- Email: [email protected]
Prevalence, risk factors, and outcomes associated with physical restraint use in mechanically ventilated adults
Journal of Critical Care 31 (2016) 31–35
Contents lists available at ScienceDirect
Journal of Critical Care journal homepage: www.jccjournal.org
Prevalence, risk factors, and outcomes associated with physical restraint use in mechanically ventilated adults☆,☆☆,★ Louise Rose, BN, MN, PhD a,b, Lisa Burry, PharmD c,d, Ranjeeta Mallick, PhD e,f, Elena Luk, BScN b, Deborah Cook, MD g,h,i, Dean Fergusson, PhD e,f, Peter Dodek, MD, MHSc j,k, Karen Burns, MD l,m, John Granton, MD n,o,p,q,r, Niall Ferguson, MD s,t, John W. Devlin, PharmD u, Marilyn Steinberg, RN v, Sean Keenan, MD w,x, Stephen Reynolds, MD w,x, Maged Tanios, MD y, Robert A. Fowler, MDCM, MS Epi t,z,a, Michael Jacka, MD aa, Kendiss Olafson, MD ab, Yoanna Skrobik, MD ac, Sangeeta Mehta, MD d,ad,⁎ a
Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON, Canada, M4N 3M5 Lawrence S. Bloomberg Faculty of Nursing, University of Toronto, 155 College St, Toronto, ON, Canada, M5T 1P8 Department of Pharmacy and Medicine, Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 d University of Toronto, Toronto, ON, Canada e Clinical Epidemiology Program, Ottawa Hospital Research Institute, 725 Parkdale Ave, Ottawa, ON, Canada, K1Y 4E9 f Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada g St Joseph’s Healthcare, 50 Charlton Ave E, Hamilton, ON, Canada, L8N 4A6 h Department of Medicine, McMaster University, Hamilton, ON, Canada i Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, ON, Canada j Division of Critical Care Medicine and Center for Health Evaluation and Outcome Sciences, St Paul’s Hospital, Vancouver, BC, Canada, V6Z 1Y6 k University of British Columbia, 1081 Burrard St, Vancouver, BC, Canada, V6Z 1Y6 l Department of Critical Care, St Michael’s Hospital, 30 Bond St, Toronto, ON, Canada, M5B 1W8 m Interdepartmental Division of Critical Care Medicine and the Li Ka Shing Institute, Toronto, ON, Canada n Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, Toronto, ON, Canada, M5G 2C4 o General Research Institute, 200 Elizabeth St, Toronto, ON, Canada, M5G 2C4 p Interdepartmental Division of Critical Care Medicine, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada q Department of Medicine, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada r Department of Physiology, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada s Critical Care and Pulmonary Medicine, University Health Network, 200 Elizabeth St, Toronto, ON, Canada, M5G 2C4 t Interdepartmental Division of Critical Care Medicine University of Toronto, Toronto, ON, Canada u School of Pharmacy, Northeastern University, 360 Huntington Ave, Boston, MA, United States, 02115 v Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 w Department of Critical Care, Royal Columbia Hospital, Division of Critical Care, 330 E Columbia St, New Westminster, BC, Canada, V3L 3W7 x University of British Columbia, Vancouver, BC, Canada y Department of Medicine, Long Beach Memorial Medical Center, 2801 Atlantic Ave, Long Beach, CA, 90806 z Department of Medicine, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON, Canada, M4N 3M5 aa Department of Anesthesiology, University of Alberta Hospitals, 8440 112 St NW, Edmonton, AB, Canada, T6G 2B7 ab Section of Critical Care, Department of Medicine, University of Manitoba, 66 Chancellors Cir, Winnipeg, MB, Canada, R3T 2N2 ac Department of Medicine, McGill University, 3605 Rue de la Montagne, Montréal, QC H3G 2M1 ad Department of Medicine and Interdepartmental Division of Critical Care Medicine, Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 b c
a r t i c l e
i n f o
Keywords: Physical restraint Chemical restraint Sedation protocol Daily sedation interruption Intensive care
a b s t r a c t Purpose: The purpose was to describe characteristics and outcomes of restrained and nonrestrained patients enrolled in a randomized trial of protocolized sedation compared with protocolized sedation plus daily sedation interruption and to identify patient and treatment factors associated with physical restraint. Methods: This was a post hoc secondary analysis using Cox proportional hazards modeling adjusted for centerand time-varying covariates to evaluate predictors of restraint use. Results: A total of 328 (76%) of 430 patients were restrained for a median of 4 days. Restrained patients received higher daily doses of benzodiazepines (105 vs 41 mg midazolam equivalent, P b .0001) and opioids (1524 vs
☆ Conflict of interest: The authors have no conflicts of interest to declare. ☆☆ ClinicalTrials.gov NCT 00675363. ★ Funding: The SLEAP trial was funded by the Canadian Institutes of Health Research. The secondary analysis of restraint use was funded by the Canadian Association of Critical Care Nurses and a Sigma Theta Tau International Lambda Pi-At-Large Research Seed Grant and the Mount Sinai Hospital Department of Medicine. ⁎ Corresponding author at: Mount Sinai Hospital, 600 University Ave, RM 18-216, Toronto, Ontario, Canada, M5G 1X5. Tel.: +1 416 586 4800x4604; fax: +1 416 586 8480. E-mail address: [email protected] (S. Mehta). http://dx.doi.org/10.1016/j.jcrc.2015.09.011 0883-9441/© 2015 Elsevier Inc. All rights reserved.
32
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
919 μg fentanyl equivalents, P b .0001), more days of infusions (benzodiazepines 6 vs 4, P b .0001; opioids 7 vs 5, P = .02), and more daily benzodiazepine boluses (0.2 vs 0.1, P b .0001). More restrained patients received haloperidol (23% vs 12%, P = .02) and atypical antipsychotics (17% vs 4%, P = .003). More restrained patients experienced unintentional device removal (26% vs 3%, P b .001) and required reintubation (8% vs 1%, P = .01). In the multivariable analysis, alcohol use was associated with decreased risk of restraint (hazard ratio, 0.22; 95% confidence interval, 0.08-0.58). Conclusions: Physical restraint was common in mechanically ventilated adults managed with a sedation protocol. Restrained patients received more opioids and benzodiazepines. Except for alcohol use, patient characteristics and treatment factors did not predict restraint use. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Physical restraints are used to promote the safety of critically ill patients; however, their use has been associated with adverse outcomes including injury to restrained limbs [1], delirium [2–4], unplanned extubation [5,6], and an increased prevalence of posttraumatic stress symptoms in intensive care unit (ICU) survivors [7,8]. Although physical restraints are often applied to prevent patient-initiated device removal, several studies indicate high failure rates [9]. One large multicenter prevalence study conducted in the United States found that 44% of patients were physically restrained at the time of device removal [10]. Given the recognized adverse physical and psychological patient consequences of physical restraints and their lack of efficacy in preventing device removal, professional society guidelines, government legislation, and hospital accreditation standards advocate that physical restraint use be minimized across all health care settings [11–13]. Physical restraint use varies substantially across countries from 0% to 100% [14] and even among hospitals in the same country [15]. In a 2013 survey of 121 French ICUs, restraints were used at least once during mechanical ventilation in more than 50% of patients; and in 65% of these ICUs, restraints were applied for more than 50% of mechanical ventilation days [16]. A prospective observational study (I-CAN-SLEAP) conducted in 2008/2009 in 51 Canadian ICUs found that 53% of 711 mechanically ventilated patients were physically restrained for an average of 4 days [17]. More recently, the SLEAP trial, a prospective randomized trial conducted in 16 tertiary ICUs in Canada and the United States comparing protocolized sedation (control group) with protocolized sedation plus daily interruption (DI) (interruption group), found that most (328/430, 76%) patients had physical restraints applied at least once during their ICU admission [18]. In this study cohort, overall mean SedationAgitation Scale (SAS) scores were 3.3 vs 3.2 in the interruption and control groups, respectively, reflecting appropriate sedation. The frequent use of physical restraints identified in the SLEAP trial was unexpected. Therefore, we conducted a secondary analysis to describe characteristics and outcomes of restrained and nonrestrained patients and to identify associations between patient and treatment factors and restraint application. 2. Methods We performed a post hoc secondary analysis of SLEAP trial data to identify factors associated with restraint use. The SLEAP trial methods have been published previously [18]. The trial was conducted in 16 tertiary ICUs in Canada and the United States from January 2008 until July 2011 following local institutional review board approvals. 2.1. Participants and procedures The SLEAP trial enrolled patients expected to require mechanical ventilation for at least 48 hours and who were receiving continuous intravenous opioid and/or benzodiazepine infusions. In the interruption and control groups, infusions of opioid (morphine, fentanyl, or hydromorphone) and benzodiazepine (midazolam or lorazepam) were
