Patient-reported Anxiety Scores Are Associated With Lower Physical Function in Patients Experiencing Orthopaedic Trauma

Abstract

Background

The outcomes of orthopaedic trauma are not solely determined by injury severity or surgical treatment. Studies of numerous orthopaedic outcomes have found that psychosocial factors are also important. Symptoms of anxiety have been linked to long-term pain and disability. Although the existence of a relationship between psychosocial factors and functional outcomes is accepted across multiple disciplines, quantification of this association in patients who have experienced orthopaedic trauma has remained limited. Measuring the anxiety experienced by these individuals and the association with long-term functional outcomes remain poorly understood.

Questions/purposes

(1) Is there an association between early postoperative anxiety symptoms and late recovery of self-reported physical function in patients with orthopaedic trauma? (2) What was the impact of other factors such as demographic variables and comorbidities on late recovery physical function scores, and how did the magnitude of these factors compare with the association with anxiety score? (3) Did patients who presented as trauma activations differ regarding their anxiety symptoms and late-recovery self-reported physical function?

Methods

A total of 1550 patients with lower extremity fractures and postoperative Patient-Reported Outcomes Measurement Information System (PROMIS) anxiety and physical function scores treated between January 1, 2014, and January 1, 2021, at an academic Level I trauma center in North America were assessed. We performed a bivariate regression between the initial PROMIS anxiety and physical function, as well as a multivariate regression including age, gender, BMI, and American Society of Anesthesiologists class to control for potential confounding variables. In a subgroup of 787 patients presenting as trauma activations, we performed a separate regression including Injury Severity Score.

Results

PROMIS anxiety was associated with decreased late-recovery physical function (β = -2.64 [95% CI -3.006 to -2.205]; p < 0.001). The relationship between PROMIS anxiety and physical function remained after controlling for confounding variables in our overall cohort (β = -2.54 [95% CI -2.93 to -2.15]; p < 0.001) and in the trauma activation cohort (β = -2.71 [95% CI -3.19 to -2.23]; p < 0.001). Age and American Society of Anesthesiologists score were associated with worse PROMIS physical function scores, while being a man was associated with better PROMIS physical function scores (age: β= -1.26 [95% CI -1.50 to -1.02]; American Society of Anesthesiologists class: β=-2.99 [95% CI -3.52 to -2.46]; men: β = 0.95 [95% CI 0.16 to 1.75]). There were no differences in initial anxiety symptoms or late-recovery physical function between patients who presented as trauma activations and those who did not. Injury Severity Scores were independently associated with worse function (β = -1.45 [95% CI -2.11 to -0.79].

Conclusion

Initial patient self-reported anxiety is negatively associated with patient-reported physical function at the final follow-up interval in a broad cohort of patients with orthopaedic lower extremity injuries undergoing surgery. Identifying patients with high initial PROMIS anxiety scores may allow us to determine which patients will report lower functional scores at the final follow-up. Future investigations could focus on the effect of psychosocial interventions such as cognitive behavioral therapy and mindfulness on functional scores.

Level of Evidence

Level III, therapeutic study.

Introduction

Orthopaedic trauma can have a devastating impact on patient mobility and function [11, 13, 14, 24]. Additionally, orthopaedic trauma is common, typically involves younger patients, and can have a substantial and lasting effect on patients. Surgeon strategies to improve functional outcomes focus on modifiable factors surrounding treatment that allow for large variance in outcomes. In addition to the nature of an injury and the surgical treatment, multiple psychologic factors are known to be associated with disability [19]. Catastrophic thinking, pain anxiety, and depression have been correlated with poorer outcomes in terms of mental health and physical health–related measures [17, 18, 20]. The magnitude and clinical significance of psychologic factors remain difficult for surgeons to measure, although there is broader recognition of their importance to patients.

There is substantial evidence about the association of psychosocial factors with long-term pain and functional and clinical outcomes; however, this evidence has focused on elective procedures [12, 18, 19], with few studies in patients with orthopaedic trauma [5, 26, 27]. Understanding the nature and magnitude of the association between initial anxiety and final function in patients who have sustained orthopaedic trauma remains limited and small in scale. Demonstrating these associations with commonly collected patient-reported outcome measures might increase awareness of anxiety in patients recovering from musculoskeletal injury and facilitate further studies of future psychosocial interventions in these patients. The Patient-Reported Outcomes Measurement Information System (PROMIS) is efficient and generalizable across injury types and measures outcomes in various health domains [7, 9]. The mental and physical components of the PROMIS have been shown to perform well against other legacy measures [1, 3, 8, 12].

