Effects of Ending the Use of Seclusion and Mechanical Restraint in the Pennsylvania State Hospital System, 2011–2020

The Pennsylvania Department of Human Services, Office of Mental Health and Substance Abuse Services (OMHSAS), was an early pioneer in reducing and ending the use of seclusion and restraint in its state hospitals and forensic centers (1–4). This renaissance in the care of people with mental illness has spread worldwide, with seclusion and restraint no longer viewed as treatments, but rather as treatment failures (1–10). Since 1990, members of past and present OMHSAS leadership teams have monitored the impact of seclusion and restraint reduction on safety and quality of care (2–4).

Despite concerns that ending the use of seclusion and restraint in mental health facilities would lead to increased assaults and injuries, the experience has been the opposite (5–9). Even though successful treatment approaches have been proven safe and effective in reducing seclusion and restraint, these containment procedures continue to be used at numerous facilities worldwide (10–14). Additionally, the use of containment procedures and forced medication continues to raise concerns about patient and staff safety worldwide (6, 15–18).

People who have experienced seclusion, restraint, or forced administration of medication perceive these procedures to be disrespectful and harmful, to violate human rights, and to compromise trust and therapeutic relationships (6, 9). Furthermore, they feel that appropriate and supportive interventions would have averted the use of these coercive procedures (6). Worldwide, interest has grown in reducing and ending the use of seclusion and restraint in all inpatient settings (10). Interventions that have been effective in avoiding the use of coercion require the commitment of all team members (19, 20).

The objective of this study was to assess ongoing efforts of ending the use of seclusion and mechanical restraint in the Pennsylvania State Hospital System and to evaluate the safety of this approach, which has been used for >30 years.

Methods

Study Setting

This study examined data from adults ages ≥18 years civilly committed to the Clarks Summit, Danville, Norristown, Torrance, Wernersville, and Warren State Hospitals in Pennsylvania (N=3,989) and those criminally committed to the Regional Forensic Centers at Norristown and Torrance State Hospitals (N=3,548) between January 1, 2011, and December 30, 2020. These facilities provide acute and subacute levels of care to people in their geographic service areas. During the decade under study, the system provided about 4.82 million days of care (civil, N=3.93 million; forensic, N=890,000) within 76 treatment units (civil, N=62; forensic, N=14).

Over the past 10 years, the patient census in the civil hospitals decreased from 1,249 in January 2011 to 915 in December 2020, a 27% (N=334) decrease. During the same period, the census in the forensic units increased by 21% (N=43), from 208 in 2011 to 251 in 2020. No closures or consolidations occurred during this period.

The Pennsylvania State Hospital System provides services by using a medical model led by a psychiatrist. Ideally, a civil hospital or forensic unit has a psychiatrist and a physician (shared among units), four direct care workers, two licensed nurses, a social worker, and program staff. Patient acuity, capacity, and census affect these ratios. The civil hospitals offer a treatment mall program based on an individual recovery approach to care and services (21).

Forensic centers provide a similar treatment approach, using a medical model led by a psychiatrist, within a secure treatment setting that is primarily focused on stabilization, competency assessment, and restoration. Both systems of care provide a full array of treatment and support services in partnership with families as well as county and community providers (22, 23).

People admitted to a civil hospital were previously served in an acute inpatient care setting. All admissions are involuntary and, according to the Mental Health Procedures Act, can last up to 180 days and may be renewed on the basis of the individual’s psychiatric condition (24).

Forensic admissions are also regulated by the Mental Health Procedures Act and arise primarily from county-level correctional facilities (23, 24). Most people admitted to a forensic center are under the jurisdiction of the criminal courts and have been determined to have a mental illness and to be in need of stabilization. Others may be exhibiting symptoms of a mental illness, and a determination may be needed about whether the admitted person is competent to stand trial. In addition, the person may have been tried for their crime, and the court may request assistance or “aid in sentencing” of the individual. This procedure assists the court in determining the best setting for people who may experience further exacerbation of symptomatology while incarcerated.