titrated hourly by the bedside nurse according to a protocol that prioritized pain management and targeted a comfortable and rousable state, with a SAS [19] (8 sites) of 3 or 4 or Richmond Agitation-Sedation Scale (RASS) [20] (8 sites) of −3 to 0. In the interruption group, continuous infusions were interrupted daily, and patients were assessed hourly for wakefulness and the ability to perform at least 3 of the following tasks on request: eye opening, tracking, hand squeezing, and toe moving. Infusions were not restarted if the patient’s SAS was 3 to 4 without them, and patients subsequently received intravenous bolus or oral sedative therapy at the discretion of the clinical team. Infusions were resumed at 50% of the previous dose and adjusted to achieve the sedation target if ongoing continuous intravenous therapy was required. Initial application and continued use of physical restraints were at the discretion of the ICU team, in accordance with restraint policies of the participating hospitals which advocated for use of restraints as a last resort to ensure patient safety. In the all sites, application of physical restraints required a physician order every 24 hours. This process was generally initiated by the bedside nurse in response to actual or anticipated threats to patient safety. 2.2. Data collection and outcome measurements We recorded patient demographic data at the time of enrolment, including presence of psychiatric disease, dementia, stroke, cardiac disease, tobacco and alcohol consumption, and Acute Physiology and Chronic Health Evaluation (APACHE) II score. We collected daily psychoactive drug exposure (opioids, benzodiazepines, antipsychotics, adjunctive oral analgesics and sedatives, and anticholinergic agents), SAS (alternatively, RASS) scores, physical restraint use, and accidental device removal (endotracheal tube, vascular catheters, gastric tube) during mechanical ventilation. Patients were screened daily for delirium by the bedside nurse using the Intensive Care Delirium Screening Checklist (ICDSC) [21]; a score of 4 or greater at any time during the study indicated delirium. Coma was defined as a SAS score of 1 or 2, or RASS score of − 5 or − 4, for 4 or more contiguous hours during 24 hours. Using a 10-point visual analogue scale (1 = very easy; 10 = difficult), registered nurses and respiratory therapists recorded their perception of workload related to study procedures twice daily. Requirement for tracheostomy, duration of mechanical ventilation, lengths of ICU and hospital stay, discharge destination, and mortality were recorded. 2.3. Statistical analysis We summarized demographic and clinical variables using descriptive statistics. For continuous variables, we report means and standard deviations (SDs) or medians and interquartile ranges (IQRs) dependent on data distribution. For dichotomous variables, we report proportions and their 95% confidence intervals (CIs). We compared continuous variables between restrained and nonrestrained patients using either Student t tests or the Wilcoxon rank sum test as appropriate and categorical variables using the χ2 test or Fisher exact test. We converted opioids to fentanyl equivalents (10 mg morphine = 2 mg hydromorphone = 100 μg
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
fentanyl), benzodiazepines to midazolam equivalents (0.5 mg lorazepam = 1 mg midazolam), and RASS to SAS scores [18]. To determine the proportion of patients within the target range for sedation, we calculated the mean SAS (SD) score for each patient over the study duration. We used a Cox proportional hazards model accounting for clustering within sites with time-varying covariates to evaluate their association with application of restraints. Time-varying covariates (total doses of benzodiazepine and opioids, device removal, administration of antipsychotic, and presence of delirium before restraint application) were ascertained by matching the previous days’ values with next days’ restraint use, as the exact time of restraint application was not recorded. As our goal was to identify predictors of restraint use in the context of the SLEAP study, we excluded patients restrained on day 1 of the study because pre-restraint data on these time-varying covariates were unavailable for these patients. A priori, we selected covariates based on clinical relevance and published evidence indicating an association. All analyses were conducted with the SAS Enterprise Guide 4.2 (SAS Institute Inc, Cary, NC) and S-Plus version 7.0 (TIBCO Software Inc, Palo Alto, CA). All tests were 2-tailed, and we considered P ≤ .05 to be statistically significant. 3. Results We identified that 328 (76%) of 430 randomized patients were restrained for a median of 4 (IQR, 1-7) days while mechanically ventilated. The median day of restraint initiation was study day 1 (IQR, 1-1). There was no difference in the use of restraints on day vs night shifts (both median of 3 shifts) or in the interruption vs control groups. Sixteen percent of the 328 restrained patients remained restrained after extubation. Baseline characteristics were similar between the restrained and nonrestrained patients, except that APACHE II score was lower in restrained patients (23 vs 28, P b .001) and more restrained patients had a history of a neurological condition (17% vs 14%, P = .047) and tobacco use (23% vs 12%, P = .05) (Table 1). Fewer restrained patients received renal replacement therapy (17% vs 32%, P = .002) or experienced coma (25% vs 58%, P b .001), but more restrained patients experienced unintentional device removal (26% vs 3%, P b .001), with 117 (82%) of 143 device removals occurring while a patient was restrained. For patients randomized to the DI group, compliance with DI was similar for restrained and nonrestrained patients (71% vs 77%, P = .37). Whereas durations of ventilation, ICU, and hospital stay were similar for restrained and nonrestrained patients, restrained patients more frequently required reintubation (8% vs 1%, P = .01). Consistent with their lower baseline severity of illness, restrained patients had a lower rate of ICU and hospital mortality and were more likely to go directly home after hospitalization (Table 2). Restrained patients received higher mean daily doses of benzodiazepines (105 vs 41 mg midazolam equivalents, P b .0001), received benzodiazepine infusions for more days (6 vs 4, P b .0001), and received more drug boluses per day (0.2 vs 0.1, P b .0001) than nonrestrained patients. Similarly, restrained patients received higher daily doses of opioids (1524 vs 919 μg fentanyl equivalents, P b .0001) and received opioid infusions for more days (7 vs 5, P = .02). More restrained than nonrestrained patients received haloperidol (23% vs 12%, P = .02) and atypical antipsychotics (17% vs 4%, P = .003). Similar numbers of patients in the restrained and nonrestrained groups received anticholinergics, antidepressants, and medication for sleep (Table 3). Among restrained patients, although there was a strong trend for the mean daily dose of benzodiazepines to be higher before restraint application, this difference did not reach statistical significance (76 vs 34 mg, P = .08). However, higher mean daily doses of opioids were administered before restraint application (1193 vs 893 μg fentanyl equivalents, P = .02). Overall, mean SAS scores were 3.4 and 2.7 for restrained vs nonrestrained patients, respectively (P b .0001), with 97% of restrained patients in target range (SAS, 3-4) on average during mechanical
33
Table 1 Baseline characteristics of restrained and never-restrained patients
Age (y), mean (SD) Female sex Admission type Medical Surgical/trauma BMI, mean (SD) APACHE II score, mean (SD) ICU admission diagnosis Bacterial/viral pneumonia Nonurinary sepsis Other respiratory disease Aspiration pneumonia COPD Postoperative respiratory failure Urinary sepsis Gastrointestinal perforation Hepatic failure Noncardiogenic pulmonary edema Other Pre-ICU conditions Alcohol use Yes No Unknown Drinks/day, mean (SD) Tobacco use Yes No Unknown Cigarettes/day, mean (SD) Any psychiatric condition Any neurological condition
Restrained (n = 328)
Never restrained (n = 93)
P value
58 (17) 141 (43)
57 (17) 43 (46)
.82 .58
279 (85) 49 (15) 31 (13) 23 (7)
75 (81) 18 (19) 29 (9) 27 (8)
.30 .05 b.001
68 (21) 53 (16) 33 (10) 13 (4) 14 (4) 8 (2) 9 (3) 8 (2) 7 (2) 8 (2) 107 (33)
18 (19) 23 (25) 10 (11) 2 (2) – 6 (6) 3 (3) 3 (3) 2 (2) 1 (1) 25 (27)
.77 .06 .85 .54 .05 .09 .73 .71 1.0 .69 .29
75 (23) 171 (52) 82 (25) 2 (2.4)
17 (18) 56 (60) 20 (22) 3 (1.5)
76 (23) 193 (59) 59 (18) 20 (11.8) 62 (19) 56 (17)
11 (12) 65 (70) 17 (18) 15 (7.9) 10 (11) 13 (14)
.38
.81 .05
.26 .07 .05
Values are number (percentage) unless otherwise indicated. Psychiatric condition includes depression, bipolar, schizophrenia, and anxiety/panic/agoraphobia. Neurological condition includes stroke, seizure disorder, dementia, neuromuscular disease, and Parkinson disease. BMI indicates body mass index; COPD, chronic obstructive pulmonary disease.
