Lost Keys: Understanding Service Providers’ Impressions of Frequent Visitors to Psychiatric Emergency Services in Singapore

For many marginalized people with mental illness, the emergency department (ED) increasingly represents their main point of contact for services (1,2)—a choice they do not make lightly (3,4). Because of system inadequacies that fail to support continuity of care, a certain portion of these individuals continue to rely on the ED to obtain care that could be provided by specialized professionals (1,5) or avoided with proper treatment of comorbid conditions (6). In some jurisdictions, the top 2% to 8% of frequent ED users consume a disproportionate amount of the resources—in excess of 20% (2,7–9; personal communication, Lucier L, 2012). Previous research has shown that various patterns of use exist among frequent users; frequent use may result from acute episodes or occur as a result of chronic conditions (7,10). This research has found that frequent users were more likely to have a diagnosis of schizophrenia or personality disorder than other diagnoses, echoing other studies (8,11,12). Among very frequent ED users, persons with mental illness are overrepresented. Some evidence suggests that among ultra-high frequent users (15 or more visits in a year), 20% have a primary diagnosis of a substance use disorder and 6% have another psychiatric diagnosis (13).

These groups have high service needs, but they also tend to receive care that is less satisfying to them (4) because of the opinions of service providers (14–16), who may regard frequent ED users as a source of unnecessary burden (17,18). Because the opinion of staff members has an impact on the quality of care, it is essential to understand their opinion of service seekers (19).

Our goal was to identify characteristics of frequent users of ED services and explore service providers’ opinions of and experiences with this particular group of service seekers.

Methods

Setting

The ED in this study represents the default source of emergency psychiatric care in Singapore. It is located in the Institute of Mental Health, the largest provider of inpatient and outpatient psychiatric care. To ensure low-barrier access to care, service seekers without financial means are referred to medical social workers and enrolled in public-benefits plans that subsidize costs (20). Persons who require nonpsychiatric medical attention are channeled to other hospitals. Individuals who do not require admission may be referred to outpatient services operating within the institute, step-down care, family service centers, or polyclinics. Service seekers may also be given the option to receive follow-up treatment at a general hospital closer to their residence.

Sources of Information

We used two data sources to inform our description: administrative data for all adult (age 18 and older) visits in 2014 and qualitative interviews with the staff who contributed to the administrative data (21). We deemed their input indispensable to understanding the administrative data because nuances are lost when researchers deal exclusively with data (19). We chose a convergent mixed-methods design to combine the two types of data at the reporting phase (22).

Quantitative Data

Information management technicians extracted administrative data according to a piloted template of necessary fields. Data management software checked for missing data, and staff made necessary changes to align the electronic records with physical case notes prior to analysis. For cases in which admission followed the visit, we consulted admission records to verify the ICD-10 psychiatric diagnosis. For cases in which the diagnosis changed, we used the diagnosis made by psychiatrists after their clinical interview rather than the diagnosis made by ED physicians. We reviewed 16,123 individual visits of 10,108 adult ED visitors for the development of the description. This represents a rate of 45.7 visits per day, with a lower rate on weekends.

The cutoff for the number of visits defining frequent use of emergency services has varied, with different definitions representing distinct groups (7,8). The definition proposed by Pasic and colleagues (7) for acute frequent users (four visits in three months) was not associated with demographic or service use variables, and the definition they proposed for chronic frequent users (visits numbering two standard deviations above the local mean) did not fit with the nonnormal distribution of the data. Other cutoffs of absolute number of visits of four or more (5,11,12) were explored. Ultimately, we chose a cutoff of five visits in a year to divide the data into proportions resembling those reported in other studies (2,9; personal communication, Lucier L, 2012).

We used random-effect logistic regression models, which took into account the longitudinal nature of the data (clustered at the individual level), to associate diagnosis with frequent use, while controlling for sex and age (23). Diagnoses were collapsed into categories to facilitate interpretation. [A table in an online supplement to this article lists ICD-10 diagnoses included in each category.] A longitudinal model was chosen because of the diagnostic variability known to result from frequent visits (24). We tested for diagnostic variability and collapsed opioid use disorder into drug use disorder and conduct disorder into disorders usually diagnosed in childhood. In 614 cases (27%), the diagnosis varied from the one made at the previous visit.

