Temporal association between physical restraint and venous thromboembolism risk among psychiatric inpatients: A matched case-control study

Abstract

BACKGROUND

Physical restraint (PR) is commonly used in psychiatric inpatient care to manage acute behavioral risks. However, restraint-related immobility may increase the risk of venous thromboembolism (VTE). Evidence regarding the temporal characteristics of restraint exposure and subsequent VTE risk remains limited.

AIM

To examine the association between PR and VTE risk among psychiatric inpatients, with a particular focus on restraint duration, cumulative restraint hours, and the interval between restraint discontinuation and VTE diagnosis.

METHODS

This single-center matched case-control study was conducted at a tertiary psychiatric hospital in China. Adult psychiatric inpatients who developed first-onset VTE during hospitalization between January 2019 and October 2021 were defined as cases. For each case, up to four controls were matched by ward, sex, age, and hospitalization duration. Exposure to PR was assessed using four indicators: Restraint use, restraint duration in days, cumulative restraint hours, and the interval from the final restraint episode to VTE diagnosis. Multivariable conditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs), adjusting for demographic characteristics, psychotropic medication use, and VTE-related risk factors.

RESULTS

A total of 127 patients with VTE and their matched controls were included. Multivariable analysis showed that PR was an independent risk factor for VTE (OR = 1.82, 95%CI: 1.16-2.85). Temporal subgroup analyses showed that restraint duration ≤ 3 days (adjusted OR = 1.67, 95%CI: 1.04-2.70) and cumulative restraint exposure exceeding 12 hours (12-24 hours: Adjusted OR = 6.27; > 24 hours: Adjusted OR = 6.81) were associated with higher odds of VTE. In addition, the first 7 days after release from the last restraint episode represented a high-risk time window for VTE occurrence (OR = 5.29, 95%CI: 2.84-9.85).

CONCLUSION

PR was independently associated with higher VTE odds in psychiatric inpatients, particularly after cumulative exposure exceeding 12 hours and during the first 7 days after restraint discontinuation.

Key Words: Physical restraint; Venous thromboembolism; Psychiatric inpatients; Case-control study; Temporal association; Risk factors

Core Tip: This matched case-control study evaluated the temporal association between physical restraint and venous thromboembolism (VTE) among psychiatric inpatients. Physical restraint was independently associated with increased VTE risk, which was particularly elevated among patients with cumulative restraint exposure exceeding 12 hours and within 7 days after restraint discontinuation. These findings highlight the importance of incorporating restraint-related temporal indicators into VTE monitoring and prevention strategies in psychiatric wards.

INTRODUCTION

Venous thromboembolism (VTE), encompassing deep vein thrombosis (DVT) and pulmonary embolism (PE), is a serious and potentially fatal complication in hospitalized patients[1]. Compared with the general inpatient population, psychiatric inpatients are exposed to a range of psychiatric-specific risk factors, including reduced mobility, catatonic states, communication barriers, and the use of physical restraint (PR), all of which may substantially increase the risk of VTE. In psychiatric inpatient care, these factors often arise in the context of acute symptom management and continuous clinical observation. Previous studies have reported that the incidence of VTE in psychiatric inpatient settings is markedly higher than that observed in community populations and is characterized by a distinct profile of risk factors[2,3].

PR is a restrictive intervention employed in psychiatric care when patients pose an imminent risk of severe self-harm or impulsive behavior, with the primary aim of ensuring safety and maintaining treatment continuity[4]. As a coercive intervention that directly restricts patient mobility, PR is highly relevant to routine inpatient management and monitoring. Several case-control studies have reported a significant association between PR exposure and VTE occurrence[5-9], including those conducted by Zhang et al[6] and Ishida et al[8]. The underlying biological mechanisms are consistent with Virchow’s triad; whereby restricted movement promotes venous stasis and may contribute to endothelial injury[10]. Most existing studies have treated PR as a dichotomous exposure (presence vs absence), with limited consideration of its temporal characteristics, such as restraint duration or cumulative exposure time, and whether these features exert graded effects on VTE risk.