ventilation compared with 57% of nonrestrained patients (P b .0001). Far more nonrestrained patients experienced SAS scores 1 to 2 during mechanical ventilation than patients who were restrained (43% vs 3%, P b .0001). No patients in either group experienced severe agitation (SAS, 5-7). Delirium incidence was higher in restrained patients compared with never-restrained patients (59% vs 33%, P b .001). Whereas nurse-perceived workload rated on a 10-point Likert visual analogue scale was higher for restrained compared with nonrestrained patients (4.2 vs 3.3, P b .0001), respiratory therapist–perceived workload was similar between the 2 groups. Table 4 provides univariable and multivariable analyses of variables a priori considered to be associated with the application of restraints using time-varying covariates measured before restraints were applied. Univariable analyses suggested that surgical or trauma patients were more likely to be restrained (hazard ratio [HR], 1.39; 95% CI, 1.021.90), as were patients with a history of a neurological condition (HR, 1.71; 95% CI, 1.08-2.72) or a psychiatric diagnosis (HR, 1.47; 95% CI, 1.11-2.00), whereas patients with a higher APACHE II score (HR, 0.83; 95% CI, 0.70-0.99) and those with a history of alcohol use (HR, 0.27; 95% CI, 0.10-0.76) were less likely to have restraints applied. In the multivariable model, only a history of alcohol use was associated with a statistically significant lower risk of restraint application (HR, 0.22; 95% CI, 0.08-0.58). 4. Discussion In this secondary analysis of SLEAP study data, we found that physical restraint use was common and that restraints were applied for prolonged periods with a similar pattern of restraint use during the day and night. Compared with nonrestrained patients, restrained
34
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
Table 2 Patient outcomes
Randomization group (SP +DI) Incidence of delirium At randomization Days of delirium, median IQR Development of ARDS during ICU admission Renal replacement therapy Duration of mechanical ventilation, median (IQR) Reintubation Coma Coma days, median (IQR) Tracheostomy Days in ICU, median (IQR) Days in hospital, median (IQR) Unintentional device removal All device removalsa Gastric tube Endotracheal tube Central venous or arterial catheter ICU mortality Hospital mortality Discharge destination Home Another hospital Chronic care facility
Table 4 Variables associated with application of restraints Restrained (n = 328)
Never restrained (n = 93)
P value
n = 167 patientsa
Univariate HR 95% CI
Multivariable HR 95% CI
162 (49) 195 (59) 40 (23) 1 (0-3) 124 (38)
50 (54) 31 (33) 3 (5) 0 (0-1) 42 (45)
.46 b.001 b.01 b.001 .20
1.06 (0.93-1.21) 1.07 (0.76-1.49)
1.01 (1.00-1.03) 1.49 (0.71-3.13)
57 (17) 9 (6-16)
30 (32) 11 (6-21)
b.01 .38
27 (8) 83 (25) 0 (0-1) 79 (24) 10 (6-18) 20 (11-42)
1 (1) 54 (58) 1 (0-4) 25 (27) 9 (4-19) 20 (5-47)
.01 b.001 b.001 .55 .17 .25
1 1.39 (1.02-1.90) 0.83 (0.70-0.99) 1.64 (0.99-2.71) 0.27 (0.10-0.76) 1.71 (1.08-2.72) 1.47 (1.11-2.00) 0.70 (0.42-1.16)
1 1.46 (0.80-2.67) 1.02 (0.99-1.06) 1.86 (0.92-3.75) 0.22 (0.08-0.58) 0.63 (0.37-1.08) 1.04 (0.60-1.78) 0.76 (0.20-2.84)
85 (26) 45 (14) 22 (7) 27 (8) 66 (20) 87 (27)
3 (3) 2 (2) 36 (39) 39 (42)
b.001 b.01 b.01 b.01 b.01 b.01
Age, 10-y increments Male Admission category Medical surgical/trauma APACHE II score, 5-point increments Current smoker History of alcohol use (≥2 drinks per day) History of any neurological condition History of psychiatric condition Sepsis Randomization group Sedation protocol and daily interruption Sedation protocol only Total benzodiazepine dose before restraint Total opioid dose before restraint Device removal before restraint Antipsychotic before restraint Delirium before restraint
1 0.78 (0.52-1.17) 1.03 (0.96-1.10) 1.00 (1.00-1.01) 3.92 (0.48-32.04) 0.69 (0.42-1.13) 1.62 (0.97-2.69)
1 0.49 (0.22-1.06) 1.07 (0.81-1.41) 1.00 (0.99-1.02) 4.04 (0.29-56.16) 1.22 (0.49-3.05) 1.42 (0.59-3.39)
120 (52) 79 (34) 32 (14)
16 (31) 22 (42) 14 (27)
.04 .04 .26
Values are number (percentage) unless otherwise indicated. Delirium was defined as ICDSC score of 4 or greater at any time during the study. Coma was defined as SAS score of 1 or 2, or RASS score of −5 or −4, for 4 or more contiguous hours during 24 hours. SP indicates sedation protocol–only group; SP + DI, sedation protocol + daily interruption group; ARDS, acute respiratory distress syndrome. a Refers to the number of patients experiencing device removal; number of patients removing specific devices sums to more than the number of patients because some removed multiple devices.
patients were less severely ill, less likely to be comatose during their ICU stay, but more likely to experience unintentional device removal. Indeed, 82% of device removals occurred while the patient was restrained. Restrained patients received higher doses of sedative drugs for more days in the ICU, suggesting that these patients were also being chemically restrained; however, restrained patients were more likely on average to be within sedation score targets, and more nonrestrained patients experienced SAS scores of 1 to 2. Use of physical restraint for
Table 3 Drug doses in restrained and never-restrained patients Restrained (n = 328) Midazolam equivalents (mg) Patients Dose/patient/day, median (IQR) Infusion days, median (IQR) Boluses/day, median (IQR) Fentanyl equivalents (μg) Patients Dose/patient/day, median (IQR) Infusion days, median (IQR) Boluses/day, median (IQR) Haloperidol Patients Days of ≥1 dose, mean (SD) Atypical antipsychoticsa Anticholinergicsb Antidepressants Zopiclone, trazadone
293 (89) 6 (0-75) 5 (2-8) 0 (0-0) 91 (98) 490 (20-1745) 5 (2-9) 1 (0-3) 76 (23) 1 (4) 54 (17) 42 (13) 38 (12) 41 (13)
Never restrained (n = 93)
P value
78 (84) 0 (0-24) 3 (1-6) 0 (0-0)
b.001 .002 b.001
302 (92) 200 (0-1200) 4 (2-7) 0 (0-3)
b.001 .11 .01
11 (12) 1 (2) 4 (4) 13 (14) 6 (6) 15 (16)
.02 .04 .003 .77 .15 .36
Values are number (percentage) unless otherwise indicated. Equivalents: 10 mg morphine = 2 mg hydromorphone = 100 μg fentanyl; 1 mg midazolam = 0.5 mg lorazepam. a Quetiapine, riseridone, olanzapine. b Anticholinergic = diphenhydramine, dimenhydrinate, prochlorperazine.
a
The model excluded patients restrained on day 1 of the study.
