Delivery of Hand Care to Patients With High Anxiety Burden

Abstract

Purpose

To determine whether patients presenting with Patient-Reported Outcome Measurement Information System (PROMIS) anxiety scores at >95th percentile of the US population undergo elective hand surgery at rates different from patients with less anxiety. Secondarily, we aimed to assess surgeon notation of these patients’ emotional states and incidences of postoperative complications.

Methods

This single-center retrospective cohort study analyzed data from new adult patients presenting for hand care between January 2019 and December 2020. Patients with initial PROMIS anxiety scores ≥70 were identified as the high anxiety burden (HAB) group. For each patient with HAB presenting with carpal tunnel syndrome, trigger finger, and distal radius fractures (n = 45), three controls were matched (n = 135). After matching, bivariate statistical analysis compared outcome variables of interest between patient groups. Sentiment analysis was used to explore if patient anxiety was realized and considered in surgical decision making.

Results

After matching, patients with HAB averaged baseline PROMIS scores in every assessed domain that were ≥1 SD worse than unaffected patients. Patients with HAB and control patients were offered surgery at a comparable rate (58% vs 47%). Among those offered surgery, patients with HAB were less likely to undergo surgery (73% vs 92%). Postoperative complications occurred more frequently in the patients with HAB (32% vs 8%). Sentiment analysis of office notes revealed that the patients with HAB had their emotional state explicitly noted more often (40% vs 24%), but the majority of patients with HAB did not have this addressed in records.

Conclusions

Patients with HAB both underwent surgery offered less frequently and when undergoing surgery, experienced more complications. Surgeons are likely to miss opportunities to positively influence extreme patient anxiety as most patients with HAB were treated without mention of their emotional state. Future investigations should explore whether preoperative anxiety alleviation could diminish these disparities.

Type of study/level of evidence

Symptom Prevalence III.

Anxiety disorders affect 40 million adults annually in the United States, making it the most common mental illness in the country.¹ This may be heightened among patients with symptomatic hand conditions because 1 in 5 patients presenting to a hand surgeon reports a degree of anxiety that could itself warrant intervention.² Anxiety, depression, and catastrophization have a known influence on patients’ quality of life and are recognized as potential factors influencing overall health and outcomes of hand surgery. For example, greater pain interference and the use of antidepressants early after distal radius fractures are predictive of less patient-reported function at 6–9 months after treatment³.

Although mounting evidence supports that anxiety and depressive symptoms have a deleterious effect on reported function and pain after hand surgery, the effect of patient psychological state on surgical decision-making remains incompletely understood.^4–6 Crijns et al⁷ determined that higher Patient-Reported Outcome Measurement Information System (PROMIS) pain interference scores were independently associated with higher rates of elective surgical treatment. Although experience with pain and anxiety exists on a continuous spectrum, and no universal threshold exists to define who might benefit from directed mental health treatment, this study was conducted to determine how markedly elevated anxiety could influence hand surgical treatment.

This study aimed to determine whether patients presenting with PROMIS scores at >95th percentile of the US population in the anxiety domain, and therefore reporting a high burden of anxiety symptoms, undergo elective hand surgery at a different rate than patients presenting with less anxiety. Secondary objectives included assessing variation in acknowledgment of patient emotional states by hand surgeons for these patients and assessing for differential postoperative complications between patient groups.

METHODS

This was a single-center retrospective cohort study. Initial data for all new patients evaluated by one of seven fellowship-trained hand surgeons at a tertiary center from January 2019 through December 2020 were collected from the electronic medical health record. The data set included all adult patients (≥18 years of age) who completed routine PROMIS anxiety (v1.0), depression (v1.0), pain interference (v1.1), and upper-extremity function (v2.0) computer adaptive tests at presentation. On all PROMIS scales, 50 represents the US population normal t score with a SD of 10.⁸ Higher scores represent more of the domain assessed (eg, higher anxiety scores are more anxiety and higher upper-extremity function scores are more function). Patients who required emergent surgery or had concomitant orthopedic injuries were excluded from our analysis. Our institutional review board deemed this study exempt because it examined data from existing health records.