Recent evidence indicates that the duration of PR may be associated with the occurrence of VTE in psychiatric patients[5]. Nevertheless, important temporal dimensions remain insufficiently explored, including cumulative restraint exposure measured in hours and the interval between restraint discontinuation and subsequent VTE diagnosis. Identification of these time-dependent risk patterns may be clinically important as psychiatric restraint is typically implemented for short-term crisis management, and critical windows of elevated VTE risk could inform more targeted monitoring and prevention strategies.

The present matched case-control study aimed to investigate the association between PR and first-onset VTE among psychiatric inpatients, with particular emphasis on temporal characteristics of restraint exposure, including restraint duration, cumulative restraint hours, and the post-restraint risk period. We hypothesized that restraint-related VTE risk would vary according to exposure duration and would remain elevated after restraint discontinuation.

MATERIALS AND METHODS

Study design and setting

This was a single-center retrospective case–control study conducted at Shenzhen Kangning Hospital, a tertiary psychiatric hospital in Shenzhen, China. The hospital comprises 29 psychiatric inpatient wards with a total of 1330 beds and admits approximately 10000 patients annually. PR is permitted in 25 wards, including acute-phase wards (n = 6), emergency treatment wards (n = 4), geriatric wards (n = 3), chronic-phase wards (n = 3), bipolar disorder wards (n = 2), addiction wards (n = 2), and one ward in psychosomatic medicine, precision psychiatry, compulsory medical treatment, psychiatric disorders secondary to physical illness, and isolation care, respectively. PR is prohibited in four wards, including adolescent, depressive disorder, anxiety disorder, and sleep disorder wards. The present study aimed to examine the association between exposure to PR and the risk of VTE among psychiatric inpatients.

Ethical approval

All study procedures were conducted in accordance with relevant national regulations and institutional ethical guidelines and complied with the Declaration of Helsinki (1975, revised in 2008). The study protocol was approved by the Ethics Committee of Shenzhen Kangning Hospital, approval No. 2023-K030-01. As this study used retrospective, de-identified data generated during routine clinical care, the requirement for informed consent was waived by the ethics committee.

Study participants

Eligible participants were Chinese patients aged 18 years or older who were hospitalized between January 1, 2019, and October 15, 2021, in psychiatric wards where PR was permitted.

Patients who developed first-onset VTE during hospitalization were defined as cases. VTE included PE (International Classification of Diseases, 10^th Revision code I26) and DVT (DVT; International Classification of Diseases, 10^th Revision codes I80.0-I80.3). All inpatients underwent routine D-dimer testing on admission and monthly thereafter during hospitalization. A D-dimer level ≥ 500 μg/L was considered positive and prompted further diagnostic evaluation, including ultrasonography. All VTE diagnoses were confirmed by contrast-enhanced computed tomography or ultrasonography, as clinically appropriate. Patients with negative imaging findings were not classified as cases.

On the date of VTE diagnosis, up to four controls were selected for each case from patients hospitalized in the same ward. Controls had no symptoms or a diagnosis of VTE during hospitalization and were matched to cases by sex and age within ± 5 years. When four eligible controls were unavailable, three matched controls were accepted. This matching strategy was used to reduce bias related to ward-level differences in restraint practices and to ensure that cases and controls had comparable opportunities for restraint exposure and clinical observation.

Data collection

Data were extracted from the hospital electronic medical record system and nursing surveillance records. Variables were selected according to previous literature and clinical relevance to VTE risk in psychiatric inpatients.

The primary outcome variable was first-onset in-hospital VTE, defined as a new diagnosis of DVT, PE, or both during hospitalization. For cases, the index date was defined as the date of VTE diagnosis. For matched controls, the index date was defined as the VTE diagnosis date of the corresponding matched case. Exposure and covariate information was assessed before or on the index date, as appropriate.