patients identified as within sedation target suggests that they were likely physically restrained even when comfortable, able to follow commands, and not exhibiting behaviors indicative of agitation. More restrained patients experienced delirium during their ICU stay and more were treated with antipsychotics compared with nonrestrained patients. Duration of mechanical ventilation and ICU and hospital lengths of stay were similar for restrained and nonrestrained patients; however, ICU and hospital mortality was higher for nonrestrained patients, and fewer returned home after hospital. Therefore, even though nonrestrained patients were more severely ill and more likely to experience comatose, thus requiring less sedative medication, they experienced worse outcomes. The frequency of restraint use is higher than what we identified for mechanically ventilated patients in the 51 Canadian ICUs participating in the I-CAN-SLEAP observational study, although the mean duration of restraint is similar (4 days) [17]. The SLEAP trial evaluated 2 sedation minimization strategies: protocolized sedation and protocolized sedation plus DI. In the final year of trial recruitment, we administered a questionnaire daily to nurses and physicians, asking whether they liked using the assigned strategy, reasons for their responses, and concerns regarding DI [22]. Undersedation was raised as a concern for both study groups by only 15% of participants and therefore may not be a key consideration for the frequent use of physical restraint. Similar to our study, others have shown [5,9] that physical restraint use is not protective against patient-initiated device removal, with 26% of restrained patients removing at least 1 device (central venous or arterial catheters, endotracheal tubes, and gastric tubes) compared with 3% of nonrestrained patients. Behaviors associated with agitation such as pulling at medical devices, climbing over bed rails, striking staff, and thrashing side to side [19] are generally cited as the primary indication for restraint application [23,24]. We also found that nonrestrained patients more frequently had a SAS score of 1 to 2 and thus were less likely to exhibit these behaviors. Although no studies demonstrate a causal relationship between physical restraint and patient-initiated device removal, several studies support an association between physical restraint use and patient-initiated device removal. Based on our findings, we hypothesize that physical restraint does not reduce agitation and may exacerbate it, thus increasing the risk of patient-initiated device removal. In our multivariable model, we found no relationship between restraint application and study randomization group, treatment factors hypothesized a priori as likely predictors of restraint application, or patient characteristics except for previous history of alcohol use, which was associated with a lower likelihood of restraint application. However, this association is likely spurious. In the I-CAN-SLEAP multicenter
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
study, we found that treatment variables associated with ever being restrained in the ICU were higher daily benzodiazepine and opioid doses, the sedation administration method (continuous vs bolus), receipt of an antipsychotic, and ever having a SAS score greater than 4 [25]. However, the I-CAN-SLEAP model cannot identify predictors of restraint application because variables were considered over the entire ICU stay and not just before restraint application. A multivariable analysis examining antecedent variables associated with delirium in the SLEAP trial found that use of physical restraints was the factor most strongly associated with an increased risk of developing delirium [4]. Although other studies have also demonstrated a relationship between restraint use and delirium [2,3], further studies are required to explore the direction of effect and to establish causality. Strengths of this study include examination of physical restraint within a randomized controlled trial that protocolized chemical restraint and targeted sedation minimization. As application of physical restraint was not protocolized within the trial, our findings provide a pragmatic perspective on restraint use in the North American context. Our study has limitations. We conducted a post hoc secondary analysis because of the unexpected high use of physical restraints. Physical restraint use was recorded as a binary (yes/no) on a daily basis only, but time of and reasons for application were not recorded. Consequently, we were not able to ascertain treatment characteristics immediately before restraint application in the multivariable model but used data (drug dose, device removal, antipsychotic, and delirium) recorded on the day prior and excluded those patients restrained at the time of study randomization. 5. Conclusion Physical restraints were commonly used in mechanically ventilated patients managed using a sedation protocol, were applied for prolonged periods, and had similar patterns of use during the day and night, suggesting that they may be overused during mechanical ventilation. Compared with nonrestrained patients, physically restrained patients experienced more patient-initiated device removal, received higher doses of benzodiazepines and opioids for more days, experienced more delirium, and received more antipsychotics but were more often within sedation target, with fewer experiencing over sedation. Prior use of alcohol was the only factor associated with decreased use of physical restraint. References [1] Evans D, Wood J, Lambert L. Patient injury and physical restraint devices: a systematic review. J Adv Nurs 2003;41:274–82. [2] Van Rompaey B, Elseviers M, Schuurmans M, Shortridge-Baggett L, Truijen S, Bossaert L. Risk factors for delirium in intensive care patients: a prospective cohort study. Crit Care 2009;13:R77.
35
[3] Micek S, Anand N, Laible B, Shannon W, Kollef M. Delirium as detected by the CAMICU predicts restraint use among mechanically ventilated medical patients. Crit Care Med 2005;33:1260–5. [4] Mehta S, Cook D, Devlin J, Skrobik Y, Meade M, Fergusson D, et al. Incidence, risk factors and outcomes of delirium in mechanically ventilated adults. Crit Care Med 2015;43:557–66. [5] Chang L, Wang K, Chao Y. Influence of physical restraint on unplanned extubation of adult intensive care patients: a case-control study. Am J Crit Care 2008;17:408–15. [6] Chevron V, Menard J-F, Richard J-C, Girault C, Leroy J, Bonmarchand G. Unplanned extubation: risk factors of development and predictive criteria for reintubation. Crit Care Med 1998;26:1049–53. [7] Hatchett C, Langley G, Schmollgruber S. Psychological sequelae following ICU admission at a level 1 academic South African hospital. S Afr J Crit Care 2010;26:52–8. [8] Jones C, Backman C, Capuzzo M, Flaatten H, Rylander C, Griffiths R. Precipitants of post-traumatic stress disorder following intensive care: a hypothesis generating study of diversity in care. Intensive Care Med 2007;33:978–85. [9] da Silva P, Fonseca M. Unplanned endotracheal extubations in the intensive care unit: systematic review, critical appraisal, and evidence-based recommendations. Anesth Analg 2012;114:1003–14. [10] Mion L, Minnick A, Leipzig R, Catrambone C, Johnson M. Patient-initiated device removal in intensive care units: a national prevalence study. Crit Care Med 2007;35:2714–20. [11] Maccioli G, Dorman T, Brown B, Mazuski J, McLean B, Kuszaj J, et al. Clinical practice guidelines for the maintenance of patient physical safety in the intensive care unit: use of restraining therapies—American College of Critical Care Medicine Task Force 2001-2002. Crit Care Med 2003;11:2665–76. [12] Government of Ontario. Patient restraints minimization act. Retrieved from http:// www.e-laws.gov.on.ca/html/statutes/english/elaws_statutes_01p16_e.htm; 2001. [on October 19th 2014]. [13] Bray K, Hill K, Robson W, Leaver G, Walker N, O'Leary M, et al. British Association of Critical Care Nurses position statement on the use of restraint in adult critical care units. Nurs Crit Care 2004;9:199–212. [14] Benbenbishty J, Adam S, Endacott R. Physical restraint use in intensive care units across Europe: the PRICE study. Intensive Crit Care Nurs 2010;26:241–5. [15] Mion L. Physical restraint in critical care settings: will they go away? Geriatr Nurs 2008;29:421–3. [16] De Jonghe B, Constantin J, Chanques G, Capdevila X, Lefratn J, Outin H, et al. Physical restraint in mechanically ventilated ICU patients: a survey of French practice. Intensive Care Med 2013;39:31–7. [17] Burry L, Williamson M, Perreault M, Rose L, Cook D, Ferguson N, et al. A prospective evaluation of sedative, analgesic, anti-psychotic and neuromuscular blocker prescribing practices in Canadian intensive care unit patients receiving mechanical ventilation. Can J Anaesth 2014;61:619–30. [18] Mehta S, Burry L, Cook D, Fergusson D, Steinberg M, Granton J, et al. A randomized trial of daily sedation interruption in mechanically ventilated critically ill patients cared for with a sedation protocol. JAMA 2012;308:1985–92. [19] Riker R, Picard J, Fraser G. Prospective evaluation of the Sedation-Agitation Scale for adult critically ill patients. 1999;27:1325–9. [20] Sessler C, Gosnell M, Grap M, Brophy G, O'Neal P, Keane K, et al. The Richmond Agitation-Sedation Scale: validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med 2002;166:1338–44. [21] Bergeron N, Dubois M, Dumont M, Dial S, Skrobik Y. Intensive Care Delirium Screening Checklist: evaluation of a new screening tool. Intensive Care Med 2001;27:859–64. [22] Rose L, Fitzgerald E, Cook D, Scott K, Steinberg M, Devlin J, et al. Clinician perspectives on protocols designed to minimize sedation. J Crit Care 2015;30:348–52. [23] Frezza E, Carleton G, Valenziano C. A quality improvement and risk management initiative for surgical ICU patients: a study of the effects of physical restraints and sedation on the incidence of self-extubation. Am J Med Qual 2000;15:221–5. [24] Pesiri A. Two-year study of the prevention of unintentional extubation. Crit Care Nurs Q 1994;17:35–9. [25] Luk E, Sneyers B, Rose L, Perreault M, Williamson D, Mehta S, et al. Predictors of physical restraint use in Canadian intensive care units. Crit Care 2014;18:R46.