State policy on the use of containment procedures (i.e., seclusion and restraint) did not change during this study. Physical restraint is defined as any hands-on control or containment of a person. Mechanical restraint involves the use of any devices to control or contain a person. The hospital system limited the use of mechanical restraint to two or four body locations with soft Velcro-type restraints only. The present policy limits patients’ time in physical restraint to a maximum of 10 minutes and prohibits restraint in the prone position. Mechanical restraint and seclusion are limited to 30-minute applications, which can be renewed with a physician’s order (25). Transfer restraints, used to move people outside the forensic centers, were not part of this study, per national reporting standards.

Starting in 2017, the civil hospitals and forensic centers, independently and at different times, initiated local policy changes to reduce the maximum allowable time in physical restraint from 10 to 3 minutes (26, 27). These changes were made with the support of new training protocols and amid growing concern that the prolonged use of physical restraint, in any position, was dangerous for everyone involved.

Data Description and Analysis

Data on 68,153 incidents of assault, aggression, and self-injurious behavior (civil, N=55,560; forensic, N=12,593) reported through the hospital risk management system during the decade under study were entered into a database and included all episodes of mechanical restraint, physical restraint, and seclusion. These data were structured by using national reporting standards and were linked to the patient information system to evaluate diagnostic and demographic differences in the use of these procedures (28).

Electrical shocking devices and chemical sprays are not permitted for use in any Pennsylvania facility. OMHSAS leadership reviewed this study and exempted it from institutional review board approval because no human testing was involved.

Rates of use and counts for each containment procedure and all safety measures were reported for comparative analysis and to evaluate changes during the study. The chosen safety measures—patient-to-patient and patient-to-staff assaults and incidents of aggression and self-injurious behavior—were selected because they were the leading causes for the use of containment procedures over the past 20 years. Additionally, during this 10-year study, these statistics were publicly reported monthly via State Hospital Risk Management Summary reports (29). The incidents reported each month by hospital and service were used as the numerator, and the monthly days of care served as the denominator. The result was multiplied by 1,000 to report incidents per 1,000 days.

A repeated measures quasi-experimental group design was used. SPSS, version 24, was used to conduct correlation and regression analyses on the variables and to assess the effect of the declining use and duration of containment procedures on all safety measures over the 10-year study (30). Statistical significance was set at p<0.05.

Results

Civil Hospitals

Containment procedures.

The frequency and duration of containment procedure use significantly declined during the 10-year study period. Mechanical restraint was applied 118 times for a total of 16,611 minutes and was last used in September 2015. Seclusion was used three times for a total of 189 minutes and was last used in July 2013 (Table 1).

Physical restraint was used 6,972 times, involving 978 individuals, for a total of 31,916 minutes. During the study period, physical restraint use went from a high of 2.6 episodes per 1,000 days (N=1,077) in 2013 to 1.1 per 1,000 days (N=380) in 2018. In 2020, the hospitals used on average 1.9 episodes of physical restraint per 1,000 days (N=654) (p<0.001) (Table 1).

The duration of physical restraint episodes significantly declined during the decade, falling from a high of 6,392 minutes in 2012 to a low of 763 minutes in 2018. Overall, the average time in physical restraint decreased from 6.6 minutes per episode in 2012 to ≤2.4 minutes in each year from 2018 to 2020 (p<0.001) (Table 1).

Proximal causes.

Incidents of aggression, defined as a verbal or physical threat of injury, were reported 12,167 times during the decade studied and were the leading reason for physical restraint use, accounting for 32.8% (N=2,283 of 6,972) of use of the procedure. Between 2011 and 2020, the number and rate of incidents of aggression showed a nonsignificant decline (Table 1).