We therefore asked: (1) Is there an association between early postoperative anxiety symptoms and late recovery of self-reported physical function in patients with orthopaedic trauma? (2) What was the impact of other factors such as demographic variables and comorbidities on late recovery physical function scores, and how did the magnitude of these factors compare with the association with anxiety score? (3) Did patients who presented as trauma activations differ regarding their anxiety symptoms and late-recovery self-reported physical function?

Patients and Methods

Study Design and Setting

This was a retrospective cross-sectional study drawn from data gathered in a longitudinally maintained database of patient-reported outcome scores. The site was a single tertiary Level I trauma center in Utah with a catchment area including major portions of the surrounding five states.

Participant Selection

Data on all patients undergoing operative fixation at the study institution between January 1, 2014, and January 1, 2021, were reviewed. Data were extracted from the database using the following Current Procedural Terminology codes: 27758, 27759, 27814, 27792, 27235, 27766, 27506, 27236, 27822, 28406, 28415, 27513, 28445, 28420, 28436, 27846, and 27511. Each patient’s record was manually reviewed to ensure there were no coding errors. Inclusion criteria were operative fixation of the femoral neck, femoral shaft, tibial shaft, rotational ankle, or hindfoot fracture; available PROMIS physical function (PF) scores during the follow-up period; and postoperative baseline PROMIS anxiety score. Exclusion criteria were multiple operations or injuries during the follow-up period and unavailable patient-reported outcome measures at the time of follow-up.

Variables and Data Sources

Available demographic information, including age, gender, BMI, smoking status, American Society of Anesthesiologists (ASA) class, and Charlson comorbidity index score, were extracted from the electronic medical record (Epic, Epic Systems). Each postoperative score was from a single patient encounter or clinic visit. A subgroup of these patients had an injury that met the criteria for mechanistic injuries, or injury severity in the emergency department for trauma activations. A subgroup analysis including the Injury Severity Scores (ISS) of these patients was performed to determine the association of injury severity with anxiety and subsequent functional outcome.

Primary and Secondary Study Outcomes

The primary aim of this study was to determine the association between initial postoperative PROMIS anxiety scores and self-reported PROMIS PF scores during late recovery in patients with lower extremity orthopaedic trauma.

The secondary study goals were to determine the association of basic demographic variables with PROMIS PF scores, and to determine whether the symptoms of anxiety were different in a subgroup of trauma activations and explained by injury severity.

Ethical Approval

Our institutional review board evaluated this study and found it exempt from formal review given the study used previously collected medical record and patient-reported outcome data (IRB# 00071850).

Sample Description

In total, 3589 operative fixations were identified. After the removal of patients undergoing additional surgeries, patients with multiple injuries treated during the study period, or patients without follow-up PROMIS scores available, 1550 unique lower extremity fractures remained in the cohort. Of these, 787 were patients with trauma activations with ISS available.

Cohort Demographics and Injury Characteristics

The mean patient age at the time of surgery was 47 ± 20 years. The mean BMI was 28 ± 7 kg/m². The overall cohort was 53% (814 of 1550) men (Table 1). Demographics were similar between the nontrauma activation and trauma activation groups with the following exceptions: The nontrauma activation group had patients with a slightly greater BMI, a lower proportion of men, and a slightly greater Charlson comorbidity index score. Regarding injury characteristics, patients with trauma activation had a greater proportion of injuries associated with high-energy mechanisms, including femoral shaft fractures (16% [129 of 787] versus 8% [61 of 763]) and tibial shaft fractures (26% [201 of 787] versus 10% [80 of 763]; p < 0.001). The nontrauma activation patient cohort had a greater proportion of rotational ankle fractures (44% versus 22%; p < 0.001). The mean ISS in the trauma activation cohort was 9.8 ± 7.9 (Table 1). Overall, PROMIS PF scores increased substantially during the follow-up period, from 31 ± 8 at 4 to 8 weeks postoperatively to 40 ± 10 at the 5- to 16-month postoperative recovery interval (Table 2).