PR exposure was the main parameter of interest and was characterized using four indicators: (1) Any use of PR before the index date (yes/no); (2) Restraint duration in days, defined as any restraint occurring within a calendar day counted as one restraint day; (3) Cumulative restraint hours, calculated as the total duration of all restraint episodes before the index date; and (4) The interval in days between the end of the final restraint episode and the index date. For patients without PR exposure, the corresponding restraint-related variables were coded as no exposure. All restraint-related information was obtained from the hospital’s standardized Impulsive Behavior Intervention Treatment Form, which documents the start and end times of each restraint episode, indications for restraint, and relevant nursing measures.

Psychiatric-related variables included the presence of a current major depressive episode and the use of antipsychotics, antidepressants, or benzodiazepines. Demographic and general clinical characteristics included age, sex, body mass index (BMI), and smoking status.

VTE-related clinical risk factors included active malignancy, history of VTE, reduced mobility, known thrombophilia, recent trauma or surgery within one month, heart and/or respiratory failure, acute infection and/or rheumatologic disorder, acute myocardial infarction or ischemic stroke, and ongoing hormonal therapy. Reduced mobility was defined as bed rest with bathroom privileges for at least 3 days, either due to the patient’s functional limitations or ordered by a physician. PR exposure alone did not meet this definition; reduced mobility required a separately documented bed-rest status. As anticoagulants were not routinely used for VTE prophylaxis in this hospital during the study period, anticoagulant-related variables were not included in the analysis.

Statistical analysis

All statistical analyses were performed using STATA version 15.0. Baseline characteristics of cases and controls are summarized using descriptive statistics. Categorical variables are presented as numbers and percentages and were compared using the χ² test or Fisher’s exact test, as appropriate. No missing data were observed for the variables included in the primary analyses; therefore, no imputation was performed.

Given the matched case-control design, conditional logistic regression was used as the primary analytical method. Four separate models were constructed to examine the associations between VTE risk and different restraint-related indicators: (1) Any PR exposure; (2) Restraint duration in days; (3) Cumulative restraint hours; and (4) The interval between the end of the final restraint episode and the index date.

Multivariable models were adjusted for BMI, smoking status, use of antipsychotics, antidepressants, and benzodiazepines, as well as VTE-related risk factors eligible for inclusion based on univariable analyses and event distribution. Variables with zero events or extremely sparse distributions were not included in the multivariable models to avoid unstable estimates. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). All statistical tests were two-sided, and P < 0.05 was considered statistically significant.

RESULTS

Study population

Between January 1, 2019, and October 15, 2021, a total of 21274 patients were admitted to psychiatric wards where PR was permitted at Shenzhen Kangning Hospital, accounting for 28211 hospitalizations. The study size was determined by the number of all eligible first-onset VTE cases identified during the study period and the availability of up to four matched controls for each case. During this period, 127 patients were newly diagnosed with VTE during hospitalization, corresponding to an incidence of 0.45% per hospitalization. Of these cases, 13 patients (10.2%) had isolated DVT, 104 patients (81.9%) had isolated PE, and 10 patients (7.9%) had both DVT and PE. The mean age of cases was 52.22 ± 16.00 years, and 53.54% were female.

Baseline characteristics

Baseline demographic and clinical characteristics of cases and controls are summarized in Table 1. No significant differences were observed between the two groups with respect to sex, age categories, BMI ≥ 30 kg/m², presence of a current major depressive episode, or use of psychotropic medications. The distribution of VTE-related risk factors - including history of VTE, reduced mobility, known thrombophilia, and recent trauma or surgery - was also comparable between cases and controls. For several variables, such as active malignancy, acute infection, and heart or respiratory failure, ORs could not be estimated due to the extremely low number of events in both groups. An inverse association between recorded smoking status and VTE was observed in the univariable analysis (OR = 0.29, 95%CI: 0.11-0.76).

Table 1 Baseline demographic and clinical characteristics of cases and controls, n (%).