Qualitative Data

We conducted 26 interviews with staff members responsible for data production. Participants gave written informed consent. Ethics approval was obtained from relevant ethics review committees. We queried their opinion of frequent visitors and explored their understanding of the types of service seeker groups they encountered. We interviewed all types of staff, including eight physicians, eight nurses, two assistant nurses, five patient service assistants, two health care assistants, and one service executive. This sample was sufficient to reach saturation of the themes reported below. Two assistant-level staff declined. Participants’ average job tenure was 4.6 years (range of two months to 16 years); some physicians had three-month rotations.

As described in the constant-comparative method of data collection (25), findings from completed interviews guided the selection of the next type of staff member interviewed. We also altered our guiding questions on the basis of emerging findings, but we included core questions (26) [see online supplement], such as “How would you describe a typical ED user?” We avoided leading questions such as “Do they make you feel angry?” We used thematic analysis to highlight content of interest in a deductive manner (27). Codes outlined at the beginning of the analysis process were related to visitors’ characteristics and presenting complaints, reasons for visits, and the impact of frequent users on staff. Interview content related to these themes was extracted. The quotes presented below represent the most frequent content and capture the emotions expressed regarding the themes. Two validation sessions held with staff at the end of the project strengthened confidence in our interpretation.

Results

Characteristics of the sample of service seekers are presented in Table 1, and Singapore census data are provided for comparison (28). The frequent-user definition of five or more visits in the year 2014 effectively distinguished the top 3% of service seekers (N=331), who were responsible for 16% of the visits, from infrequent users. For all ED visitors, physicians most commonly diagnosed psychotic disorders, adjustment disorders, and depressive disorders during the ED visit or the subsequent admission. The disposition of ED visits did not differ between frequent and nonfrequent users.

Characteristics of Frequent Visitors

A regression analysis based on the administrative data indicated that service seekers diagnosed as having substance use disorders, psychotic disorders, or personality disorders had significantly increased odds of being frequent users, compared with those diagnosed as having depression (Table 2). Only those who were kept for observation without a diagnosis had lower odds of frequent ED visits compared with those diagnosed as having depression. Frequent users had a consistent visit rate over the week, unlike the patterns of infrequent users, which were lower on weekends.

Service providers’ descriptions of frequent users were very consistent. Nurses universally listed persons with addictive disorders, those with personality disorders, and those brought in by police (forensic cases) as the frequent users most commonly encountered. Although a sizable proportion of service seekers had a diagnosis of psychosis, it was mentioned inconsistently as a characteristic of frequent users, and participants stated that psychosis was a more modest contributor than other factors to the phenomenon of frequent use. Physicians did not list the same groups as nurses; they spoke more of the social issues that were evident in the service seekers’ explanations of their complaints, including family disputes and housing instability.

Perceived Reasons for Frequent Visits

The reasons for frequent use cited by participants were related to seeking comfort and inclusion. “The feeling of going to IMH [the Singapore Institute of Mental Health] is that people understand them, rather than the outside world. Because they know that ‘Oh, we’re patients here. So people know us.’ But in the outside world nobody knows them. If they act strange, they can’t explain to that person—a normal person—that ‘I’m suffering from this disease, or illness.’ They don’t know how to explain to others, so that’s why when they come here, they feel they belong.”

Participants recognized the double impact of stigma, which pushed those with a history of ED visits to return to the ED for the comfort they derived from being in a familiar place. But for infrequent visitors, the stigma of visiting the ED acted as a barrier to service seeking.

Caregivers also played a role in driving patterns of attendance by seeking admission for their kin for respite purposes. “Maybe with family members, I think it’s a bit trickier. You know, that [family members’] intention for bringing patients in is different. They have higher expectations. . . . For example, there’s no clear indication for admission, but families insist she’s unwell and needs to be treated as an inpatient.”

Other reasons for frequent visits listed by participants included missing appointments, running out of medication, seeking controlled substances, and nonemergency concerns.

Impact of Frequent Users on Service Providers

Service seekers who were perceived to be visiting the ED for comfort or because they were experiencing symptoms of a relapse elicited sympathy from service providers, unlike those who were intoxicated or seeking controlled substances or those who had personality disorders and who manipulated the staff by mentioning self-harm to gain admission. Participants questioned the legitimacy of these visits. Such users elicited strong emotional responses of powerlessness and hopelessness. “Personality disorder: they always threaten us. Threatening, they always tell us they want to go and jump . . . all this. And then drug addicts are always asking for admission to the addiction ward, which we can’t do here. Emergency is not detox. But if you tell them, they also won’t understand. . . . They will still come back [to the ED].”