Table 1.

Cohort demographics

Parameter	Overall group (1550)	Nontrauma activations (n = 763)	Trauma activations (n = 787)	p value
BMI in kg/m2, mean ± SD	28 ± 6.8	28 ± 6.8	27 ± 6.7	0.045
Age in years, mean ± SD	47 ± 20	48 ± 20	46 ± 20	0.12
Men, % (n)	53 (814)	48 (368)	57 (446)	< 0.01
ASA score, mean ± SD	2.2 ± 0.9	2.1 ± 0.9	2.2 ± 0.9	0.71
Charlson comorbidity index, mean ± SD	1.4 ± 2.5	1.5 ± 2.5	1.3 ± 2.4	0.04
Income in USD, mean ± SD	95,746 ± 31,269	95,655 ± 31,576	95,837 ± 30,981	0.91
PROMIS anxiety score, mean ± SD	57 ± 10	57 ± 10	57 ± 10	0.64
Injury Severity Score, mean ± SD			9.8 ± 7.9
Fracture type, % (n)				< 0.001
Ankle	31 (473)	39 (298)	22 (175)
Calcaneus	8 (117)	1 (6)	6 (49)
Distal femur	5 (80)	5 (34)	6 (46)
Femoral neck	15 (229)	12 (93)	17 (136)
Femoral shaft	12 (190)	8 (61)	16 (129)
Talus	6 (89)	6 (42)	6 (47)
Tibial shaft	18 (281)	10 (80)	3 (20)

A t-test was used for comparing continuous variables; a chi-square test was used for comparing categorical variable. ASA = American Society of Anesthesiologists; PROMIS = Patient-Reported Outcomes Measurement Information System.

Table 2.

Early follow-up PROMIS physical function scores stratified by trauma activation

Period	Overall group	Nontrauma activations	Trauma activations	p value
4 to 8 weeks	31 ± 8	31 ± 8	31 ± 8	0.58
10 to 14 weeks	38 ± 8	38 ± 8	37 ± 9	0.02
5 to 16 months	40 ± 10	41 ± 10	40 ± 10	< 0.01

All variables are reported as the mean ± SD. A t-test was used for comparing continuous viables. PROMIS = Patient-Reported Outcomes Measurement Information System.

Analytic Approach

Descriptive statistics were used to summarize patient characteristics and demonstrate demographic differences between patients with trauma activation and nontrauma activation patients. For demographic and outcome variable comparisons between groups, a t-test was used for continuous variables, while the chi-squared test was used for categorical variables. PROMIS PF scores were plotted by time from surgery, and local estimated scatterplot smoothing curve fitting was applied to empirically determine appropriate outcome measurement intervals. Bivariable analyses were conducted using a least-squares best-fit linear regression analysis to identify associations between patient characteristic variables and PROMIS PF scores. The multiple linear regression was used to control for covariates in the overall cohort. An additional multiple regression model was developed for the trauma activation group to determine the association between ISS and PROMIS PF scores. Confidence intervals are reported at 95%, and a p value < 0.05 was used to determine statistical significance. All statistical analyses were conducted using R Statistical Software (the R Foundation).

Results

Association of Anxiety and Physical Function

We found a strong association between early postoperative anxiety symptoms and late-recovery self-reported function. We found that for a 10-point increase in initial PROMIS anxiety score, there was an associated decrease in long term PROMIS PF (β = -2.64 [95% CI -3.07 to -2.21]; p < 0.001) (Table 3). When controlling for confounding variables, the relationship between PROMIS anxiety and PF scores remained in our overall cohort (β = -2.54 [95% CI -2.93 to -2.15]; p < 0.001) (Table 4). This relationship was also maintained in the trauma activation cohort (β = -2.71 [95% CI -3.19 to -2.23]; p < 0.001) (Table 5).

Table 3.

Bivariable regression for PROMIS physical function at 5 to 16 months

Factor	Estimate (95% CI)	p value
BMI	-2.56 (-3.20 to -1.93)	< 0.001
Age	-1.99 (-2.20 to -1.79)	< 0.001
Men gender	2.29 (1.48 to 3.10)	< 0.001
American Society of Anesthesiologists class	-4.79 (-5.23 to -4.35)	< 0.001
PROMIS anxiety	-2.64 (-3.07 to -2.21)	< 0.001

PROMIS = Patient-Reported Outcomes Measurement Information System.

