Management of Mental Health Crises Among Youths With and Without ASD: A National Survey of Child Psychiatrists

Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental disorder marked by persistent deficits in social communication and interaction along with restricted and repetitive behaviors (1). Roughly 70% of children with ASD have at least one co-occurring psychiatric diagnosis, and over 30% have two or more co-occurring psychiatric diagnoses (2–4). Both externalizing problems (such as aggression, disruptive behavior, and self-injury) and internalizing problems (such as anxiety and depression) are more common among children with ASD than among typically developing children (5) and youths with intellectual disability (6). These conditions are often persistent (7) and highly impair functioning of both the child (8,9) and the family (10).

Despite the significant relationship between ASD and mental illness, qualified mental health providers who are available to work with this population are scarce (11–13). Of particular concern is a national shortage of child psychiatrists with expertise in ASD. One major reason for this deficit is that medical schools and psychiatry training programs offer limited clinical experiences and didactics in ASD (14). Marrus and colleagues (14) surveyed general psychiatry and child psychiatry training directors and found that trainees see one to five individuals with ASD or intellectual disability per year and receive only a few hours of didactic training per year about these conditions.

Given the high rates of psychopathology among youths with ASD, coupled with the shortage of providers, this population is at risk of experiencing mental health crises. The term mental health crisis is defined as “an acute disturbance of thought, mood, or behavior that requires immediate intervention and the resources available to manage the situation are not available at the time and place of occurrence” (15). Currently there are no instruments to assess the presence of a mental health crisis among children or adolescents. As such, one approach to assessing the prevalence of crises is by reviewing emergency department (ED) data, given that individuals experiencing an acute psychiatric event, such as explosive aggression and suicidal behavior, often present to the ED. Current data show that individuals with ASD have higher rates of ED visits for psychiatric purposes compared with those without ASD (16–18), suggesting that the prevalence of mental health crises among youths with ASD may be high.

It is critical to understand whether clinicians feel prepared and have adequate resources to manage mental health crises among youths with ASD. Gathering the perspective of child and adolescent psychiatrists on these topics is particularly relevant because children with ASD may present with severe psychopathology by the time they finally get an appointment with a child psychiatrist. Some may be close to experiencing a mental health crisis. No study, to our knowledge, has examined this topic.

This study examined whether child psychiatrists differed in their management of mental health crises among youths with and without ASD. Using online survey data, we examined several facets of crisis management, including whether child psychiatrists are willing and prepared to see youths in crisis in their outpatient practice and whether they have access to sufficient resources to manage these events.

Methods

Sample

A custom online survey was administered to members of the American Academy of Child and Adolescent Psychiatry in February 2015. To be included in the survey, the participant must be a child and adolescent psychiatrist, have actively seen patients within the past three months, and practice in the United States. Child psychiatrists who did not meet these criteria or who were not interested in participating (N=172) could “opt out of the study.” Informed consent was obtained. The study was approved by the local institutional review board.

Mental Health Crisis Survey

The survey assessed three domains of crisis management: willingness to work with youths who are experiencing a mental health crisis (items 4, 6, and 7), comfort level in managing a mental health crisis (items 5 and 8), and assessment of external resources during a crisis (items 1–3, 9, and 10). All responses were based on a 5-point Likert scale, with lower scores representing greater access to a resource or a more positive attribution about a particular service. [A list of survey items is available as an online supplement to this article.]

Creation of Psychiatrist Groups

Psychiatrists were divided into two groups on the basis of the percentage of youths with ASD they reported seeing in their outpatient practice. Those who reported that youths with ASD accounted for 6% to 15% or more of their caseload were considered to routinely treat this population. Those who reported that less than 6% of their patients had ASD served as the comparison group because they hardly saw any individuals with ASD. The two groups received identical crisis survey questions with the only difference being that psychiatrists who routinely saw youths with ASD received questions that pertained specifically to this population. A total of 492 and 374 psychiatrists reported on youths with ASD and youths without ASD, respectively.

Efforts to Increase Response Rates

To maximize response rates, basic survey design principles—such as personalizing the recruitment e-mail and highlighting the brevity of the survey and the incentive (19)—were initially implemented. Second, a randomized-incentive design was implemented (20) to increase response rates in a randomly selected subsample. This design was chosen because it would be impractical to summon the resources needed to achieve a high response rate for the entire cohort. This procedure resulted in two study arms. The randomized arm (N=400) submitted psychiatrists to a five-step intensive protocol, which initially provided each participant with a $10 gift code to Amazon.com for completing the survey. Nonresponders received two additional follow-up e-mails with an increased incentive. The remaining nonresponders in the random sample were mailed a hard copy of the survey and received a reminder phone call. Psychiatrists who were not randomly selected for the randomized-incentive protocol (N=3,792) were entered into a raffle for one of five $200 gift codes to Amazon.com. [A figure illustrating participant flow through the study design, including survey response and completion rates, is available in the online supplement.]

