Figures
Abstract
Background
This study contrasted annual rates of difficult behaviours in emergency departments among cohorts of individuals who were homeless and low-income housed and examined predictors of these events.
Methods
Interviews in 1999 with men who were chronically homeless with drinking problems (CHDP) (n = 50), men from the general homeless population (GH) (n = 61), and men residing in low-income housing (LIH) (n = 58) were linked to catchment area emergency department records (n = 2817) from 1994 to 1999. Interview and hospital data were linked to measures of difficult behaviours.
Results
Among the CHDP group, annual rates of visits with difficult behaviours were 5.46; this was 13.4 (95% CI 10.3–16.5) and 14.3 (95% CI 11.2–17.3) times higher than the GH and LIH groups. Difficult behaviour incidents included physical violence, verbal abuse, uncooperativeness, drug seeking, difficult histories and security involvement. Difficult behaviours made up 57.54% (95% CI 55.43–59.65%), 24% (95% CI 19–29%), and 20% (95% CI 16–24%) of CHDP, GH and LIH visits. Among GH and LIH groups, 87% to 95% were never involved in verbal abuse or violence. Intoxication increased all difficult behaviours while decreasing drug seeking and leaving without being seen. Verbal abuse and violence were less likely among those housed, with odds ratios of 0.24 (0.08, 0.72) and 0.32 (0.15, 0.69), respectively.
Conclusions
Violence and difficult behaviours are much higher among chronically homeless men with drinking problems than general homeless and low-income housed populations. They are concentrated among subgroups of individuals. Intoxication is the strongest predictor of difficult behaviour incidents.
Citation: Svoboda T (2015) Difficult Behaviors in the Emergency Department: A Cohort Study of Housed, Homeless and Alcohol Dependent Individuals. PLoS ONE 10(4): e0124528. https://doi.org/10.1371/journal.pone.0124528
Academic Editor: Vineet Gupta, University of California San Diego, UNITED STATES
Received: August 5, 2014; Accepted: March 15, 2015; Published: April 28, 2015
Copyright: © 2015 Tomislav Svoboda. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper.
Funding: This work was supported by grants from the Central East Health Information Partnership, the former City of Toronto Shelter, Housing and Support Division, Physician Services Incorporated, and St. Michael's Hospital Research Department as well as a post-doctoral fellowship from the Canadian Institutes for Health Research to Tomislav Svoboda.
Competing interests: The author has declared that no competing interests exist.
Introduction
Individuals who are homeless use emergency departments more heavily than the general population, frequently accessing them as sources of primary care.[1–6] Barriers to accessing care in such settings has been described in the general homeless population and the role of difficult behaviours or interactions with staff has been raised.[7,8] Currently the only available data about homelessness and difficult behaviours shows that homelessness and alcohol use are risk factors.[9,10] Individuals who are chronically homeless and heavy alcohol users form a recognizable subset of the homeless population and are anecdotally known to experience more difficulties with behaviours.[11–15] Terms such as “public inebriates,” “chronic inebriates,” and “frequent fliers” have been used in the medical and lay literature and may reflect a level of visible importance for this subgroup that has led to stigmatization rather than understanding.[16,17] Despite their visibility, policy relevance and need for understanding, information is generally anecdotal with few epidemiological studies of this group. Misunderstandings or misplaced opinions of difficult behaviours in these populations may serve to further stigmatize and marginalize rather than help these already vulnerable groups. Difficult behaviours generally refer to physical or verbal abuse or violence that might discourage staff from providing care due to the threat these behaviours may pose.[18–21] Absent or less well described are non-threatening behaviours that still pose significant barriers to complete assessments, treatments or proper discharge and follow-up plans. These might include difficult histories or leaving before being seen. This retrospective cohort analysis of emergency department records was carried out to answer the following clinically and policy relevant questions: 1) what difficult behaviours are reported by care providers? 2) what proportion of individuals engage in difficult behaviours? 3) how frequently? and 4) what are the risk factors and distribution of different difficult behaviours? These questions were explored in three policy relevant groups: those who are chronically homeless and heavy users of alcohol, those who are in the general homeless population and those who live in low-income housing.
