is possible that this rate is lower than would be expected for traditional, face-to-face care due to an increased sense of control over the situation should a patient require activation of emergency services, transportation to an emergency department, or initiation of involuntary civil commitment. However, future studies would be necessary to test this hypothesis.

While approximately 10% of respondents initiated the process for involuntary civil commitment while performing tele-behavioral health, narrative comments suggested many clinicians had frustrating experiences when doing so. In addition to expressing fears and anxieties regarding encounters that necessitated emergency safety planning, as well as an impression from clinicians that their risk assessments were impacted by limitations of virtual care, what was perhaps most concerning was the avoidance of using involuntary commitment due to workflow. Some clinicians opted to send the patient to the emergency department or call the police for a welfare check rather than petitioning for involuntary commitment. Based on narrative comments, this avoidance may partly be due to the increased perceived burden and challenges associated with pursuing civil commitment while caring for patients from afar, including the need for notarization services and identifying the patient’s local magistrate.

While less restrictive alternatives to involuntary hospitalization are always preferable, there are clinical encounters that necessitate civil commitment to ensure the safety of the patient and society. Relying on law enforcement or local emergency departments as a safety net is not always an ideal or realistic contingency plan. The protocols that many large academic medical centers previously instituted to manage acute risk must adapt to the changes that accompany providing care from afar. These changes include providing care to patients who are located across the state, and thus are under the jurisdiction of different police agencies and court systems. Additionally, these patients may have different local services, such as availability of acute mental health care such as mobile crisis teams. Clinicians may be familiar to the protocols in the surrounding areas of a large academic department, but less familiar with the procedures of other counties and cities. Creating an accessible, centralized resource for clinicians working virtually to assist with emergent safety planning may be beneficial. An additional step would be to digitize commitment paperwork to reduce clinician burden during the petition process, which often utilizes paper forms and notary services. A digital petition would further minimize the challenges identified in this survey including accessing the requisite paperwork, identifying magistrate and notary information, accessing a fax machine or other secure communication method, and knowing where to send the petition.

This survey of behavioral health clinicians practicing within a large, academic medical center revealed that 10.8% of clinicians petitioned a patient for civil commitment. This was over a time period in which the department conducted over 110,000 virtual visits. It is difficult to say how this rate compares to that of in person care. While there is much in the psychiatric and legal literature relating to civil commitment, accurate tracking of its prevalence is sparse. In fact, a 2021 article revealed that North Carolina was one of many states without data on involuntary psychiatric hospitalizations [19]. A 2007 survey of 1500 members of the American Psychiatric Association (APA) revealed that while 62% of respondents had direct experience with civil commitment in the prior two years, the percentage of the respondents’ patients involuntarily hospitalized was low, at a median of 0.03% and mean of 0.1% [20]. Tracking the rate of civil commitment and patients throughout the process are crucial to gaining a better understanding of this contentious intervention and whether there are differences between in-person and virtual care [21].

Historically, involuntary commitment statutes required an in-person assessment, though some states have statutorily defined this assessment to include telehealth, including North Carolina, as well as California’s Lanterman-Petris-Short Act, commonly referred to as a 5150 [22]. However, North Carolina statute language excludes an evaluation completed entirely over the telephone, or other means of assessing a patient without visualization of the respondent [23]. This creates challenges for individuals who are providing telephonic behavioral health care for patients who are unable to access or utilize two-way audio-visual services. Given that certain states and the Center for Medicare Services have made permanent adjustments to allow audio-only behavioral health to be reimbursed, this seems to create a situation that may put some clinicians providing audio-only treatment in situations where they are concerned about patient safety but are unable to initiate civil commitment processes. There is limited case law to guide clinicians in this matter, although there is precedent that if a psychiatrist is familiar with a patient through a treating relationship, a telephone interview meets statutory requirements for a personal examination [24]. Given the limited guidance present, behavioral health clinicians conducting audio-only care should be aware of their limited ability to enact involuntary civil commitment, depending on local laws.

These survey data are associated with limitations in that it only included clinicians from one academic psychiatry department. It is possible that mental health clinicians in the community and at other institutions have encountered different scenarios and implemented different protocols to manage emergent safety concerns while providing virtual care. We also note that physicians were relatively underrepresented in our sample compared to the academic departments’ clinician ratio (46% of respondents were physicians, while 59% of those eligible to respond were physicians). We do not anticipate that this would meaningfully bias our results in any way, although it is possible that certain types of behavioral health clinicians are more likely to utilize certain types of proactive or emergent safety planning. Another limitation is that this survey data only focused on self-reported measures. While there is central reporting for inpatients who are civilly committed, there is no tracking of outpatients who may have presented to emergency departments across the state or for those petitioned for commitment in other jurisdictions. Thus, correlating the survey data to reportable behavior was not possible. Future directions would include measures of clinician behavior as related to the acuity of their outpatient patient panel, such as the utilization of emergency services.

As best practices for virtual care continue to be established, it is critical to address risk assessments and appropriate risk mitigation strategies to keep patients and society safe. Clinicians need increased training on the importance of proactive safety planning, as well as assessing safety and understanding when the limitations of telepsychiatry impede an adequate safety assessment. Clear protocols for emergent safety planning should be implemented at all clinical sites, with knowledge of local resources and services depending on the patient’s residence. These protocols need to be adaptable and centrally located so that clinicians can easily access them during patient encounters. These processes will be important as continued utilization of virtual care for behavioral health develops.

Declaration of Conflicting Interests

The authors declare that there is no conflict of interest. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.