Data Analysis

Addressing missing data was the first priority of the analysis. Missingness rates ranged from .05% of the data set for information about psychiatrists’ age to 7.5% of the data set for information about average wait time. Whereas the average rate of missingness per variable was low (3.7%), only 80% of the entire data set had complete information on all variables. To address this problem, we used multiple imputations via chained equations to create 25 complete data sets (21).

The second step was to maximize comparability between the two psychiatrist groups who did and did not routinely see youths with ASD. Inverse probability of treatment weights (IPTWs), a propensity score–based method, were employed to balance observed covariates between these groups (22,23). Another use of IPTWs is to account for selection (or nonresponse) bias (23). We utilized the randomized-incentive design to calculate a second set of IPTWs to account for this bias. Among all respondents, weights were calculated as the inverse of the probability of being in the randomized group and the inverse of 1 minus the probability of being in the nonrandomized group, with calculations conditioned on all covariates listed in Table 1 and Table 2. This procedure balanced the respondents from the nonrandomized group with the respondents from the randomized-incentive group, under the assumption that the latter better reflects the target population since they were randomly selected and the response rate was much higher when compared with the nonrandomized group. The two sets of IPTWs were then combined, by multiplying the weights, and were stabilized by dividing the combined weights by the marginal probability of being in the group that reported on youths with ASD.

Last, mean differences between the groups that did or did not report on patients with ASD in each of the mental health crisis survey items were assessed. This analysis was conducted by using a doubly robust linear regression model (24). To better understand the success of the weighting procedure at equating the two groups of psychiatrists, unweighted and weighted means are presented, along with standardized mean differences (SMDs) [see online supplement]. After the weighting procedure, the SMD for all items was <.01, reflecting excellent balance (<1% difference in the covariate) between groups. All analyses were conducted in Stata 12.0 by using the mi suite to create and handle the multiple imputations. Alpha was set at .05 for determining statistical significance.

Results

Demographic Characteristics

Both groups of child psychiatrists were predominantly Caucasian and non-Hispanic (Table 1) and were evenly distributed over the number of years in practice (Table 2). Psychiatrists who reported on youths with ASD were slightly younger (p<.01). Both groups had comparable geographic dispersion, except in the North Central and Midwest regions, where there were slightly more psychiatrists who reported on youths with ASD compared with psychiatrists who reported on youths without ASD (p<.01).

Clinical Practice Differences

Clinical practice characteristics were similar between the two groups with respect to time spent on patient care (primary activity), the proportion of physicians working in a teaching hospital, time spent in an inpatient or outpatient setting, and adults seen in their practice (all p≥.1) (Table 2). However, psychiatrists who reported on youths with ASD saw more patients (p<.01), conducted more evaluations (p=.06), and had longer wait times (p<.01) compared with psychiatrists who reported on youths without ASD. The weighting procedure successfully balanced the two groups with respect to these and other group differences [see online supplement].

Crisis Management

Overall, 24% (N=622) of psychiatrists did not “frequently” or “often” accept a child with a history of mental health crisis. In the weighted analyses, there were no significant group differences regarding willingness to accept new patients with a history of mental health crisis or in the use of office designs that allow the psychiatrist to manage a child experiencing a crisis (Table 3). Psychiatrists who reported on youths with ASD, however, were more likely to keep an appointment open in case a child had a mental health crisis (p<.05).

Both groups were equally comfortable treating youths in crisis. Yet, psychiatrists reporting on youths with ASD were slightly less likely to focus on developing emergency crisis plans for youths with ASD compared with psychiatrists reporting on youths without ASD (p<.05). Psychiatrists reporting on youths with ASD also indicated having less access to other mental health professionals, such as psychologists and social workers, and to a psychiatric crisis evaluation center that could facilitate an inpatient admission if the child was in crisis (both p<.05). Psychiatrists who reported on youths with ASD were also less likely to recommend that parents take their child to the ED during a mental health crisis and endorsed less confidence in the ability of ED health care professionals and the police to manage youths with ASD in a safe and developmentally appropriate manner during a crisis (all p<.05).

Discussion

Child psychiatrists were willing to see children in crisis, regardless of whether they reported on youths with or without ASD. However, one key difference is that child psychiatrists reported having fewer external resources when managing mental health crises among youths with ASD versus youths without ASD. Typically, when a child has a mental health crisis, psychiatrists either call emergency responders (police or 911), send the child to the ED with the parent, or work with the parent in the office to seek inpatient hospitalization through an urgent crisis evaluation center. In this study, we found that a gap in care was evident across all three of these services for youths with ASD.

