Patient Variables Associated with the Therapeutic Use of Seclusion and Restraint in Inpatient Psychiatry



A retrospective chart review of psychiatric inpatient admissions at St. Barnabas Hospital for the year 2006 was undertaken to determine the extent to which certain patient demographic variables were associated with the therapeutic use of seclusion or restraint (S/R) on the acute inpatient psychiatric unit.  Variables of interest were the patient’s age, sex, psychiatric diagnosis (psychotic or not psychotic), type of admission (voluntary or emergency) and race.  It was hypothesized that patients who experience S/R will be younger, carry a psychotic diagnosis (schizophrenia, schizoaffective disorder, psychotic disorder NOS), and be admitted on an emergency legal status to the psychiatric units.  Researchers found that patients who experienced S/R events were significantly younger (p ≤ .001) and more likely to be admitted under emergency legal status (p ≤ .05).  Other variables were not significantly related to the experience of seclusion or restraint.  Suggestions for further research and clinical practice are made.


Monitoring the use of seclusion and restraint (S/R) within psychiatric units is mandated by both state and federal agencies and continued efforts to decrease the use of S/R are widely noted. 1 2 3  While abundant research on S/R is represented in the literature, most articles have focused strictly on prevention efforts and not adequately addressed demographic variables (notably age and diagnosis) of patients experiencing S/R events while on the acute psychiatric inpatient unit. 4

Monitoring of S/R in psychiatry has been an important task to ensure that these interventions are used only when other de-escalation techniques have failed and a patient is in danger of harming himself or others.  S/R interventions are never to be used as a punishment or as a convenience.  While most mental health professionals agree that limiting the use of S/R by implementing other strategies is the best direction to take, 4 5 6 7 8  more recently, some mental health advocates have sought to eliminate the use of S/R altogether. 9 10 11  Currently, methods for de-escalating a patient or patients in a crisis situation include: providing reassurance, medication administration, comforting, verbal limit-setting, separation of individuals, imposing a time-out, an explanation of consequences for the behavior, and the offering of a distraction, such as a snack or beverage.  All Nursing and Security staff members are in-serviced annually on de-escalation techniques.

Existing research supports the use of de-escalation techniques to minimize the use of S/R 1 11 12 and yet, further reductions are sought since S/R events are often experienced as negative events by the patient. 11 13  It was expected that this study could add to the literature on the use of S/R and possibly identify variables for further study, which may assist performance improvement efforts to further reduce the use of S/R as a therapeutic intervention in inpatient psychiatry.

The investigators expected to identify patient variables associated with the use of S/R in acute inpatient psychiatry which could, in turn, inform further performance improvement efforts.  We hypothesized that:

  1. The patients’ age would be predictive of the experience of S/R interventions. Specifically, younger patients would more likely experience S/R interventions.
  2. The patients’ diagnoses (psychotic/not psychotic) would be predictive of the experience of S/R interventions, expecting that psychotic patients would more likely experience S/R events.
  3. Patients’ admission status would be predictive of the experience of S/R interventions.  Specifically, patients with an emergency status would be more likely to experience S/R interventions.



A retrospective chart review of all admissions to the adult acute psychiatric units was conducted for the calendar year 2006.  Those patients who remained on the unit at the time of the analyses were excluded.  Demographic variables were recorded and comparisons made between patients who experienced one or more S/R events and patients who did not. The Department of Psychiatry routinely records data on the use of S/R and this database was utilized for the present study.


Demographics and Sample Characteristics

The sample consisted of all admissions to the adult psychiatric inpatient service at St. Barnabas Hospital during the calendar year 2006 (N = 1039).  Some individual patients were admitted more than once during the year.  Three admissions were excluded as the patients remained hospitalized at the time of these analyses.  The sample contained 596 men and 443 women.  The age of the sample ranged from 18 to 77 years with a mean of 38.25 years (SD = 12.523).  The frequencies of each racial category in the sample as per the admission records were as follows: Black (f = 433), Hispanic (f = 490), Caucasian (f = 52), Indian (f = 6) and Other/Unknown (f = 58).