Patient-to-staff assaults occurred 7,856 times in the civil hospitals, with 23.8% (N=1,657 of 6,972) resulting in the use of physical restraint. During the study period, the rate of patient-to-staff assaults was unchanged, from 1.92 (N=868) per 1,000 days in 2011 to 1.90 (N=633) per 1,000 days in 2020 (Table 1).

Incidents of self-injurious behavior occurred 10,942 times and accounted for 17.7% (N=1,237) of the physical restraint procedures reported. Incidents of this type significantly declined from 3.20 (N=1,404) per 1,000 days in 2012 to 0.22 (N=77) per 1,000 days in 2020 (p=0.016) (Table 1).

Between 2011 and 2020, a total of 24,595 patient-to-patient assaults were reported, of which 4.1% (N=1,013) required the use of physical restraint. During this period, patient-to-patient assaults increased from 6.5 (N=2,947) per 1,000 days in 2011 to 6.7 (N=2,349) per 1,000 days in 2020 (Table 1).

Effect.

After each use of a containment procedure, the patient was assessed for physical injury. Overall, 84.1% (N=5,866) of the episodes of physical restraint were reported to result in no injury, and 7.9% (N=548) included an abrasion, scratch, or hematoma. Only 1.4% (N=101) were reported to include a bruise or contusion, which were the most serious injuries linked to an incident involving the use of physical restraint.

Diagnostic differences.

Between 2011 and 2020, people with a diagnosis of schizophrenia and related psychotic disorders accounted for 73% (N=2,912) of those served in the six civil hospitals (N=3,989), followed by people diagnosed as having a personality disorder, at 21% (N=838) (31).

Of the 978 individuals involved in an episode of physical restraint in the civil hospitals during this study, 52% (N=507) had a primary diagnosis of schizophrenia and related psychotic disorders and were involved in 37.0% (N=2,578) of the physical restraint episodes. People with a personality disorder diagnosis represented 15% (N=147) of the individuals involved in an episode of physical restraint and accounted for 20.9% (N=1,458) of the episodes. Individuals with mood disorders accounted for 20.2% (N=1,407) of the episodes. Individuals with co-occurring diagnoses of an intellectual disability disorder and a serious mental illness (N=21) were physically restrained 248 times.

Gender and racial-ethnic differences.

Women represented 42% (N=1,668 of 3,989) of the people served in the civil hospitals; 26% (N=434) were involved in an episode of physical restraint, representing 59.1% (N=4,123) of all physical restraint episodes reported during the study period. Of the 434 women involved in a physical restraint episode, 80% (N=348) were White, the largest racial-ethnic subgroup; these women represented 47.2% (N=3,292) of all physical restraint events reported. Black women (N=65) accounted for 8.3% (N=582) of all physical restraint events. Women belonging to all other racial-ethnic groups (Hispanic, Pacific Islander, Asian, Native American, and unknown racial-ethnic identity) accounted for 2% (N=78) of the population in the civil hospitals, with 20 of these individuals accounting for 3.4% (N=234) of all physical restraint episodes.

Men represented 58% (N=2,321 of 3,989) of the individuals in the civil hospitals, with 576 (25%) men being involved in a physical restraint episode, representing 40.8% (N=2,847) of all physical restraint episodes reported. Of these 576 men, 71% (N=409) were White, the largest racial-ethnic subgroup, representing 29.9% (N=2,087) of all physical restraint episodes. Black men (N=119, 21% of men involved in physical restraint) accounted for 6.8% (N=477) of all physical restraint procedures. Men from all other racial-ethnic groups (Hispanic, Pacific Islander, Asian, Native American, and unknown racial-ethnic identity) accounted for 1% (N=48) of the population in the civil hospitals. All individuals in this group were involved in physical restraint, accounting for 4.1% (N=283) of all episodes.

Differences in age and length of residence.

Individuals between ages 20 and 29 years were involved in most of the containment procedures reported, accounting for 37.1% (N=2,587) of physical restraint events. People ages ≥65 were involved in 7.1% (N=492) of physical restraint occurrences, and those ages <21 accounted for 2.1% (N=144).