Contents lists available at ScienceDirect
Journal of Critical Care journal homepage: www.jccjournal.org
Prevalence, risk factors, and outcomes associated with physical restraint use in mechanically ventilated adults☆,☆☆,★ Louise Rose, BN, MN, PhD a,b, Lisa Burry, PharmD c,d, Ranjeeta Mallick, PhD e,f, Elena Luk, BScN b, Deborah Cook, MD g,h,i, Dean Fergusson, PhD e,f, Peter Dodek, MD, MHSc j,k, Karen Burns, MD l,m, John Granton, MD n,o,p,q,r, Niall Ferguson, MD s,t, John W. Devlin, PharmD u, Marilyn Steinberg, RN v, Sean Keenan, MD w,x, Stephen Reynolds, MD w,x, Maged Tanios, MD y, Robert A. Fowler, MDCM, MS Epi t,z,a, Michael Jacka, MD aa, Kendiss Olafson, MD ab, Yoanna Skrobik, MD ac, Sangeeta Mehta, MD d,ad,⁎ a
Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON, Canada, M4N 3M5 Lawrence S. Bloomberg Faculty of Nursing, University of Toronto, 155 College St, Toronto, ON, Canada, M5T 1P8 Department of Pharmacy and Medicine, Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 d University of Toronto, Toronto, ON, Canada e Clinical Epidemiology Program, Ottawa Hospital Research Institute, 725 Parkdale Ave, Ottawa, ON, Canada, K1Y 4E9 f Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada g St Joseph’s Healthcare, 50 Charlton Ave E, Hamilton, ON, Canada, L8N 4A6 h Department of Medicine, McMaster University, Hamilton, ON, Canada i Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, ON, Canada j Division of Critical Care Medicine and Center for Health Evaluation and Outcome Sciences, St Paul’s Hospital, Vancouver, BC, Canada, V6Z 1Y6 k University of British Columbia, 1081 Burrard St, Vancouver, BC, Canada, V6Z 1Y6 l Department of Critical Care, St Michael’s Hospital, 30 Bond St, Toronto, ON, Canada, M5B 1W8 m Interdepartmental Division of Critical Care Medicine and the Li Ka Shing Institute, Toronto, ON, Canada n Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, Toronto, ON, Canada, M5G 2C4 o General Research Institute, 200 Elizabeth St, Toronto, ON, Canada, M5G 2C4 p Interdepartmental Division of Critical Care Medicine, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada q Department of Medicine, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada r Department of Physiology, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada s Critical Care and Pulmonary Medicine, University Health Network, 200 Elizabeth St, Toronto, ON, Canada, M5G 2C4 t Interdepartmental Division of Critical Care Medicine University of Toronto, Toronto, ON, Canada u School of Pharmacy, Northeastern University, 360 Huntington Ave, Boston, MA, United States, 02115 v Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 w Department of Critical Care, Royal Columbia Hospital, Division of Critical Care, 330 E Columbia St, New Westminster, BC, Canada, V3L 3W7 x University of British Columbia, Vancouver, BC, Canada y Department of Medicine, Long Beach Memorial Medical Center, 2801 Atlantic Ave, Long Beach, CA, 90806 z Department of Medicine, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON, Canada, M4N 3M5 aa Department of Anesthesiology, University of Alberta Hospitals, 8440 112 St NW, Edmonton, AB, Canada, T6G 2B7 ab Section of Critical Care, Department of Medicine, University of Manitoba, 66 Chancellors Cir, Winnipeg, MB, Canada, R3T 2N2 ac Department of Medicine, McGill University, 3605 Rue de la Montagne, Montréal, QC H3G 2M1 ad Department of Medicine and Interdepartmental Division of Critical Care Medicine, Mount Sinai Hospital, 600 University Ave, Toronto, ON, Canada, M5G 1X5 b c
a r t i c l e
i n f o
Keywords: Physical restraint Chemical restraint Sedation protocol Daily sedation interruption Intensive care
a b s t r a c t Purpose: The purpose was to describe characteristics and outcomes of restrained and nonrestrained patients enrolled in a randomized trial of protocolized sedation compared with protocolized sedation plus daily sedation interruption and to identify patient and treatment factors associated with physical restraint. Methods: This was a post hoc secondary analysis using Cox proportional hazards modeling adjusted for centerand time-varying covariates to evaluate predictors of restraint use. Results: A total of 328 (76%) of 430 patients were restrained for a median of 4 days. Restrained patients received higher daily doses of benzodiazepines (105 vs 41 mg midazolam equivalent, P b .0001) and opioids (1524 vs
☆ Conflict of interest: The authors have no conflicts of interest to declare. ☆☆ ClinicalTrials.gov NCT 00675363. ★ Funding: The SLEAP trial was funded by the Canadian Institutes of Health Research. The secondary analysis of restraint use was funded by the Canadian Association of Critical Care Nurses and a Sigma Theta Tau International Lambda Pi-At-Large Research Seed Grant and the Mount Sinai Hospital Department of Medicine. ⁎ Corresponding author at: Mount Sinai Hospital, 600 University Ave, RM 18-216, Toronto, Ontario, Canada, M5G 1X5. Tel.: +1 416 586 4800x4604; fax: +1 416 586 8480. E-mail address: [email protected] (S. Mehta). http://dx.doi.org/10.1016/j.jcrc.2015.09.011 0883-9441/© 2015 Elsevier Inc. All rights reserved.
32
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
919 μg fentanyl equivalents, P b .0001), more days of infusions (benzodiazepines 6 vs 4, P b .0001; opioids 7 vs 5, P = .02), and more daily benzodiazepine boluses (0.2 vs 0.1, P b .0001). More restrained patients received haloperidol (23% vs 12%, P = .02) and atypical antipsychotics (17% vs 4%, P = .003). More restrained patients experienced unintentional device removal (26% vs 3%, P b .001) and required reintubation (8% vs 1%, P = .01). In the multivariable analysis, alcohol use was associated with decreased risk of restraint (hazard ratio, 0.22; 95% confidence interval, 0.08-0.58). Conclusions: Physical restraint was common in mechanically ventilated adults managed with a sedation protocol. Restrained patients received more opioids and benzodiazepines. Except for alcohol use, patient characteristics and treatment factors did not predict restraint use. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Physical restraints are used to promote the safety of critically ill patients; however, their use has been associated with adverse outcomes including injury to restrained limbs [1], delirium [2–4], unplanned extubation [5,6], and an increased prevalence of posttraumatic stress symptoms in intensive care unit (ICU) survivors [7,8]. Although physical restraints are often applied to prevent patient-initiated device removal, several studies indicate high failure rates [9]. One large multicenter prevalence study conducted in the United States found that 44% of patients were physically restrained at the time of device removal [10]. Given the recognized adverse physical and psychological patient consequences of physical restraints and their lack of efficacy in preventing device removal, professional society guidelines, government legislation, and hospital accreditation standards advocate that physical restraint use be minimized across all health care settings [11–13]. Physical restraint use varies substantially across countries from 0% to 100% [14] and even among hospitals in the same country [15]. In a 2013 survey of 121 French ICUs, restraints were used at least once during mechanical ventilation in more than 50% of patients; and in 65% of these ICUs, restraints were applied for more than 50% of mechanical ventilation days [16]. A prospective observational study (I-CAN-SLEAP) conducted in 2008/2009 in 51 Canadian ICUs found that 53% of 711 mechanically ventilated patients were physically restrained for an average of 4 days [17]. More recently, the SLEAP trial, a prospective randomized trial conducted in 16 tertiary ICUs in Canada and the United States comparing protocolized sedation (control group) with protocolized sedation plus daily interruption (DI) (interruption group), found that most (328/430, 76%) patients had physical restraints applied at least once during their ICU admission [18]. In this study cohort, overall mean SedationAgitation Scale (SAS) scores were 3.3 vs 3.2 in the interruption and control groups, respectively, reflecting appropriate sedation. The frequent use of physical restraints identified in the SLEAP trial was unexpected. Therefore, we conducted a secondary analysis to describe characteristics and outcomes of restrained and nonrestrained patients and to identify associations between patient and treatment factors and restraint application. 2. Methods We performed a post hoc secondary analysis of SLEAP trial data to identify factors associated with restraint use. The SLEAP trial methods have been published previously [18]. The trial was conducted in 16 tertiary ICUs in Canada and the United States from January 2008 until July 2011 following local institutional review board approvals. 2.1. Participants and procedures The SLEAP trial enrolled patients expected to require mechanical ventilation for at least 48 hours and who were receiving continuous intravenous opioid and/or benzodiazepine infusions. In the interruption and control groups, infusions of opioid (morphine, fentanyl, or hydromorphone) and benzodiazepine (midazolam or lorazepam) were