It was difficult for staff to know how best to serve these individuals because of the absence of suitable alternatives. The situation was especially problematic for service seekers who could not be handled in other settings because they verbalized thoughts of self-harm or were intoxicated or aggressive or experienced social issues. “They go out, they drink, then they are shouting . . . screaming. And then the police bring them here. But they are not psychotic. They are drunk. But they walk in, and we have to assess them. When they are intoxicated, we cannot just leave them on the street. . . . But after they sober up, they want to leave. Because they know they have no psychiatric issue.”

The absence of suitable alternatives was frequently related to the absence of social supports. The staff viewed this as an important barrier to offering client-centered care because it limited their options for secure outpatient care. “These are groups of patients that actually come back to us a lot. But we also find it difficult to help this group. Because their support is very lacking. No family member, so how are we going to monitor their progress in the community?”

The staff highlighted their desire to help, frustration at not being able to reliably ensure continuity of care, and uncertainty about the effectiveness of treatments offered to frequent users. However, they also noted the personal emotional toll. “They can become quite irritated with you. . . . They will demand admission. And I will see there is no need. And they come to the point of threatening you. . . . I won’t say you have to be rude. But sometimes you have to be a bit hard. Because you cannot admit everyone. So they get a bit negative. You feel mentally exhausted.”

Discussion

Our goal was to describe frequent users of Singapore’s dedicated psychiatric ED and explore staff perceptions of the reasons for frequent visits. By using qualitative interviews to complement the administrative data, we added a layer of detail that is typically absent in descriptions of frequent users derived from administrative data (19).

The diagnostic groups represented by frequent users of Singapore’s psychiatric ED were similar to those in urban centers in other countries. Personality disorders and psychotic disorders have been consistently reported (5–7,10). Therefore, services developed in other contexts may generalize to service seekers in Singapore (29). However, the high rate of hospital admission after ED visits in our study must be considered. Deinstitutionalization in Singapore has not progressed at the same speed as in other developed countries (30), and thus the medical system retains a high capacity for inpatient treatment. This may indicate that the demand for community services in the past has been low; however, the scarcity of community-based alternatives likely perpetuates high admission rates (31).

Our participants had strong emotional responses to ED visitors who expressed a desire to harm themselves, echoing previous work on nurses’ opinions of individuals whose presenting complaint is a desire to self-harm (14,15). Emotional responses were especially pronounced in regard to service seekers with personality disorders who expressed intentions to self-harm. However, anger toward frequent users, contrary to expectations, was not particularly strong, which may reflect staff’s understanding of the reasons for frequent ED visits, rather than the effect of social desirability during the interview. Participants sympathized with service seekers’ feelings of belonging. Participants worried about the care service seekers would receive in the absence of social support and, as a result, anticipated their return to the ED.

The stigma of mental illness, which may explain the service gap documented in local studies (32), played an important role in study participants’ explanations for frequent visits. Stigmatizing social environments outside the institute and the sense of belongingness within the institute led some service seekers to visit the ED without clear medical indication for admission. This finding is in line with qualitative research conducted in other large urban centers, which highlighted that service users seek reassurance and resolution (4). According to our study participants, frequent users’ desired resolution was admission as a reprieve from the stress of living in the community. In addition, some frequent use was driven by family members seeking respite.

Dealing with frequent users who were intoxicated was a substantial challenge. This finding has been reported in several studies in other countries and is becoming the focus of targeted interventions (33–36). In the United Kingdom and elsewhere, intoxicated persons who visit the ED also commonly have traumatic injuries or a psychiatric illness. In our study, participants reported that these individuals visited the ED because of social problems but not invariably because of psychiatric illness. Because some frequent ED visits can be explained by social problems, expanding medical services will likely treat the symptoms and not address the cause (34).

Although interventions that target frequent users directly may be effective in reducing ED visits (29), reductions may also be accomplished with non-ED interventions, such as patient education (37). Because inadequate social support, owing to caregiver burnout or absence, may be responsible for an important portion of repeat visits, non-ED interventions that support caregivers may be effective. Relying on emergency services to provide respite services is suboptimal (38). Further research is needed to confirm the opinions of service providers and determine how existing continuity-of-care strategies may be adapted to suit the local context and bolster social support.

Conclusions

Qualitative interviews can be used to complement comprehensive administrative data and describe service seekers who present frequently to emergency services. Use of qualitative interviews allowed us to explore perceived reasons for frequent use and to explain why addressing the causes of frequent use may be difficult. Because administrative data may not include information about social causes of frequent use, it is essential to understand the societal pressures that influence the decisions of service providers and service seekers. Ignoring these social factors may lead to erroneous conclusions about which services should be implemented.