People hospitalized for <1 year were involved in 37.0% (N=2,579) of the physical restraint events, and those who were inpatients for <90 days accounted for 15.3% (N=1,064) of the procedures. Individuals with a hospital stay of ≥5 years were involved in 25.7% (N=1,795) of all physical restraint episodes reported.

STAT medication use.

In March 2005, after an 18-month study, the hospital system discontinued the use of PRN (pro re nata, i.e., “as the need arises”) orders for psychiatric indications. It was felt that requiring a physician’s STAT (statim, i.e., “now, immediately”) decision on the need for the one-time use of additional medication was a best practice (32, 33).

During the study period, data for STAT medication administered for psychiatric indications in the civil hospitals were available from 2015 to 2020. These data revealed an overall decrease in the number of medications administered. However, when adjusted per 1,000 days, STAT medication use was unchanged (Table 1). During these 6 years, the hospitals averaged 18.2 STAT medication orders per 1,000 days (N=558 per month).

Effect of declining use of physical restraint on patient assaults resulting in any injury.

During this study, we also assessed the effect of the declining use of physical restraint on patient assaults that resulted in any patient or staff injury. A Pearson correlation analysis showed a positive and moderate relationship between the reduced use of physical restraint and patient-patient assaults with any patient injury (r=0.320, N=120, p<0.001). A similar correlation was observed for the declining use of physical restraint with any patient-staff assault resulting in any staff injury (r=0.301, N=120, p<0.001).

Forensic Centers

The frequency and duration of containment procedure use at the forensic centers declined between 2011 and 2020. During the study period, 2,567 containment procedures, involving 841 individuals and lasting 9,383 minutes, were reported. Seclusion was used once during the period, in 2012, and lasted 345 minutes. Mechanical restraint was applied 10 times, involving seven different people for a total of 432 minutes, and was last applied in 2014. Physical restraint was used 2,546 times, involving 833 different people, with 8,606 minutes recorded (Table 2).

Physical restraint use significantly declined from a high of 3.8 episodes per 1,000 days in 2012 to 2.5 in 2020 (p<0.001). The duration of physical restraint episodes significantly declined from 6.8 minutes in 2011 to ≤2.5 minutes in each year from 2017 to 2020 (p<0.001) (Table 2).

Proximal causes.

Over the 10-year period of this study, patient-to-patient assaults were the leading incident type in the forensic centers, with 8,085 reported. These incidents were also the primary reason for physical restraint use, accounting for 36% (N=907) of the episodes reported. During the study, the rate of patient-to-patient assaults significantly declined from a high of 13.2 per 1,000 days in 2012 to 7.6 in 2020 (p<0.001) (Table 2).

Acts of aggression were the second most frequent incident type to occur during the study, with 2,344 reported. Physical restraint was used in 740 of these incidents, representing 29% of the procedures. Incidents of aggression significantly declined from a high of 5.4 per 1,000 days in 2012 to a low of 0.7 in 2020 (p<0.001) (Table 2).

Patient-to-staff assaults occurred 1,386 times during this study, with 48% (N=663) requiring the use of physical restraint. The rate of patient-to-staff assaults significantly declined from a high of 2.1 per 1,000 days (N=162) in 2012 to 1.3 (N=133) in 2020 (p<0.001) (Table 2).

Incidents of self-injurious behavior were reported 778 times, of which 139 involved the use of physical restraint, or 5% of physical restraint events. During the study, the rate of self-injurious incidents significantly declined from a high of 2.1 per 1,000 days (N=156) in 2011 to a low of 0.1 (N=10) in 2020 (p<0.001) (Table 2). Property damage, attempting to escape, and other incident types accounted for 4% (N=96) of the physical restraint procedures.

Effects.