There are several reasons why child psychiatrists may be less inclined to seek assistance from an ED or emergency responders when managing youths with ASD during a crisis. First, concerns about calling 911 or the police may stem from media reports documenting harmful interactions between youths with ASD and the police (25,26). Second, reluctance to send a child to the ED may be driven by previous reports indicating the use of high levels of chemical and physical restraint to manage agitated youths with ASD in this setting (27). These types of experiences can be traumatic for youths with ASD and their families, could result in physical injuries, and may aggravate psychopathology in the future. In addition to provider concerns, our clinical experience suggests parents are fearful of the emergency response system, including the ED, and would prefer to “ride out” the crisis at home rather than engage these systems. These perceptions may also influence a psychiatrist’s medical decision making during crisis situations.

Child psychiatrists may also hesitate to send families to an ED because the psychiatrists are aware of the shortage of inpatient psychiatric units for children with ASD. Currently, there are only nine specialized inpatient units in the country serving youths with developmental disabilities (28). General child psychiatric inpatient units may refuse to admit youths with ASD or may accept these children but lack the necessary resources to provide multidisciplinary programming (28). Long wait times are also a concern. Among youths in general, psychiatric ED visits have been steadily increasing over the past decade (29,30), even as the number of EDs across the United States is shrinking (29). The confluence of these events may affect the ability of ED clinicians to provide timely medical care because of overcrowding and increased wait times. The treating psychiatrist is therefore faced with the dilemma of deciding whether to subject a child to potentially long wait times and use of restraints in the ED or to manage the crisis in an outpatient setting until an inpatient bed becomes available.

Our data show that other factors also constrain management of crises among youths with ASD. We found that psychiatrists treating youths with ASD were slightly less likely to have access to other mental health specialists compared with psychiatrists treating youths without ASD. These professionals, such as behavioral psychologists and social workers, can target important contributing factors, such as behavioral factors, caregiver stress, and lack of wraparound services, that the psychiatrist may not be fully equipped to address. We also found that child psychiatrists seeing youths with ASD were less likely to help families develop emergency crisis plans. Families may therefore be at a loss about what to do when their child’s behavior escalates. Working with families to develop crisis plans could help reduce adverse outcomes. Crisis plans could include psychopharmacologic treatments, such as the use of PRN medications; specific behavioral interventions; and encouragement for parents to proactively connect with local emergency responders, such as police and ED clinicians, about how best to manage a crisis involving their child.

Although resources are limited, there are several promising initiatives underway to improve crisis management for youths with ASD. For example, crisis intervention teams are being developed to build community partnerships between law enforcement agencies and families, schools, and community mental health organizations (31). Similar programs are also being developed to help train ED providers in the management of crises involving youths with ASD (32). Another area of research involves testing whether community-based crisis intervention models, such as the START (Systemic, Therapeutic, Assessment, Resources, and Treatment) program, can divert ED visits for individuals with a developmental disability (33). Finally, academic and governmental alliances, such as the LEND (Leadership Education in Neurodevelopmental Disabilities) program, can continue to help increase the workforce of mental health professionals working with youths with ASD.

One finding of note from this study is that 24% (N=201) of psychiatrists did not “frequently” or “often” accept a child with a history of mental health crisis (N=43 missing). This suggests that parents of children with psychopathology may have difficulty obtaining an appointment with a child psychiatrist. There may be many reasons for this finding, including concerns about a lack of resources in the office for managing dangerous behaviors. Further research is needed to better understand factors that may contribute to this practice pattern for psychiatrists generally, and specifically for those managing youths with ASD. A greater understanding is also needed of the proportion of psychiatrists who accept adults with ASD experiencing a mental health crisis and differences in the management of crises involving patients with ASD between psychiatrists who are specialists in ASD and those who see only a few of these youths in their general practice. Last, this finding suggests that frontline providers, such as pediatricians, may benefit from tools that assist them in managing mental health crises while families seek out a psychiatrist for their child.

The foremost limitation of this study was that the data presented are not representative of all U.S. child psychiatrists. Selection effects also are related to survey nonresponse, which was higher than desired and may have yielded a biased sample; however, our response rate is consistent with previous survey research among child psychiatrists (34,35). To offset nonresponse bias, we utilized IPTWs within the randomized-incentive design. In an era of decreasing survey response rates across all fields (36), our novel use of the randomized design may spawn greater study of cost-effective approaches that can offset growing concerns about selection bias in national surveys. Beyond concerns about nonresponse bias, additional limitations included the potential for confounding due to systematic differences between the groups that remain unmeasured (for example, rurality) as well as the accuracy and specificity of provider responses. For the latter, this was particularly a concern for the ASD group because some clinicians may have reported on a population that represented a minority of their overall clientele. Finally, the goal of this study was to compare differences in the way child psychiatrists manage mental health crises between youths with and without ASD. We chose to examine this question by creating groups based on the number of children with ASD that providers see in their practice. A different study design, such as one that compares crisis management strategies for youths with and without ASD among clinicians who routinely see youths with ASD, could possibly yield different findings.

Conclusions

This study showed that child and adolescent psychiatrists have limited external resources to manage mental health crises among youths with ASD. These findings underscore the need for increasing the acute care resources available to parents and clinicians who are involved in managing this population.