Methods
Design
This study analyzed data from a larger study evaluating the impact of a harm reduction program on individuals who were chronically homeless with severe drinking problems (CHDP) using individuals within low income housing (LIH) and general homeless (GH) populations as non-equivalent controls.[22] A retrospective cohort analysis was used. This approach is useful when resource constraints and characteristics of a study population make it difficult to follow cohorts over time. For this analysis, rates of difficult behaviours found in emergency department records during a six year period prior to subject interviews were compared among the three groups of men.
Ethics Statement
Independent research ethics review and approval of the larger study was obtained from ethics review committees in the University of Toronto Office of Research Services, St. Michael's Hospital, the University Health Network, and St. Joseph’s Health Centre. Written informed consent was obtained from all participating study subjects.
Setting and Participants
All adult men were invited to participate from a 60 bed shelter-based alcohol harm reduction program (CHDP) designed for chronically homeless men unable to abstain from alcohol and rejected from typical shelters due to their drinking patterns; a 250 bed men’s shelter for the general homeless population (GH) and three low income housing sites (LIH). The subjects were recruited in April and May 1999. Data from hospital records were abstracted for the period January 1, 1994, to April 14, 1999 inclusive. The two control group samples were identified using systematic sampling with random start, a type of sampling which approximates a simple random sample. Every third individual entering the participating sites during blocks of time spanning the whole time that individuals enter the sites were invited to meet with an interviewer. Harm reduction program recruits not meeting alcohol dependence criteria were excluded from the analysis (n = 9) as they represented residents sheltered for problems unrelated to alcohol use.
Free and informed consent for vulnerable subjects
Recruitment methods to obtain free and informed consent for study participation included recruiter discussion, written information on the consent form, researcher availability to answer questions any time during the study, recruiter training, and competence assessments where recruiters asked subjects to express in their own words 1) the purpose of the study, 2) that only researchers would access their records and 3) how they could withdraw from the study at any time. Participation was hidden from agency staff and express statements were made to clients that their care would not be disadvantaged by non-participation. Additional techniques shown to improve comprehension included: using a lower reading level [23], using larger typeface [24], quizzing [25] and having multiple individuals provide the information.[26]
Data sources and variables
Subject records in all 9 downtown Toronto hospitals were reviewed and validated and study-specific instruments determined subjects’ alcohol dependence status, history of homelessness, and length of time residing in the study hospitals’ catchment area. Data from physician, nurse, and paramedic elements of emergency department records were abstracted using computer forms and pre-defined coding lists and definitions. Abstracted data included: 1) intoxication; and head injury 2) behaviours including: violent behaviour (aggressive, threatening, abusive, etc.); verbal abuse; uncooperative; drug seeking; difficult historian; leaving against medical advice; leaving without being seen by physician; and leaving before discharge and 3) interventions including: presence or involvement of security staff during medical encounter; escort out by security staff; and use of restraints. Behaviours were rated as severe if they included related qualifiers (e.g. “very,” “++,” etc.); a raised voice or threats to staff; or if security or restraints were involved.
The rates of emergency department visits were determined using the total group observation time. Subjects with observation times greater than 6 months were used to determine what proportion of subjects accounted for the majority of annual difficult incidents that occurred for each group.
Reducing Sources of Bias
CHDP clients were approached up to three separate times to reduce the chances of non-selection among more severely intoxicated clients. To maximize response rate in all groups we provided a $20 honorarium for completed interviews, and utilized staff familiar and trusted by clients to introduce recruitment staff using uniform scripts. Standardized training, scripts and probes, supervision by the principal investigator and blinding to group membership reduced interviewer measurement biases. To reduce hospital chart abstraction biases and increase reliability, the abstractors were blind to study group membership, and each record was reviewed twice by two independent teams. To reduce missing hospital records with poorly recorded identifiers due to intoxicated presentations, records were hand-searched by two independent study staff, then further probabilistically matched to the interview subjects. Multiple identifiers were collected including health insurance numbers, and alias names. Data from the time subjects reported living outside the hospital catchment areas, were excluded. Power calculations for detecting changes in social service use in the original study determined sample sizes.