Psychiatric admissions in 2006 consisted of emergency admissions, in which the patient is admitted without his or her permission because he or she is deemed to be a danger to self or others (f = 534).  There were fewer patients admitted with voluntary admission status, in which the patient requests admission (f = 503).  There were only two involuntary admissions, wherein the patient is involuntarily admitted to psychiatry from a medical unit because he or she is deemed to be a danger to self or others (f = 2).  There were similar numbers of admissions to the two psychiatric units, Kane 2 (f = 526) and Kane 3 (f = 505).  There were many more psychotic diagnoses (f = 816) than non-psychotic (f = 223).  The length of stay in the hospital psychiatric service ranged from 1 to 387 days, with a mean stay of 17.40 days (SD = 28.653).

Comparing Kane 2 and Kane 3

The inpatient psychiatric service at St. Barnabas Hospital consists of two separate units.  In order to justify analyzing all admissions in 2006 across the two units, some analyses were necessary to ensure that the units are similar.  When compared on the average length of stay for a patient, Kane 2 (M = 16.45, SD = 25.328) and Kane 3 (M = 17.91, SD = 31.024) did not differ significantly.  The units did, however, differ slightly on the use of seclusions and restraints.  Kane 2 had significantly more seclusions than Kane 3 (See Table 1).  In addition, Kane 2 had slightly more restraint events and more combined seclusion and restraint events, but neither of these differences was statistically significant (See Table 1).

Seclusion and Restraint

The mean number of restraint events per admission (M = 0.13, SD = 0.661) was similar to the mean number of seclusion events (M = 0.14, SD = 0.705).  The average number of seclusion and restraint events combined per admission was 0.27 (SD = 1.180).  There were more patients with zero seclusion or restraint events (f = 913) than there were with one or more (f = 126), and among patients who did have seclusion or restraint events there were more with only one event (f = 78) than with more than one event (f = 48).

The principal aim of this research project was to identify observable patient characteristics that are predictive of the number of S/R events.  This would allow identification of patients at risk for S/R events.  In order to do this, several variables were identified as possible predictors for a multiple regression analysis.  The variables examined included diagnosis (psychotic or non-psychotic), sex, legal admission status, race, age and length of stay on the unit.  A stepwise entry method was used.  Only two of the original variables investigated (age and length of stay) entered into the regression model.  This two-variable model achieved significance in predicting the number of S/R events (F = 23.570, p ≤ .001), yet accounted for a negligible amount of the variance in number of S/R events (r2 = .044).

Mean comparisons were also conducted to examine any differences between those patients who had S/R events and those who did not.  Once again, age (p ≤ .001) and length of stay (p ≤ .001) were the only variables found to vary significantly between those patients with zero seclusions and restraints and those with one or more.  Patients who were secluded or restrained were younger (M = 34.37, SD = 10.983) on average than those who were not (M = 38.78, SD = 12.634).  They also tended to spend more days on the psychiatric unit (M = 32.00, SD = 48.670) than patients who were not secluded or restrained (M = 15.39, SD = 24.005).

In order to study the group of admissions who were secluded or restrained more than once (n = 48), patients in this group were compared to those who were only secluded or restrained once (n = 78), and those who were never secluded or restrained (n = 913).  Those admissions with more than one S/R event comprised 4.6% of the total sample and yet accounted for 71.8% of the total number of S/R events for admissions in 2006.  When compared on age, those with more than one event (M = 31.54, SD = 11.231) were younger than those who had been secluded or restrained only once (M = 36.12, SD = 10.524, p ≤ .05) as well as those who had never been secluded or restrained (M = 38.78, SD = 12.634, p ≤ .001).  A chi-square test revealed a significant relationship between the category (zero, one or several seclusion and restraint events) and the admission status of the patient (χ2 = 10.969, p ≤ 0.05).  Examination of the frequencies and expected counts revealed that patients with one or more than one event were more likely to be admitted with an emergency status rather than a voluntary status.  Length of stay was much longer (p ≤ 0.001) for those patients who had more than one event (M = 34.58, SD = 30.561), than for those with none (M = 15.39, SD = 24.005).  Those patients with several events (M = 34.58, SD = 30.561) did not significantly differ on length of stay from those with only one (M = 30.41, SD = 57.172).