Table 4.

Multivariable model for PROMIS physical function at 5 to 16 months

Factor	Estimate (95% CI)	p value
BMI	-0.85 (-1.51 to -0.19)	0.01
Age	-1.26 (-1.50 to -1.02)	< 0.001
Gender	0.95 (0.16 to 1.75)	0.02
American Society of Anesthesiologists class	-2.99 (-3.52 to -2.46)	< 0.001
PROMIS anxiety	-2.54 (-2.93 to -2.15)	< 0.001

PROMIS = Patient-Reported Outcomes Measurement Information System.

Table 5.

Multivariable model for PROMIS physical function at 5 to 16 months for the trauma activation subset

Factor	Estimate (95% CI)	p value
BMI	-1.15 (-1.95 to -0.34)	< 0.01
Age	-1.98 (-2.29 to -1.67)	< 0.001
Gender	1.15 (0.17 to 2.12)	0.02
American Society of Anesthesiologists class	-2.19 (-2.93 to -1.45)	< 0.01
PROMIS anxiety	-2.71 (-3.19 to -2.23)	< 0.001
Injury Severity Score	-1.45 (-2.11 to -0.79)	< 0.001

PROMIS = Patient-Reported Outcomes Measurement Information System.

Association of Demographics and Comorbidities With Physical Function

Increasing age, BMI, and ASA class were strongly associated with worse PROMIS PF during the long-term recovery time period, while men demonstrated an association with improved physical function (Table 3). After controlling for confounding, an increase in ASA score had a similar effect to a two-decade increase in age (β = -2.99 [95% CI -3.52 to -2.46]; p < 0.001 vs β = -1.26 [95% CI -1.50 to -1.02]; p < 0.001) (Table 4). BMI had a smaller effect (β = -0.85 [95% CI -1.51 to -0.19]; p = 0.01) as did men gender (β = 0.95 [95% CI 0.16 to 1.75]; p = 0.02).

Anxiety and Physical Function of Trauma Activation Patients

PROMIS anxiety scores and PROMIS PF scores were similar between the nontrauma activation and trauma activation patient cohorts. PROMIS anxiety was again independently associated with a lower self-reported physical function score. The overall anxiety scores between the trauma activation and nontrauma activations were the same (57 ± 10 versus 57 ± 10; p = 0.641) (Table 1). PROMIS PF was the same between the nontrauma activation and trauma activation cohorts during early recovery (31 ± 8 versus 31 ± 8; p = 0.575, mid-recovery (38 ± 8 versus 37 ± 9; p = 0.02), and late recovery periods (41 ± 10 versus 40 ± 10; p = 0.003). The association of the PROMIS anxiety was similar to our overall cohort (β = -2.71 [95% CI -3.19 to -2.23]; p < 0.001) even when accounting for injury severity score. Injury Severity Score was associated with an independent decrease in PROMIS physical function (β = -1.45 [95% CI -2.11 to -0.79]; p < 0.001) (Table 5).

Discussion

There is evidence that preoperative psychologic factors are associated with pain and functional outcomes in patients with orthopaedic conditions; however, this is poorly documented in patients who have sustained trauma, and there are barriers to effective screening and quantification of functional impact. We found initial anxiety scores were negatively associated with final self-reported functional scores. Older age, increasing BMI, and increasing ASA class also had a negative association with the final functional outcome. In these data, a 10-point (one standard deviation) increase in the initial anxiety score was associated with the equivalent functional score as a two-decade increase in age. The trauma activation subgroup had similar demographics and initial PROMIS anxiety scores to the overall group, and although ISS was independently associated with the final function, the PROMIS anxiety score coefficient remained similar.

Limitations

We were not able to differentiate between state anxiety and trait anxiety because of a lack of preoperative data on patient anxiety, which is a common weakness plaguing research in the trauma population. This impacts therapeutic treatments because we could not differentiate between anxiety caused by the acute traumatic incident and anxiety as a baseline trait. We also lacked demographic and injury-type data on patients treated at our institution who had no follow-up appointments, allowing a possible selection bias. Outcomes including complications, reoperations, pain scores, and opioid use were also not examined. The minimum clinically important difference in this population was not clearly defined; however, most PROMIS minimum clinically important differences have been determined to be approximately 5, because the tool is designed to have a standard deviation of 10 (one-half the SD definition of the minimum clinically important difference = approximately 5) [21]. In this study’s model, a 10-point increase in anxiety score was associated with a decrease in functional score (-2.5). This association may only be clinically relevant for patients reporting anxiety scores far higher than the population mean.