Analyses
Confidence intervals for proportions of visits involving difficult behaviors used a normal approximation of the binomial distribution.[27] Variables predicting the occurrence of difficult behaviours were determined using multivariate logistic regression, controlling for individual subjects and using a binary distribution for the outcome (SAS 9.0 Glimmix Procedure, 2010).
Results
Of the 227 individuals approached for the study, 178 (78%) agreed to participate. In the CHDP group, 9 did not meet alcohol dependence criteria and were excluded from further analysis. Table 1 compares demographic, mental health measures and emergency department visit rates among the three groups. The CHDP group tended to be older, homeless longer and have more emergency department visits than the other groups.
The study spanned 641 subject years and 2817 emergency department records were reviewed. The absolute annual crude rate of emergency department visits with any difficult behaviour was 5.46 (95% CI 5.16–5.77) for the CHDP group; this was 13.4 (95% CI 10.3–16.5) times higher than the GH group with a rate of 0.408 (95% CI 0.315–0.500) and 14.3 (95% CI 11.2–17.3) times higher than the LIH group with a rate of 0.383 (95% CI 0.304–0.462). More than half (0.575, 95% CI 0.554–0.597) of the visits by the CHDP group included difficult behaviours. This was more than double the proportion for the GH group (0.24, 95% CI 0.19–0.29) and the LIH group (0.20, 95% CI 0.16–0.24), who were not significantly different (p = 0.18).
Table 2 shows a similar proportional breakdown of difficult behaviours among the three groups. There were more verbal abuse incidents and security staff interventions in the CHDP group. Uncooperative behaviours within the medical encounter were higher in the CHDP group. In the groups combined, severe difficult behaviours were 18.4% (95% CI 16.4–20.4%) of all difficult behaviours, and severe violent behaviours were 43.2% (95% CI 36.8–49.5%) of all violent behaviours; the three groups were not significantly different from each other.
Table 3 summarizes what proportion of subjects accounted for most of the difficult behaviour incidents. In all types of incidents the proportion was much higher in the CHDP group. The median proportion of visits with any difficult behaviour among subjects in the CHDP group was 0.50 (IQR 0.35–0.59). This was significantly higher than for the GH and LIH groups, which had medians of 0.00 (IQR 0.0–0.23) and 0.00 (IQR 0.0–0.33), respectively.
Table 4 shows the proportion of subjects who fell within each of four different ranges of difficult behaviour rates. For example, 38% of those in the CHDP group were noted to be drug seeking, with 2% drug seeking more than half of the time, while in the GH and LIH groups only 5% of subjects were ever involved in drug-seeking behaviours and these occurred in less than 50% of their visits.
In Table 5, the multivariate modeling of all groups combined shows that difficult behaviours were significantly associated with intoxication. Subjects were intoxicated with alcohol alone 85.8% (95% CI 84.0–87.7%) of the time while the remaining intoxication involved other substances (9.5%, 95% CI 8.0–11%). Types of intoxication were similar across groups. Leaving without being seen and drug seeking were the only difficult behaviours that were less likely among those who were intoxicated; the remaining difficult behaviours were more likely among those intoxicated. The proportion of difficult behaviours in which intoxication was implicated was significantly greater in the CHDP group for verbal abuse (77%, 95% CI 72–81%, p = 0.01), physical violence (81%, 95% CI 75–85, p = 0.02), and leaving without being seen (49%, 95% CI 43–54%, p = 0.001).
Table 6 provides difficult behaviour descriptions found in the emergency records.
Interpretation
This is the first cohort study examining difficult behaviours in emergency departments among homeless, and low-income housed individuals. Difficult behaviours occurred in a quarter to a half of all emergency department encounters in these groups. Uncooperative behaviours made up three-quarters of difficult behaviours across all groups, while up to a fifth of difficult behaviours included violence. Nearly half of violent behaviours were severe, resulting in restraint or security interventions.