Variable	Cases (n = 127)	Controls (n = 423)	OR (95%CI)	P value
Sex
Female	68 (53.54)	211 (49.88)	1.00 (ref.)	-
Age (years)
≤ 40	32 (25.20)	132 (31.21)	1.00 (ref.)	-
41-60	52 (40.94)	180 (42.55)	1.90 (0.53-7.19)	0.31
61-74	34 (26.77)	84 (19.86)	6.50 (0.71-59.90)	0.10
≥ 75	9 (7.09)	27 (6.38)	2.20 (0.09-56.39)	0.63
BMI ≥ 30 kg/m²
Yes	2 (1.57)	17 (4.02)	0.37 (0.08-1.69)	0.20
Smoking
Yes	6 (4.72)	52 (12.29)	0.29 (0.11-0.76)	0.01
Current major depressive episode
Yes	8 (6.30)	21 (4.96)	0.96 (0.39-2.39)	0.94
Psychotropic medications
Antipsychotics (yes)	101 (79.53)	345 (81.56)	1.21 (0.66-2.21)	0.53
Antidepressants (yes)	31 (24.41)	75 (17.73)	1.24 (0.69-2.34)	0.47
Benzodiazepines (yes)	12 (9.45)	44 (10.40)	0.66 (0.28-1.52)	0.33
Hypertension
Yes	24 (18.90)	65 (15.37)	1.11 (0.61-2.02)	0.53
Active cancer
Yes	0 (0.00)	1 (0.24)	-	-
History of VTE
Yes	5 (3.94)	13 (3.07)	1.04 (0.72-1.49)	0.82
Reduced mobility
Yes	7 (5.51)	13 (3.07)	1.90 (0.75-4.82)	0.18
Thrombophilia
Yes	4 (3.15)	9 (2.13)	1.37 (0.38-4.98)	0.63
Recent trauma/surgery (≤ 1 month)
Yes	4 (3.15)	5 (1.18)	2.26 (0.60-8.58)	0.23
Heart/respiratory failure
Yes	0 (0.00)	5 (1.18)	-	-
Acute infection/rheumatic disease
Yes	0 (0.00)	6 (1.42)	-	-
Acute myocardial infarction/stroke
Yes	0 (0.00)	1 (0.24)	-	-
Hormonal therapy
Yes	0 (0.00)	3 (0.71)	-	-

Odds ratio for variables with zero events could not be estimated. OR: Odds ratio; CI: Confidence interval; VTE: Venous thromboembolism; BMI: Body mass index.

Association between PR and VTE risk

The results of the multivariable conditional logistic regression analyses are shown in Table 2. After adjustment for BMI, smoking status, psychotropic medication use, and eligible VTE-related risk factors, PR was significantly associated with an increased risk of VTE (adjusted OR = 1.82, 95%CI: 1.16-2.85).

Table 2 Multivariable conditional logistic regression analyses for the association between physical restraint and venous thromboembolism, n (%).

Model	Cases (n = 127)	Controls (n = 423)	aOR (95%CI)
Model 1 PR exposure
PR (yes)	50 (39.37)	116 (27.42)	1.82 (1.16-2.85)
Model 2 PR duration in days
≤ 3	39 (30.71)	100 (23.64)	1.67 (1.04-2.70)
4-7	8 (6.30)	12 (2.84)	2.62 (0.93-7.40)
≥ 8	3 (2.36)	4 (0.95)	2.84 (0.52-15.65)
Model 3 cumulative PR hours
≤ 12	30 (23.62)	103 (24.35)	1.25 (0.75-2.08)
12-24	11 (8.66)	8 (1.89)	6.27 (1.98-19.90)
> 24	9 (7.09)	5 (1.18)	6.81 (1.92-24.36)
Model 4 interval between last PR and VTE diagnosis
≤ 7	37 (29.13)	34 (8.04)	5.29 (2.84-9.85)
8-14	3 (2.36)	31 (7.33)	0.38 (0.11-1.32)
≥ 15	10 (7.87)	51 (12.06)	0.82 (0.37-1.80)

All models were adjusted for age, body mass index, smoking status, use of antipsychotics, antidepressants, benzodiazepines, and venous thromboembolism-related risk factors with non-zero events. aOR: Adjusted odds ratio; CI: Confidence interval; PR: Physical restraint; VTE: Venous thromboembolism.