Of the 2,546 physical restraint procedures reported in the forensic centers, 79% (N=2,017) involved no injury; 10% (N=256) resulted in an abrasion, scratch, or hematoma; and 2% (N=60) led to bruising, contusion, or discoloration. The most serious injuries reported involved five different events: two involved fractures, and three involved lacerations requiring sutures. After a detailed review of these events, only the three incidents involving lacerations were attributable to the use of physical restraint.

Gender and racial-ethnic differences.

Women represented 20% (N=711 of 3,548) of the people served in the forensic centers during the study period; 26% (N=187) were involved in an episode of physical restraint, representing 29% (N=732) of all physical restraint procedures in the forensic centers. Black women represented 46% (N=330) of all women served and 58% (N=109) of women involved in a physical restraint episode, with involvement in 419 physical restraint events. White women made up 48% (N=342) of all women served in the forensic centers and 37% (N=70) of women who were involved in a physical restraint event, totaling 292 events. Physical restraint was used more frequently with Black women than with White women, but Black women’s time in restraint was shorter, averaging 3.1 minutes per episode, compared with 6.4 minutes for White women, over the same period. Asian and Hispanic women and women with unspecified ethnicity represented 5% (N=39) of women in the forensic centers; eight of these individuals were involved in a physical restraint event, totaling 21 events.

Men represented 80% (N=2,837 of 3,548) of those served in the forensic centers and were involved in 71% (N=1,813) of all physical restraint episodes in these units. Black men represented 44% (N=1,252) of the men served in the centers and were involved in 47% (N=1,207) of all restraint events reported. White men represented 48% (N=1,348) of the male population and accounted for 16% (N=416) of all physical restraint events. The mean duration of a restraint episode for Black men was 2.4 minutes and for White men was 5.3 minutes. Asian and Hispanic men and men with unspecified ethnicity represented 8% (N=227) of the men served and accounted for 7% (N=190) of physical restraint episodes in the forensic centers.

Diagnostic differences.

People with a primary diagnosis of schizophrenia and related psychotic disorders accounted for approximately 58% (N=2,057 of 3,548) of the people served in the forensic centers during this study and were involved in 51% (N=1,297) of physical restraint events. People diagnosed as having personality disorders were involved in 14% (N=352) of the physical restraint events, followed by people with mood disorders who were involved in 14% (N=346) of such events.

STAT medication use.

During the study period, 53.7% (N=6,762 of 12,593) of all containment events reported included a STAT order for psychotropic medication. The rate of STAT orders went from a high of 10.6 per 1,000 days (N=832) in 2012 to a low of 5.0 per 1,000 days (N=418) in 2016. During the study, the mean rate of STAT medication orders linked to an incident was 7.6 per 1,000 days (N=6,762) (Table 2).

Effect of declining use of physical restraint on patient assaults.

As with the results from the civil hospitals, the effects of the declining use of physical restraint on patient-to-patient and patient-to-staff assaults resulting in any injury over the 10-year period of this study were assessed in the forensic centers. A Pearson correlation analysis showed a positive and moderate association between the decreasing frequency of physical restraint episodes and patient-to-staff assaults with any staff injury (r=0.485, N=120, p<0.001). Additionally, the same analysis of the relationship between the declining use of physical restraint and patient-to-patient assaults with any patient injury exhibited a positive and moderate relationship (r=0.327, N=120, p<0.001).

Discussion

This study was a continuation of an examination of Pennsylvania’s efforts to reduce the use of containment procedures in its hospital system (2–4). The results confirm and extend the findings from examinations of the relationship between the decreasing use of all containment procedures and patient-to-patient and patient-to-staff assaults resulting in injuries to patients and staff, thus reinforcing the cultural transformation in the hospital system’s approach to services in support of recovery (1).

The racial-ethnic, gender, and age differences in the use of physical restraint observed in this study warrant further study. This analysis should include the added variables of age and length of stay, because 70% (N=1,788) of the restraint episodes reported in state forensic centers involved people ages <40.