This study challenges conclusions that might be drawn from other studies suggesting that being homeless is an independant risk factor for difficult behaviours in the emergency department. [10,18,19,28,29] These studies considered emergency room records as the units of analysis in a single point in time. In contrast, this study followed individuals over time to examine their rates of presenting with difficult behaviours. This approach suggests that the vast majority of individuals in these populations do not exhibit difficult behaviours and when they do, it is because they are presenting intoxicated or otherwise acutely brain injured rather than being homeless or having a drinking problem. Rates of difficult behaviours were not found to be significantly different between the GH population and the LIH population, while the rates among the frequently intoxicated CHDP population were 14.3 times greater. Among the GH and LIH groups, 87% and 90%, respectively, were never involved in any recorded verbal abuse, while 90% and 95%, respectively, did not present with violent behaviours. In the CHDP group, more individuals exhibited difficult behaviours; however, they occurred in less than 10% of visits for 56% and 72% of subjects and less than half the time for the rest (Table 4). Out of 17 different measures examined, intoxication was a moderate to strong predictor of 11. Being identified as a problem drinker or being housed was significant for 1 or two. Membership in the chronically homeless problem drinking group was not a significant predictor of difficult behaviours when intoxication was taken into account. These types of variable interactions, suggest that the housing effect may be confounded by lower intoxicated presentation rates in the housed group. This analysis cannot resolve whether housing reduces substance use or whether reduced substance use makes it easier to be housed as the effects of changing housing status were not evaluated. Growing evidence that housing interventions lead to reductions in hospital utilization support the first interpretation.[30] Studies of difficult behaviours in emergency departments suggest that alcohol is implicated in 50% to 77% of abuse-of-staff incidents and 3% to 11% of incidents of leaving without being seen, however tests of association were not made in these studies. [18,20,31–33]
Multivariate analyses in this study suggest that being unconscious at some point during the visit is a strong predictor of being restrained, being verbally abusive, leaving against medical advice, and leaving before discharge. Unconsciousness was predictive of difficult behaviours independent of intoxication; this may suggest that both are the effects of an injured brain. This association with loss of consciousness was not found in the studies reviewed.[20,31–33] Emergency department resources may not be sufficient to meet the needs of many patients with severe intoxication or other injuries compounded by severe psychosocial problems and their resultant difficult behaviours.[34] Table 6 gives qualitative evidence suggestive of the how challenging and possibly inappropriate the current approach to care for these vulnerable populations can be for both the care providers and the patients. Difficult behaviours range from “refusing BP” to “setting bedsheets on fire” to being “put in 4 point restraint for staff safety”.
This study has several limitations. Difficult behaviours were likely undercounted to varying degrees due to recording bias. [20] Fernandes’ study of a comparable inner city hospital suggests a rate of 330 assaults in a year based on staff recall. [19] This study suggests 98. Over-recall in Fernandes’s study and under-recording in the current study may provide upper and lower bounds to the actual number. Other potential sources of undercount include exclusion of individuals from participation due to severe mental illness, severe intoxication, or not appearing as scheduled as well as including individuals in a shelter based harm reduction program that aimed to improve behaviours. The results of the current study may not be generalizable to other jurisdictions, subgroups, such as women and youth not included in this study, and present circumstances due to the age of the data. Older and more recent studies of difficult behaviours during this time suggestion that there has not been a change.[7–10] Some authors have suggested that difficult behaviours may increase as resources become more stretched.[34]
Intervention studies aimed at assisting those who are homeless especially those who are chronically homeless and heavy alcohol users should be a high priority. Approaches that reduce inappropriate and difficult emergency department visits, especially for those who are likely to present intoxicated, are needed. They are beginning to appear.[12,35,36] These studies however do not address or describe the impact of interventions on difficult behaviors. This study raises important topics for further research and program development, especially for vulnerable subgroups such as chronically homeless, alcohol-dependent men. These include the further evaluation and development of promising community and hospital-based interventions for those with severe problems related to heavy alcohol use, such as housing first, staff sensitivity and de-escalation skills training, compassionate care, harm reduction, and abstinence-based programs. These interventions may both reduce visits to the emergency department and rates of difficult behaviours.[22,30,35–37] Better understanding of staff, client, environmental, and care approach differences among emergency departments and within the community could also provide promising results.