The present study sought to identify trends in the implementation of seclusions and restraints (S/R) based on patient characteristics.  Such an investigation was deemed necessary both as a means of ensuring the equitable use of S/R interventions and as a means of discovering patterns that would help to focus interventions to reduce the necessity of S/R interventions.

The first hypothesis proposed at the outset of this study was that there exists a relationship between age and S/R events with younger patients experiencing more events.  This hypothesis was supported, as patients who experienced S/R events were significantly younger than those who did not experience any.  Additionally, age proved to be a significant predictor of the number of S/R events in a multiple regression analysis.  In fact, the age of the patient was the most powerful predictor of S/R events examined in this study.

The second hypothesis was that there exists a relationship between diagnosis and the experience of S/R events, specifically that psychotic patients would be more likely to experience S/R events.  This hypothesis was not supported by the present data.

The third hypothesis proposed was that there would exist a relationship between legal admission status and the number of S/R events.  Specifically, it was proposed that those patients with an emergency status would be more likely than those with a voluntary status to experience S/R events.  This hypothesis was supported, to an extent, by the present data.  A chi-square analysis revealed that those patients who experienced one or more than one S/R event were more likely to be admitted with an emergency status.

It became apparent in analyzing the S/R data for admissions from 2006 that there was a small subset of admissions that accounted for the majority of S/R events.  This being the case, it makes practical sense to highlight the various characteristics of these patients so that similar patients may be more readily and effectively targeted for alternate interventions in the future.

The subset of patients with multiple S/R events was substantially younger than the patients who had only one S/R event and the patients with zero S/R events.  This pattern suggests that younger patients are at higher risk for repeated S/R events.  It was also found that patients with an emergency status were more likely to have more than one S/R event, suggesting that those patients with such a legal status have a higher risk of S/R events.  Finally, length of stay was found to be much longer for those patients with more than one event than for those with none.  This relationship could be indicative of the fact that the risk for more S/R events increases as the patient remains on the unit for longer.  It is more likely, however, that those patients who are secluded and/or restrained are likely to be more severely ill and exhibit more dangerous behavior and are therefore more difficult to discharge to community placement, resulting in a longer length of stay.

These findings suggest that high-risk patients can be identified, to some extent, by certain patient characteristics.  Those patients who are younger and who are admitted on an emergency status should be targeted for alternate and preventive interventions.  It would also be wise to immediately target patients for alternate interventions after the first S/R event so that further events can be prevented.

The methods currently employed at St. Barnabas Hospital as alternative interventions include providing reassurance, medication administration, comforting, verbal limit-setting, separation of individuals, imposing a time-out, an explanation of consequences for the behavior, and the offering of a distraction, such as a snack or beverage.  The psychiatric staff also engages in a multidisciplinary review of any patient who has experienced more than one recent S/R event.  Behavior intervention plans are often implemented as a result.

It is important to note several substantial limitations of the present study.  First, this study was conducted at a single site, and therefore draws on a specific local population.  The local population served by St. Barnabas Hospital is that of the South Bronx in New York City, which may differ in significant ways from the general population.  This fact limits the generalizability of the present results to other settings and populations.  Second, because the present data was organized by admissions and not by patient, some data points represent the same patient during different admissions.  This fact may have exaggerated or minimized some findings.  Third, the utilization of billing codes (DRG codes) to provide diagnostic information may have oversimplified patient diagnoses.