The forensic transformation from “custody and control” to a recovery-supporting approach is intended to enable people to manage their illness and to reduce contact with the criminal legal system upon return to community living. Notably, the assertion that efforts to reduce containment procedures “may not be without risk” may remain an open, empirical question (12). However, results from this ongoing study reveal that reduced use of containment procedures leads to a safer and healthier environment of care and treatment for patients and staff without further traumatizing patients or their caregivers and without resorting to “administrative separation” as a substitute for the uses of seclusion (34).

Pennsylvania’s ongoing emphasis on staff training, deescalation techniques, psychiatric emergency response teams, and the multipronged approach noted by Rudnick (35) contribute to a person-centered approach to care and treatment with significant benefits to staff and patients. Marked reduction in aggression, patient-to-patient and patient-to-staff assaults, and instances of self-injurious behaviors in the civil hospitals and forensic centers during this period all reinforce the utility of this approach.

The hospital system’s sustained efforts in support of this initiative provide ample evidence for the replicability of this approach in other settings. Components of such an initiative include the importance of administrative and clinical leadership; sustained training and support of frontline staff; a recovery-supporting approach in which external and internal advocates, family members, and peers are involved in providing guidance; and a data-driven approach in which all staff and, in particular, union officials give feedback and input and in which the gradual introduction of changes in the protocol is driven by data. During the 10-year study period, reducing the length of time a person could be held in physical restraint from 10 to 3 minutes was a key improvement, further minimizing injury and trauma to both patients and staff.

In a recent study examining the sustainability of these initiatives in the Netherlands, Vruwink and colleagues (36) reported:

We are pleased to report that the program continues.

A more recent systematic review of the effects of the uses of seclusion and restraint concluded that “the identified literature strongly suggests that seclusion and restraints have deleterious physical and psychological consequences” (8). The findings of the present study reveal that adherence to the six core strategies for reducing seclusion and restraint (leadership, using data to inform practice, staff development, restraint and seclusion reduction tools, consumer role in patient settings, and debriefing techniques) do serve a forensic population with equal outcomes (37), thus further reducing the “militarization” of hospital-based services (38).

A review of the 2020 data revealed a possible effect of the COVID-19 pandemic on patient behavior and use of restraints. The rates per 1,000 days for each indicator—physical restraints, aggression, patient-to-patient assaults, patient-to-staff assaults, and self-injurious behaviors—were lower than the aggregate averages for these indicators for the 2011–2019 period. This observation requires additional examination over the duration of the pandemic to determine what factors may have contributed to this effect.

Over the past 30 years, many programmatic changes occurred in Pennsylvania’s mental health system that have contributed to the importance of individualizing recovery. These changes included a shift toward use of community-based service providers, who encouraged the expansion of such services. Although these changes did not always occur at the same time at every agency, hospital, or center, they helped reduce violence by reinforcing relationships, building effective communication skills, and promoting conflict resolution. Other changes included the introduction of certified peer specialists, wellness recovery action planning, dialectical behavior therapy, cognitive-behavioral therapy, trauma-informed care, and deescalation skills of every type. These approaches, and many others, are outlined in the Substance Abuse and Mental Health Services Administration’s “Promoting Alternatives to the Use of Seclusion and Restraint” (39). As a prelude to the transformation, OMHSAS invited Joel Slack of RESPECT International to offer presentations to all state hospital staff, patients, and community partners about his personal journey of recovery, imparting a message of respect and hope (40).

Conclusions

The findings of this study provide compelling evidence that uses of seclusion and restraints can be reduced or eliminated in both civil and forensic populations, with benefits to both the persons being served and their support staff. A key change during the 10-year study period was a reduction in the maximum time spent in physical restraint from 10 to 3 minutes. Additional contributing factors included a recovery-supporting clinical approach and continuous adherence to the six core strategies for seclusion and restraint reduction.