This study also lacked physician-reported outcome scores and relied solely on patient-reported outcomes. A patient’s anxiety influences their self-perceived function, and therefore, the effect reported here may not be owing to a difference in function (such as ambulation, stamina, and independence) but simply their own interpretation of their function. Although patient-reported outcomes are important, more objective outcomes (such as the timed-up-and-go test) can sometimes provide a clearer picture of function that is not clouded by the mental aspects of recovery.

Early Anxiety and Recovery of Physical Function

We found that higher levels of early postoperative anxiety were associated with lower levels of recovered physical function. Identifying variability in a patient’s functional outcome has been assumed to be partly because of psychosocial influences. Nunziato et al. [17] found that cognitive bias was associated with poor satisfaction while the magnitude of surgery did not affect pain, opioid refills, or satisfaction. The LEAP study group [6] found that in patients with severe lower extremity injuries, anxiety had an important role in the persistence of acute pain, but they did not evaluate the relationship with functional outcomes. A further study [26] prospectively studied the association of psychologic factors with disability and pain intensity in 136 patients who sustained trauma and found that catastrophic thinking in the early period was the sole significant predictor of disability in the late-recovery follow-up period. They also assessed for depression (Center for Epidemiological Studies Depression Scale), post-traumatic stress disorder (PTSD Checklist), and anxiety about pain (Pain Anxiety Symptom Scale Short Form 20); however, these were not significant in their models. Catastrophic thinking and anxiety are likely related, and we do not believe our findings contradict those of Vranceanu et al [26]. Similarly, a prospective study in the Netherlands of 101 patients who experienced trauma found a correlation between symptoms of depression and posttraumatic stress disorder at enrollment, as well as with disability at 5 to 8 months after trauma [16]. Future studies might compare types of anxiety, thinking patterns, and preexisting mental health symptoms with functional outcomes. Although this work [16] demonstrates one relationship, the precise psychologic determinants of recovery remain open to discussion.

Other Factors and Recovery of Physical Function

We found that BMI, age, and ASA class were negatively associated with lower levels of recovered physical function, while gender was not associated with a difference in physical function. The effect of an increase in ASA class was similar to a 10-point increase in anxiety score and approximately two decades of increased age. A 10-point increase in BMI was associated with approximately half the coefficient of one decade of age. These associations refer to absolute final, not recovered, self-reported function. Intuitively, weight, comorbidities, and age negatively impact a patient’s function, and this has been corroborated across a variety of disciplines and populations [1, 10, 15, 23]. More importantly, anxiety remains independently associated with lower physical function when accounting for these other associated variables.

Association of Trauma Activation With Recovery of Physical Function

We found that patients in a trauma activation setting at the time of initial presentation did not differ in terms of the relationship between anxiety and recovered physical function compared with patients who were not in a trauma activation setting. There are two differences in patients presenting as a trauma activation—their presenting mechanism or resuscitation and their care pathway. Patients presenting as trauma activations undergo advanced trauma life-support evaluations by our general surgery trauma team, which could be an independent stressor and induce perioperative anxiety. However, anxiety scores were not different between these two cohorts. The ISS was initially designed to predict survival and has been found to correlate with mortality, morbidity, and hospitalization [2, 4, 22, 25]. The inclusion of the ISS in our model did not change our primary finding that PROMIS anxiety was independently associated with lower physical function.

Conclusion

We demonstrated that initial patient self-reported anxiety scores are negatively associated with patient-reported physical function at the final follow-up in a broad cohort of patients undergoing surgery for orthopaedic lower extremity injuries. In our model, clinical relevance may only be reached in patients reporting anxiety well above (one to two standard deviations) the population mean. Identifying patients with high initial PROMIS anxiety scores may allow us to determine which patients will report lower functional scores at the final follow-up examination. This may guide future interventions to improve overall function after lower extremity trauma. Future investigations could focus on the effect of psychosocial interventions such as cognitive behavioral therapy and mindfulness on functional scores.