titrated hourly by the bedside nurse according to a protocol that prioritized pain management and targeted a comfortable and rousable state, with a SAS [19] (8 sites) of 3 or 4 or Richmond Agitation-Sedation Scale (RASS) [20] (8 sites) of −3 to 0. In the interruption group, continuous infusions were interrupted daily, and patients were assessed hourly for wakefulness and the ability to perform at least 3 of the following tasks on request: eye opening, tracking, hand squeezing, and toe moving. Infusions were not restarted if the patient’s SAS was 3 to 4 without them, and patients subsequently received intravenous bolus or oral sedative therapy at the discretion of the clinical team. Infusions were resumed at 50% of the previous dose and adjusted to achieve the sedation target if ongoing continuous intravenous therapy was required. Initial application and continued use of physical restraints were at the discretion of the ICU team, in accordance with restraint policies of the participating hospitals which advocated for use of restraints as a last resort to ensure patient safety. In the all sites, application of physical restraints required a physician order every 24 hours. This process was generally initiated by the bedside nurse in response to actual or anticipated threats to patient safety. 2.2. Data collection and outcome measurements We recorded patient demographic data at the time of enrolment, including presence of psychiatric disease, dementia, stroke, cardiac disease, tobacco and alcohol consumption, and Acute Physiology and Chronic Health Evaluation (APACHE) II score. We collected daily psychoactive drug exposure (opioids, benzodiazepines, antipsychotics, adjunctive oral analgesics and sedatives, and anticholinergic agents), SAS (alternatively, RASS) scores, physical restraint use, and accidental device removal (endotracheal tube, vascular catheters, gastric tube) during mechanical ventilation. Patients were screened daily for delirium by the bedside nurse using the Intensive Care Delirium Screening Checklist (ICDSC) [21]; a score of 4 or greater at any time during the study indicated delirium. Coma was defined as a SAS score of 1 or 2, or RASS score of − 5 or − 4, for 4 or more contiguous hours during 24 hours. Using a 10-point visual analogue scale (1 = very easy; 10 = difficult), registered nurses and respiratory therapists recorded their perception of workload related to study procedures twice daily. Requirement for tracheostomy, duration of mechanical ventilation, lengths of ICU and hospital stay, discharge destination, and mortality were recorded. 2.3. Statistical analysis We summarized demographic and clinical variables using descriptive statistics. For continuous variables, we report means and standard deviations (SDs) or medians and interquartile ranges (IQRs) dependent on data distribution. For dichotomous variables, we report proportions and their 95% confidence intervals (CIs). We compared continuous variables between restrained and nonrestrained patients using either Student t tests or the Wilcoxon rank sum test as appropriate and categorical variables using the χ2 test or Fisher exact test. We converted opioids to fentanyl equivalents (10 mg morphine = 2 mg hydromorphone = 100 μg
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
fentanyl), benzodiazepines to midazolam equivalents (0.5 mg lorazepam = 1 mg midazolam), and RASS to SAS scores [18]. To determine the proportion of patients within the target range for sedation, we calculated the mean SAS (SD) score for each patient over the study duration. We used a Cox proportional hazards model accounting for clustering within sites with time-varying covariates to evaluate their association with application of restraints. Time-varying covariates (total doses of benzodiazepine and opioids, device removal, administration of antipsychotic, and presence of delirium before restraint application) were ascertained by matching the previous days’ values with next days’ restraint use, as the exact time of restraint application was not recorded. As our goal was to identify predictors of restraint use in the context of the SLEAP study, we excluded patients restrained on day 1 of the study because pre-restraint data on these time-varying covariates were unavailable for these patients. A priori, we selected covariates based on clinical relevance and published evidence indicating an association. All analyses were conducted with the SAS Enterprise Guide 4.2 (SAS Institute Inc, Cary, NC) and S-Plus version 7.0 (TIBCO Software Inc, Palo Alto, CA). All tests were 2-tailed, and we considered P ≤ .05 to be statistically significant. 3. Results We identified that 328 (76%) of 430 randomized patients were restrained for a median of 4 (IQR, 1-7) days while mechanically ventilated. The median day of restraint initiation was study day 1 (IQR, 1-1). There was no difference in the use of restraints on day vs night shifts (both median of 3 shifts) or in the interruption vs control groups. Sixteen percent of the 328 restrained patients remained restrained after extubation. Baseline characteristics were similar between the restrained and nonrestrained patients, except that APACHE II score was lower in restrained patients (23 vs 28, P b .001) and more restrained patients had a history of a neurological condition (17% vs 14%, P = .047) and tobacco use (23% vs 12%, P = .05) (Table 1). Fewer restrained patients received renal replacement therapy (17% vs 32%, P = .002) or experienced coma (25% vs 58%, P b .001), but more restrained patients experienced unintentional device removal (26% vs 3%, P b .001), with 117 (82%) of 143 device removals occurring while a patient was restrained. For patients randomized to the DI group, compliance with DI was similar for restrained and nonrestrained patients (71% vs 77%, P = .37). Whereas durations of ventilation, ICU, and hospital stay were similar for restrained and nonrestrained patients, restrained patients more frequently required reintubation (8% vs 1%, P = .01). Consistent with their lower baseline severity of illness, restrained patients had a lower rate of ICU and hospital mortality and were more likely to go directly home after hospitalization (Table 2). Restrained patients received higher mean daily doses of benzodiazepines (105 vs 41 mg midazolam equivalents, P b .0001), received benzodiazepine infusions for more days (6 vs 4, P b .0001), and received more drug boluses per day (0.2 vs 0.1, P b .0001) than nonrestrained patients. Similarly, restrained patients received higher daily doses of opioids (1524 vs 919 μg fentanyl equivalents, P b .0001) and received opioid infusions for more days (7 vs 5, P = .02). More restrained than nonrestrained patients received haloperidol (23% vs 12%, P = .02) and atypical antipsychotics (17% vs 4%, P = .003). Similar numbers of patients in the restrained and nonrestrained groups received anticholinergics, antidepressants, and medication for sleep (Table 3). Among restrained patients, although there was a strong trend for the mean daily dose of benzodiazepines to be higher before restraint application, this difference did not reach statistical significance (76 vs 34 mg, P = .08). However, higher mean daily doses of opioids were administered before restraint application (1193 vs 893 μg fentanyl equivalents, P = .02). Overall, mean SAS scores were 3.4 and 2.7 for restrained vs nonrestrained patients, respectively (P b .0001), with 97% of restrained patients in target range (SAS, 3-4) on average during mechanical
33
Table 1 Baseline characteristics of restrained and never-restrained patients
Age (y), mean (SD) Female sex Admission type Medical Surgical/trauma BMI, mean (SD) APACHE II score, mean (SD) ICU admission diagnosis Bacterial/viral pneumonia Nonurinary sepsis Other respiratory disease Aspiration pneumonia COPD Postoperative respiratory failure Urinary sepsis Gastrointestinal perforation Hepatic failure Noncardiogenic pulmonary edema Other Pre-ICU conditions Alcohol use Yes No Unknown Drinks/day, mean (SD) Tobacco use Yes No Unknown Cigarettes/day, mean (SD) Any psychiatric condition Any neurological condition
Restrained (n = 328)
Never restrained (n = 93)
P value
58 (17) 141 (43)
57 (17) 43 (46)
.82 .58
279 (85) 49 (15) 31 (13) 23 (7)
75 (81) 18 (19) 29 (9) 27 (8)
.30 .05 b.001
68 (21) 53 (16) 33 (10) 13 (4) 14 (4) 8 (2) 9 (3) 8 (2) 7 (2) 8 (2) 107 (33)
18 (19) 23 (25) 10 (11) 2 (2) – 6 (6) 3 (3) 3 (3) 2 (2) 1 (1) 25 (27)
.77 .06 .85 .54 .05 .09 .73 .71 1.0 .69 .29
75 (23) 171 (52) 82 (25) 2 (2.4)
17 (18) 56 (60) 20 (22) 3 (1.5)
76 (23) 193 (59) 59 (18) 20 (11.8) 62 (19) 56 (17)
11 (12) 65 (70) 17 (18) 15 (7.9) 10 (11) 13 (14)
.38
.81 .05
.26 .07 .05
Values are number (percentage) unless otherwise indicated. Psychiatric condition includes depression, bipolar, schizophrenia, and anxiety/panic/agoraphobia. Neurological condition includes stroke, seizure disorder, dementia, neuromuscular disease, and Parkinson disease. BMI indicates body mass index; COPD, chronic obstructive pulmonary disease.