Acknowledgments
This work was supported by grants from the Central East Health Information Partnership; the former City of Toronto Shelter, Housing and Support Division; Physician Services Incorporated; and St. Michael’s Hospital Research Department, as well as a post-doctoral fellowship from the Canadian Institutes for Health Research to Dr. Svoboda.
Author Contributions
Conceived and designed the experiments: TS. Analyzed the data: TS. Contributed reagents/materials/analysis tools: TS. Wrote the paper: TS.
References
- 1. Hwang SW. Homelessness and health. Canadian Medical Association Journal. 2001;164: 229–233. pmid:11332321
- 2. Mandelberg JH, Kuhn RE, Kohn MA. Epidemiologic analysis of an urban, public emergency department’s frequent users. Academic Emergency Medicine. 2000;7: 637–646. pmid:10905642
- 3. Kushel MB, Vittinghoff E, Haas JS. Factors associated with the health care utilization of homeless persons. JAMA. 2001;285: 200–206. pmid:11176814
- 4. Lissauer T, Richman S, Tempia M, Jenkins S, Taylor B. Influence of homelessness on acute admissions to hospital 763. Archives of Disease in Childhood. 1993;69: 423–429. pmid:8259871
- 5. Padgett DK, Struening EL, Andrews H, Pittman J. Predictors of emergency room use by homeless adults in New York City: the influence of predisposing, enabling and need factors. Social Science & Medicine. 1995;41: 547–56.
- 6. Pilossoph-Gelb S, Mower WR, Ajaelo I, Yang SC. Psychosocial difficulties and emergency department use 451. Academic Emergency Medicine. 1997;4: 589–92. pmid:9189192
- 7.
Mason K, Khandor E. The Street Health Report 2007 [Internet]. Toronto: Street Health; 2007 Sep. Available: c:\references\Street Health Report 2007.pdf
- 8.
Ambrosio E, Crowe C, Hardill K. The Street Health Report [Internet]. 1992. Available: c:\References\The Street Health Report.pdf
- 9. Mayer D. Refusal of care and discharging “difficult” patients from the emergency department. Ann Emerg Med. 1990;19: 1436–1446. pmid:2240758
- 10. McNiel DE, Binder RL. Psychiatric emergency service use and homelessness, mental disorder, and violence. Psychiatr Serv. 2005;56: 699–704. pmid:15939946
- 11. Giesbrecht NA, Giffen PJ, Lambert S, Oki G. Changes in the social control of skid row inebriates in Toronto: assessments by skid row informants. CanJPublic Health. 1981;Revue Canadienne de Sante Publique. 72: 101–104. pmid:6165452
- 12. Thornquist L, Biros M, Olander R, Sterner S. Health care utilization of chronic inebriates. Academic Emergency Medicine. 2002;9: 300–308. pmid:11927454
- 13. Leonard EC. The treatment of Philadelphia inebriates. From temperance reform to “secret cure” 60. American Journal on Addictions. 1997;6: 1–10. pmid:9097866
- 14. Cox GB, Walker RD, Freng SA, Short BA, Meijer L, Gilchrist L. Outcome of a controlled trial of the effectiveness of intensive case management for chronic public inebriates. Journal of Studies on Alcohol. 1998;59: 523–532. pmid:9718104
- 15. Greene J. Serial Inebriate Programs: What to do About Homeless Alcoholics in the Emergency Department. Annals of Emergency Medicine. 2007;49: 792–793.
- 16. Fulde GWO. The Homeless and the Emergency Department: a Special Relationship. The Medical Journal of Australia. 2003;179: 651–652. pmid:14636146
- 17.
CLEGG and ASSOCIATES. Recommended Array of Services of Chronic Public Inebriates. Seattle, WA: Seattle-King County Department of Public Health; 1993.