The present study has generated some promising directions for future research and clinical practice.  First, in culling the relevant data from the patients’ medical records, it became apparent that certain improvements in data collection could be made.  Data is consistently recorded including the time, date and duration of an S/R event and the reason for the intervention; it would be beneficial to record even more data.  Specifically, it would be helpful to record the circumstances preceding and coinciding with the incident that precipitated the S/R event.  It would also be helpful to record which staff members and/or patients were present at the time of the incident.  Such information would facilitate functional behavior analyses that could reduce or eliminate the use of S/R for a given patient.

Another direction for future research that may yield some valuable information would be to conduct extended debriefing interviews with patients who experienced S/R events immediately following the event.  Such interviews could probe for the patients’ own understanding of what happened and which interventions might be of help to that patient in avoiding future S/R events.


Table 1

Mean number of seclusion, restraint and combined events per admission on Kane 2 and Kane 3.

Seclusions* Restraints Seclusions and restraints
Kane 2 0.19 (SD = 0.930) 0.13 (SD = 0.690) 0.32 (SD = 1.435)
Kane 3 0.08 (SD = 0.327) 0.12 (SD = 0.592) 0.19 (SD = 0.793)

*Denotes a result significant difference between the two units at or beyond the α = .05 level.

Table 2

Mean age and length of stay by number of seclusion or restraint (S/R) events.

Age (years) 38.78 (SD = 12.634) 36.12 (SD = 10.524) 31.54 (SD = 11.231)
Length of Stay (days) 15.39(SD = 24.005) 30.41(SD = 57.172) 34.58 (SD = 30.561)


  1. Smith GM, Davis RH, Bixler EO, Lin H-M, Altenor A, Altenor RJ, Hardenstine DB, Kopchick GA. Pennsylvania state hospital system’s seclusion and restraint reduction program. Psychiatric Services 2005; 56: 1115-1122.

  2. Fisher WA. Restraint and seclusion: A review of the literature. American Journal of Psychiatry 1994; 151: 1584-1591.

  3. Phillips P, Nasr SJ. Seclusion and restraint and prediction of violence. American Journal of Psychiatry 1983; 140: 229-232.

  4. Tunde-Ayinmode M, Little J. (2004). Use of seclusion in the acute inpatient unit. Australian Psychiatry 2004; 12: 347-351.

  5. Liberman RP. Elimination of seclusion and restraint: A reasonable goal? Psychiatric Services 2006; 57: 576.

  6. Herzog A. Letter to the editor. Psychiatric Services 2005; 56: 1622.

  7. LeBel J, Goldstein R. The economic cost of using restraint and the value added by restraint reduction or elimination. Psychiatric Services 2005; 56: 1109-1114.

  8. Glover RW. Reducing the use of seclusion and restraint: A NASMPD priority. Psychiatric Services 2005; 56: 1141-1142.

  9. Busch AB. Introduction to the special section. Psychiatric Services 2005; 56: 1104.

  10. Curie CG. Special section on seclusion and restraint: Commentary: SAMHSA’s commitment to eliminating the use of seclusion and restraint. Psychiatric Services 2005; 56: 1139-1143.

  11. Robins CS, Sauvageot JA, Cusack KJ, Suffoletta-Marierle S, Frueh BC.Consumers’ perceptions of negative experiences and the “sanctuary harm” in psychiatric settings. Psychiatric Services 2005; 56: 1134-1140.

  12. Donat DC. Encouraging alternatives to seclusion, restraint, and reliance on PRN drugs in a public psychiatric hospital. Psychiatric Services 2005; 56: 1105-1108.

  13. rueh BC, Knapp RG, Cusack KJ, Grubaugh AL Sauvageot JA, Cousins VC, Yim E, Robins CS, Monnier J, Heirs FG. Patients’ reports of traumatic or harmful experiences within the psychiatric setting. Psychiatric Services 2005; 56: 1123-1133.