Further analyses of temporal restraint characteristics showed that restraint duration of ≤ 3 days was associated with an increased risk of VTE (adjusted OR = 1.67, 95%CI: 1.04-2.70). Although the ORs were also elevated for restraint duration of 4-7 days and ≥ 8 days, the confidence intervals were wide and did not reach statistical significance.

When cumulative restraint exposure was examined in hours, cumulative restraint of ≤ 12 hours was not significantly associated with VTE risk (adjusted OR = 1.25, 95%CI: 0.75-2.08). In contrast, cumulative restraint exposure of 12-24 hours (adjusted OR = 6.27, 95%CI: 1.98-19.90) and > 24 hours (adjusted OR = 6.81, 95%CI: 1.92-24.36) was associated with higher odds of VTE. However, these estimates had wide confidence intervals, particularly for cumulative exposure > 24 hours, indicating limited precision.

The strongest association was observed when the final restraint episode ended within 7 days before the index date (adjusted OR = 5.29, 95%CI: 2.84-9.85). No statistically significant increase in VTE risk was observed for intervals of 8-14 days or ≥ 15 days.

DISCUSSION

In this single-center retrospective matched case-control study of psychiatric inpatients, PR was independently associated with higher odds of VTE after adjustment for BMI, smoking status, psychotropic medication use, and available VTE-related clinical risk factors. The association also varied according to the temporal characteristics of restraint exposure. Cumulative restraint exposure exceeding 12 hours was associated with higher odds of VTE, and the strongest association was observed when the final restraint episode ended within 7 days before VTE diagnosis. These findings suggest that PR should not be considered solely as a binary exposure, as its cumulative duration and temporal proximity to VTE diagnosis may also be clinically relevant.

Our findings are consistent with previous studies which reported an association between PR and increased VTE risk among psychiatric inpatients. Several investigations, including case-control studies by Tsuda et al[11] and Ishida et al[8], as well as cohort studies by Funayama et al[9] and Baandrup et al[12], have identified PR as a significant risk factor for thromboembolic events. The present study extends these findings by focusing on temporal features of restraint exposure, including restraint duration in days, cumulative restraint hours, and the interval between restraint discontinuation and VTE diagnosis. Compared with studies that treated restraint exposure simply as present or absent, this approach may provide a more clinically informative description of when VTE risk is likely to increase.

An important contribution of this study is the incorporation of a temporal, dynamic perspective on restraint-related VTE risk, which has been largely overlooked in previous research. With respect to restraint duration, our findings suggest that even relatively short periods of restraint may be associated with an increased risk of VTE. This observation differs from conclusions drawn from commonly used VTE risk assessment tools, which typically emphasize prolonged immobility as a key risk factor[13,14].

Several mechanisms could plausibly contribute to this association. Restricted lower-extremity movement may reduce venous return, while insufficient fluid intake during restraint could exacerbate a hypercoagulable state. Agitation and struggling during restraint have also been proposed to contribute to endothelial injury[15]. However, these physiological pathways were not directly assessed in the present study and should be regarded as potential explanations rather than established mechanisms. Differences in patient characteristics, restraint implementation practices, and diagnostic timing may also partly account for inconsistencies between our findings and those reported in large administrative database studies[16].

In addition to restraint duration measured in days, cumulative restraint exposure emerged as another important temporal dimension of VTE risk. Our findings suggest that restraint-related thrombotic risk may develop relatively early during cumulative exposure. In contrast, some previous studies have suggested that short delays in pharmacological prophylaxis following restraint initiation may be safe[17]. Existing psychiatric VTE risk assessment tools also commonly include restraint exposure based on predefined time thresholds[18]. Taken together, these findings indicate that restraint-related VTE risk may emerge earlier than currently assumed in some clinical frameworks, highlighting the need for further investigation to determine the optimal timing of preventive measures during the early phase of restraint exposure.