ventilation compared with 57% of nonrestrained patients (P b .0001). Far more nonrestrained patients experienced SAS scores 1 to 2 during mechanical ventilation than patients who were restrained (43% vs 3%, P b .0001). No patients in either group experienced severe agitation (SAS, 5-7). Delirium incidence was higher in restrained patients compared with never-restrained patients (59% vs 33%, P b .001). Whereas nurse-perceived workload rated on a 10-point Likert visual analogue scale was higher for restrained compared with nonrestrained patients (4.2 vs 3.3, P b .0001), respiratory therapist–perceived workload was similar between the 2 groups. Table 4 provides univariable and multivariable analyses of variables a priori considered to be associated with the application of restraints using time-varying covariates measured before restraints were applied. Univariable analyses suggested that surgical or trauma patients were more likely to be restrained (hazard ratio [HR], 1.39; 95% CI, 1.021.90), as were patients with a history of a neurological condition (HR, 1.71; 95% CI, 1.08-2.72) or a psychiatric diagnosis (HR, 1.47; 95% CI, 1.11-2.00), whereas patients with a higher APACHE II score (HR, 0.83; 95% CI, 0.70-0.99) and those with a history of alcohol use (HR, 0.27; 95% CI, 0.10-0.76) were less likely to have restraints applied. In the multivariable model, only a history of alcohol use was associated with a statistically significant lower risk of restraint application (HR, 0.22; 95% CI, 0.08-0.58). 4. Discussion In this secondary analysis of SLEAP study data, we found that physical restraint use was common and that restraints were applied for prolonged periods with a similar pattern of restraint use during the day and night. Compared with nonrestrained patients, restrained
34
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
Table 2 Patient outcomes
Randomization group (SP +DI) Incidence of delirium At randomization Days of delirium, median IQR Development of ARDS during ICU admission Renal replacement therapy Duration of mechanical ventilation, median (IQR) Reintubation Coma Coma days, median (IQR) Tracheostomy Days in ICU, median (IQR) Days in hospital, median (IQR) Unintentional device removal All device removalsa Gastric tube Endotracheal tube Central venous or arterial catheter ICU mortality Hospital mortality Discharge destination Home Another hospital Chronic care facility
Table 4 Variables associated with application of restraints Restrained (n = 328)
Never restrained (n = 93)
P value
n = 167 patientsa
Univariate HR 95% CI
Multivariable HR 95% CI
162 (49) 195 (59) 40 (23) 1 (0-3) 124 (38)
50 (54) 31 (33) 3 (5) 0 (0-1) 42 (45)
.46 b.001 b.01 b.001 .20
1.06 (0.93-1.21) 1.07 (0.76-1.49)
1.01 (1.00-1.03) 1.49 (0.71-3.13)
57 (17) 9 (6-16)
30 (32) 11 (6-21)
b.01 .38
27 (8) 83 (25) 0 (0-1) 79 (24) 10 (6-18) 20 (11-42)
1 (1) 54 (58) 1 (0-4) 25 (27) 9 (4-19) 20 (5-47)
.01 b.001 b.001 .55 .17 .25
1 1.39 (1.02-1.90) 0.83 (0.70-0.99) 1.64 (0.99-2.71) 0.27 (0.10-0.76) 1.71 (1.08-2.72) 1.47 (1.11-2.00) 0.70 (0.42-1.16)
1 1.46 (0.80-2.67) 1.02 (0.99-1.06) 1.86 (0.92-3.75) 0.22 (0.08-0.58) 0.63 (0.37-1.08) 1.04 (0.60-1.78) 0.76 (0.20-2.84)
85 (26) 45 (14) 22 (7) 27 (8) 66 (20) 87 (27)
3 (3) 2 (2) 36 (39) 39 (42)
b.001 b.01 b.01 b.01 b.01 b.01
Age, 10-y increments Male Admission category Medical surgical/trauma APACHE II score, 5-point increments Current smoker History of alcohol use (≥2 drinks per day) History of any neurological condition History of psychiatric condition Sepsis Randomization group Sedation protocol and daily interruption Sedation protocol only Total benzodiazepine dose before restraint Total opioid dose before restraint Device removal before restraint Antipsychotic before restraint Delirium before restraint
1 0.78 (0.52-1.17) 1.03 (0.96-1.10) 1.00 (1.00-1.01) 3.92 (0.48-32.04) 0.69 (0.42-1.13) 1.62 (0.97-2.69)
1 0.49 (0.22-1.06) 1.07 (0.81-1.41) 1.00 (0.99-1.02) 4.04 (0.29-56.16) 1.22 (0.49-3.05) 1.42 (0.59-3.39)
120 (52) 79 (34) 32 (14)
16 (31) 22 (42) 14 (27)
.04 .04 .26
Values are number (percentage) unless otherwise indicated. Delirium was defined as ICDSC score of 4 or greater at any time during the study. Coma was defined as SAS score of 1 or 2, or RASS score of −5 or −4, for 4 or more contiguous hours during 24 hours. SP indicates sedation protocol–only group; SP + DI, sedation protocol + daily interruption group; ARDS, acute respiratory distress syndrome. a Refers to the number of patients experiencing device removal; number of patients removing specific devices sums to more than the number of patients because some removed multiple devices.
patients were less severely ill, less likely to be comatose during their ICU stay, but more likely to experience unintentional device removal. Indeed, 82% of device removals occurred while the patient was restrained. Restrained patients received higher doses of sedative drugs for more days in the ICU, suggesting that these patients were also being chemically restrained; however, restrained patients were more likely on average to be within sedation score targets, and more nonrestrained patients experienced SAS scores of 1 to 2. Use of physical restraint for
Table 3 Drug doses in restrained and never-restrained patients Restrained (n = 328) Midazolam equivalents (mg) Patients Dose/patient/day, median (IQR) Infusion days, median (IQR) Boluses/day, median (IQR) Fentanyl equivalents (μg) Patients Dose/patient/day, median (IQR) Infusion days, median (IQR) Boluses/day, median (IQR) Haloperidol Patients Days of ≥1 dose, mean (SD) Atypical antipsychoticsa Anticholinergicsb Antidepressants Zopiclone, trazadone
293 (89) 6 (0-75) 5 (2-8) 0 (0-0) 91 (98) 490 (20-1745) 5 (2-9) 1 (0-3) 76 (23) 1 (4) 54 (17) 42 (13) 38 (12) 41 (13)
Never restrained (n = 93)
P value
78 (84) 0 (0-24) 3 (1-6) 0 (0-0)
b.001 .002 b.001
302 (92) 200 (0-1200) 4 (2-7) 0 (0-3)
b.001 .11 .01
11 (12) 1 (2) 4 (4) 13 (14) 6 (6) 15 (16)
.02 .04 .003 .77 .15 .36
Values are number (percentage) unless otherwise indicated. Equivalents: 10 mg morphine = 2 mg hydromorphone = 100 μg fentanyl; 1 mg midazolam = 0.5 mg lorazepam. a Quetiapine, riseridone, olanzapine. b Anticholinergic = diphenhydramine, dimenhydrinate, prochlorperazine.
a
The model excluded patients restrained on day 1 of the study.