- 18. Mahoney BS. The extent, nature, and response to victimization of emergency nurses in Pennsylvania. J Emerg Nurs. 1991;17: 282–291; discussion 292–294. pmid:1921065
- 19. Fernandes CMB, Bouthillette F, Raboud JM, Bullock L, Moore CF, Christenson JM, et al. Violence in the emergency department: a survey of health care workers. CMAJ. 1999;161: 1245–1248. pmid:10584084
- 20. Vardy J, Mansbridge C, Ireland A. Are emergency department staffs’ perceptions about the inappropriate use of ambulances, alcohol intoxication, verbal abuse and violence accurate? Emerg Med J. 2009;26: 164–168. pmid:19234002
- 21. James A, Madeley R, Dove A. Violence and aggression in the emergency department. Emerg Med J. 2006;23: 431–434. pmid:16714500
- 22.
Svoboda TJ. Measuring the “Reduction” in a Harm Reduction Program for Homeless Men with Problems Related to Alcohol and Severe Mental Illness. University of Toronto. 2005.
- 23. Young D, Hooker D, Freeberg F. Informed consent documents: Increasing comprehension by reducing reading level. IRB. 1990;12: 1–5. pmid:11651265
- 24. Taub H, Baker M, Kline G, Sturr J. Comprehension of informed consent information by young-old through old-old volunteers. Exp Aging Res. 1987;13: 173–178. pmid:3505870
- 25. Taub HA, Baker MT, Kline G, Sturr J. Effect of repeated testing upon comprehension of informed consent materials by elderly volunteers. Exp Aging Res. 1983;9: 135–138. pmid:6641771
- 26. Tankanow R, Sweet B, Weiskopf J. Patients’ perceived understanding of informed consent in investigational drug studies. Am J Hosp Pharm. 1992;49: 633–635. pmid:1598945
- 27. Agresti A, Coull BA. Approximate Is Better than “Exact” for Interval Estimation of Binomial Proportions. The American Statistician. 1998;52: 119–126.
- 28. Pane GA, Winiarski AM, Salness KA. Aggression directed toward emergency department staff at a university teaching hospital. Ann Emerg Med. 1991;20: 283–286. pmid:1996823
- 29. Crilly J, Chaboyer W, Creedy D. Violence towards emergency department nurses by patients. Accid Emerg Nurs. 2004;12: 67–73. pmid:15041007
- 30. Parker D. Housing as an Intervention on Hospital Use: Access among Chronically Homeless Persons with Disabilities. J Urban Health. 2010;87: 912–919. pmid:21125341
- 31. Henson VL, Vickery DS. Patient self discharge from the emergency department: who is at risk? Emerg Med J. 2005;22: 499–501. pmid:15983086
- 32. Mohsin M, Forero R, Ieraci S, Bauman AE, Young L, Santiano N. A population follow-up study of patients who left an emergency department without being seen by a medical officer. Emerg Med J. 2007;24: 175–179. pmid:17351221
- 33. Goodacre S, Webster A. Who waits longest in the emergency department and who leaves without being seen? Emerg Med J. 2005;22: 93–96. pmid:15662055
- 34. Morrison LJ. Abuse of emergency department workers: an inherent career risk or a barometer of the evolving health care system? CMAJ. 1999;161: 1262–1263. pmid:10584087
- 35. Larimer ME, Malone DK, Garner MD, Atkins DC, Burlingham B, Lonczak HS, et al. Health care and public service use and costs before and after provision of housing for chronically homeless persons with severe alcohol problems. JAMA. 2009;301: 1349–1357. pmid:19336710
- 36. Podymow T, Turnbull J, Coyle D, Yetisir E, Wells G, Podymow T, et al. Shelter-based managed alcohol administration to chronically homeless people addicted to alcohol.[see comment]. CMAJ Canadian Medical Association Journal. 2006;174: 45–49. pmid:16389236
- 37. Redelmeier DA, Molin JP, Tibshirani RJ. A randomised trial of compassionate care for the homeless in an emergency department [see comments] 568. Lancet. 1131;345: 1131–4. pmid:7723543