The higher odds of VTE diagnosis observed within 7 days after restraint discontinuation represent another clinically important finding. A similar concentration of risk during the early treatment period has been reported for antipsychotic exposure[19], although this temporal parallel does not establish a shared underlying mechanism. Previous studies have also reported that PR commonly lasts approximately one week[9], suggesting that restraint exposure and subsequent diagnostic evaluation may cluster within a relatively short period.

Venous stasis and endothelial injury have been proposed as biologically plausible pathways contributing to thrombogenesis in restrained patients[20]. However, these processes were not directly assessed in the present study, and the observed temporal association does not establish that restraint directly caused VTE. Residual confounding and differences in clinical surveillance or diagnostic timing may also have contributed to this finding. Nevertheless, the results support continued clinical monitoring during the early post-restraint period.

The inverse association between recorded smoking status and VTE observed in the univariable analysis should be interpreted cautiously, as it may reflect residual confounding, selection bias, or limited statistical power rather than a biologically protective effect. More broadly, psychiatric inpatients may be exposed to setting-specific risk factors, including acute symptom exacerbation, antipsychotic treatment, and abrupt mobility restriction related to PR; antipsychotic use has previously been associated with first-time VTE[21]. Risk assessment tools developed for general hospital settings may therefore not fully capture these psychiatric-specific factors[13,14]. There is a need to develop and validate VTE risk assessment tools tailored to psychiatric settings[8], incorporating restraint characteristics, catatonic features, and electroconvulsive therapy to improve risk stratification and clinical applicability.

Several limitations of this study should be acknowledged. First, the single-center retrospective design may limit the generalizability of the findings. Second, although routine D-dimer screening was performed, universal imaging screening for asymptomatic VTE was not conducted, and some asymptomatic events may therefore have been undetected. Third, despite multivariable adjustment, residual confounding cannot be excluded as hydration status, degree of immobility, restraint modality, catatonia severity, and use of mechanical thromboprophylaxis were not captured in the retrospective dataset. These factors may have influenced both restraint exposure and VTE risk. Finally, several temporal subgroups contained few exposed patients. In particular, the estimates for restraint duration ≥ 8 days and cumulative restraint exposure > 24 hours were based on low counts and had wide confidence intervals. These estimates should therefore be interpreted cautiously and cannot support firm conclusions regarding an exposure-response pattern.

Despite these limitations, this study provides clinically relevant evidence that restraint-related VTE risk among psychiatric inpatients may be time-dependent. The findings suggest that psychiatric wards should not only record whether restraint was used, but also monitor cumulative restraint exposure and maintain clinical vigilance during the early period after restraint discontinuation. Efforts to minimize unnecessary restraint duration, maintain hydration when clinically feasible, encourage safe mobilization after restraint removal, and promptly evaluate suspected thromboembolic symptoms may contribute to safer inpatient psychiatric care.

CONCLUSION

This study showed that PR was independently associated with an increased risk of VTE among psychiatric inpatients. The association varied according to temporal characteristics of restraint exposure, particularly cumulative restraint hours and the interval after restraint discontinuation. Patients with cumulative restraint exposure exceeding 12 hours and those within the first 7 days after the final restraint episode may represent clinically important groups for strengthened VTE monitoring.

These findings support the incorporation of restraint-related temporal indicators into VTE risk assessment and prevention strategies in psychiatric inpatient settings. Future multicenter prospective studies are needed to validate these temporal patterns, clarify underlying mechanisms, and develop more refined VTE prevention protocols for restrained psychiatric inpatients.

ACKNOWLEDGEMENTS

We are deeply grateful to the Information Technology Department of Shenzhen Kangning Hospital for their assistance with data extraction and technical support. We also acknowledge the contributions of clinical and nursing staff involved in routine data documentation.