patients identified as within sedation target suggests that they were likely physically restrained even when comfortable, able to follow commands, and not exhibiting behaviors indicative of agitation. More restrained patients experienced delirium during their ICU stay and more were treated with antipsychotics compared with nonrestrained patients. Duration of mechanical ventilation and ICU and hospital lengths of stay were similar for restrained and nonrestrained patients; however, ICU and hospital mortality was higher for nonrestrained patients, and fewer returned home after hospital. Therefore, even though nonrestrained patients were more severely ill and more likely to experience comatose, thus requiring less sedative medication, they experienced worse outcomes. The frequency of restraint use is higher than what we identified for mechanically ventilated patients in the 51 Canadian ICUs participating in the I-CAN-SLEAP observational study, although the mean duration of restraint is similar (4 days) [17]. The SLEAP trial evaluated 2 sedation minimization strategies: protocolized sedation and protocolized sedation plus DI. In the final year of trial recruitment, we administered a questionnaire daily to nurses and physicians, asking whether they liked using the assigned strategy, reasons for their responses, and concerns regarding DI [22]. Undersedation was raised as a concern for both study groups by only 15% of participants and therefore may not be a key consideration for the frequent use of physical restraint. Similar to our study, others have shown [5,9] that physical restraint use is not protective against patient-initiated device removal, with 26% of restrained patients removing at least 1 device (central venous or arterial catheters, endotracheal tubes, and gastric tubes) compared with 3% of nonrestrained patients. Behaviors associated with agitation such as pulling at medical devices, climbing over bed rails, striking staff, and thrashing side to side [19] are generally cited as the primary indication for restraint application [23,24]. We also found that nonrestrained patients more frequently had a SAS score of 1 to 2 and thus were less likely to exhibit these behaviors. Although no studies demonstrate a causal relationship between physical restraint and patient-initiated device removal, several studies support an association between physical restraint use and patient-initiated device removal. Based on our findings, we hypothesize that physical restraint does not reduce agitation and may exacerbate it, thus increasing the risk of patient-initiated device removal. In our multivariable model, we found no relationship between restraint application and study randomization group, treatment factors hypothesized a priori as likely predictors of restraint application, or patient characteristics except for previous history of alcohol use, which was associated with a lower likelihood of restraint application. However, this association is likely spurious. In the I-CAN-SLEAP multicenter
L. Rose et al. / Journal of Critical Care 31 (2016) 31–35
study, we found that treatment variables associated with ever being restrained in the ICU were higher daily benzodiazepine and opioid doses, the sedation administration method (continuous vs bolus), receipt of an antipsychotic, and ever having a SAS score greater than 4 [25]. However, the I-CAN-SLEAP model cannot identify predictors of restraint application because variables were considered over the entire ICU stay and not just before restraint application. A multivariable analysis examining antecedent variables associated with delirium in the SLEAP trial found that use of physical restraints was the factor most strongly associated with an increased risk of developing delirium [4]. Although other studies have also demonstrated a relationship between restraint use and delirium [2,3], further studies are required to explore the direction of effect and to establish causality. Strengths of this study include examination of physical restraint within a randomized controlled trial that protocolized chemical restraint and targeted sedation minimization. As application of physical restraint was not protocolized within the trial, our findings provide a pragmatic perspective on restraint use in the North American context. Our study has limitations. We conducted a post hoc secondary analysis because of the unexpected high use of physical restraints. Physical restraint use was recorded as a binary (yes/no) on a daily basis only, but time of and reasons for application were not recorded. Consequently, we were not able to ascertain treatment characteristics immediately before restraint application in the multivariable model but used data (drug dose, device removal, antipsychotic, and delirium) recorded on the day prior and excluded those patients restrained at the time of study randomization. 5. Conclusion Physical restraints were commonly used in mechanically ventilated patients managed using a sedation protocol, were applied for prolonged periods, and had similar patterns of use during the day and night, suggesting that they may be overused during mechanical ventilation. Compared with nonrestrained patients, physically restrained patients experienced more patient-initiated device removal, received higher doses of benzodiazepines and opioids for more days, experienced more delirium, and received more antipsychotics but were more often within sedation target, with fewer experiencing over sedation. Prior use of alcohol was the only factor associated with decreased use of physical restraint. References [1] Evans D, Wood J, Lambert L. Patient injury and physical restraint devices: a systematic review. J Adv Nurs 2003;41:274–82. [2] Van Rompaey B, Elseviers M, Schuurmans M, Shortridge-Baggett L, Truijen S, Bossaert L. Risk factors for delirium in intensive care patients: a prospective cohort study. Crit Care 2009;13:R77.
35
[3] Micek S, Anand N, Laible B, Shannon W, Kollef M. Delirium as detected by the CAMICU predicts restraint use among mechanically ventilated medical patients. Crit Care Med 2005;33:1260–5. [4] Mehta S, Cook D, Devlin J, Skrobik Y, Meade M, Fergusson D, et al. Incidence, risk factors and outcomes of delirium in mechanically ventilated adults. Crit Care Med 2015;43:557–66. [5] Chang L, Wang K, Chao Y. Influence of physical restraint on unplanned extubation of adult intensive care patients: a case-control study. Am J Crit Care 2008;17:408–15. [6] Chevron V, Menard J-F, Richard J-C, Girault C, Leroy J, Bonmarchand G. Unplanned extubation: risk factors of development and predictive criteria for reintubation. Crit Care Med 1998;26:1049–53. [7] Hatchett C, Langley G, Schmollgruber S. Psychological sequelae following ICU admission at a level 1 academic South African hospital. S Afr J Crit Care 2010;26:52–8. [8] Jones C, Backman C, Capuzzo M, Flaatten H, Rylander C, Griffiths R. Precipitants of post-traumatic stress disorder following intensive care: a hypothesis generating study of diversity in care. Intensive Care Med 2007;33:978–85. [9] da Silva P, Fonseca M. Unplanned endotracheal extubations in the intensive care unit: systematic review, critical appraisal, and evidence-based recommendations. Anesth Analg 2012;114:1003–14. [10] Mion L, Minnick A, Leipzig R, Catrambone C, Johnson M. Patient-initiated device removal in intensive care units: a national prevalence study. Crit Care Med 2007;35:2714–20. [11] Maccioli G, Dorman T, Brown B, Mazuski J, McLean B, Kuszaj J, et al. Clinical practice guidelines for the maintenance of patient physical safety in the intensive care unit: use of restraining therapies—American College of Critical Care Medicine Task Force 2001-2002. Crit Care Med 2003;11:2665–76. [12] Government of Ontario. Patient restraints minimization act. Retrieved from http:// www.e-laws.gov.on.ca/html/statutes/english/elaws_statutes_01p16_e.htm; 2001. [on October 19th 2014]. [13] Bray K, Hill K, Robson W, Leaver G, Walker N, O'Leary M, et al. British Association of Critical Care Nurses position statement on the use of restraint in adult critical care units. Nurs Crit Care 2004;9:199–212. [14] Benbenbishty J, Adam S, Endacott R. Physical restraint use in intensive care units across Europe: the PRICE study. Intensive Crit Care Nurs 2010;26:241–5. [15] Mion L. Physical restraint in critical care settings: will they go away? Geriatr Nurs 2008;29:421–3. [16] De Jonghe B, Constantin J, Chanques G, Capdevila X, Lefratn J, Outin H, et al. Physical restraint in mechanically ventilated ICU patients: a survey of French practice. Intensive Care Med 2013;39:31–7. [17] Burry L, Williamson M, Perreault M, Rose L, Cook D, Ferguson N, et al. A prospective evaluation of sedative, analgesic, anti-psychotic and neuromuscular blocker prescribing practices in Canadian intensive care unit patients receiving mechanical ventilation. Can J Anaesth 2014;61:619–30. [18] Mehta S, Burry L, Cook D, Fergusson D, Steinberg M, Granton J, et al. A randomized trial of daily sedation interruption in mechanically ventilated critically ill patients cared for with a sedation protocol. JAMA 2012;308:1985–92. [19] Riker R, Picard J, Fraser G. Prospective evaluation of the Sedation-Agitation Scale for adult critically ill patients. 1999;27:1325–9. [20] Sessler C, Gosnell M, Grap M, Brophy G, O'Neal P, Keane K, et al. The Richmond Agitation-Sedation Scale: validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med 2002;166:1338–44. [21] Bergeron N, Dubois M, Dumont M, Dial S, Skrobik Y. Intensive Care Delirium Screening Checklist: evaluation of a new screening tool. Intensive Care Med 2001;27:859–64. [22] Rose L, Fitzgerald E, Cook D, Scott K, Steinberg M, Devlin J, et al. Clinician perspectives on protocols designed to minimize sedation. J Crit Care 2015;30:348–52. [23] Frezza E, Carleton G, Valenziano C. A quality improvement and risk management initiative for surgical ICU patients: a study of the effects of physical restraints and sedation on the incidence of self-extubation. Am J Med Qual 2000;15:221–5. [24] Pesiri A. Two-year study of the prevention of unintentional extubation. Crit Care Nurs Q 1994;17:35–9. [25] Luk E, Sneyers B, Rose L, Perreault M, Williamson D, Mehta S, et al. Predictors of physical restraint use in Canadian intensive care units. Crit Care 2014;18:R46.