Patients with PROMIS anxiety scores ≥70 (2 SD above the mean) at presentation comprised the high anxiety burden (HAB) group.⁹ This group was compared with all other new patients according to demographic data, International Classification of Diseases, 10th Revision diagnostic codes and PROMIS scores.¹⁰ The HAB group was subdivided into the three common specific diagnoses. A group of three matched control subjects per patient with HAB was identified for each diagnostic subgroup (carpal tunnel syndrome [CTS], trigger finger, and distal radius fracture) according to primary diagnosis, sex, race, age (± 5 years), and treating surgeon.

The hand conditions prompting evaluations were characterized by the severity of pathology and the duration of symptoms. Patients in the CTS subgroup were categorized into three stages as described by Alfonso et al.¹¹ That system distinguishes predominately nocturnal symptoms (stage 1) from those with additional daytime symptoms or subjective weakness (stage 2) with a final category for those with thenar atrophy or a loss of two-point discrimination (stage 3). Patients with trigger fingers were categorized based on the Green classification with grade 1 being pain and tenderness at the A-1 pulley, grade 2 exhibiting catching digits, grade 3 signifying patients with locking of a digit that was passively correctable, and grade 4 denoting patients with a fixed locked digit.¹² Distal radius fractures were described based on the presence or absence of displacement and intraarticular extension. Postoperative complications were identified through a manual chart review to identify any surgical or medical occurrences deviating from the normal course. A priori, infection was defined as requiring antibiotics or debridement.

Sentiment analysis

Sentiment analysis describes the study of opinions, evaluations, attitudes, and emotions as expressed in natural language text.¹³ Medical sentiment refers to attitude, thought, or judgment with respect to the health of an individual expressed through language (eg, judgment, vagueness, certainty, etc) concerning a medical condition and its appearances or consequences for an individual.¹⁴ Sentiment scores of nursing notes have been demonstrated to be a predictor of patient mortality and sepsis.^13,15,16 Additionally, social psychology literature has previously demonstrated that biased language in clinical notes has been shown to affect subsequent clinician attitudes and decision-making.^17–20

In this study, sentiment analysis was used to assess clinician attitude and surgical decision-making to explore if patient anxiety was realized and if it appeared related to surgical decision-making. Manual review based on previously defined criteria for sentiment analysis has been suggested as the reference standard.^13,21 All clinic notes underwent sentiment analysis for both groups by one author (C.D.C.). Each clinical note was evaluated based on predefined criteria for presence or absence of qualifying language related to the patient’s medical condition. Notes were scored as either having a surgeon acknowledge the patient’s emotional state or not. This was supplemented by a summative count of the number of positive or negative qualifiers.

Statistical methods

Descriptive statistics were generated to describe the study participant groups. Unpaired Student t tests and chi-square tests were used for bivariate analyses comparing variables between the patients with HAB and all other patients. Variables included demographic data, PROMIS scores, and presenting diagnoses.

Once matching groups (1:3) were established among patients with CTS, trigger finger, and distal radius fractures, the groups were compared using summary bivariate statistics. Variables compared included PROMIS scores, frequency of explicit acknowledgment of the patient emotional state on sentiment analysis, frequency of surgery being recommended, percentage of patients undergoing recommended surgery, and incidence of postoperative complications. Statistical significance was set at P < .05. PROMIS score differences were considered clinically relevant beyond a threshold of 4 points based on the minimal clinically important difference.^22–24

For the sentiment analysis, a subset of 15 patient records were analyzed by a second reviewer (S.C.R.). The absolute agreement for all measures of the qualitative content analysis between reviewers was >80%. The Cohen’s kappa (κ = 0.80) for the sentiment analysis of surgeon sentiment score indicated moderate interrater reliability.²⁵ Qualifier counts had moderate reliability for positive qualifiers (κ = 0.65) and negative qualifiers (κ = 0.75). Each of these reliability statistics indicated acceptable reliability for inclusion of these data.

This study was designed based on a sample size calculation to power our primary statistical analysis of surgical rates between groups. Designed for 3:1 matching, with α = 0.05, power of 0.8, and a meaningful difference in surgical rates of 40% versus 60%, the study would require 65 patients with HAB and 195 controls. We chose the time window for inclusion to capture approximately 10,000 new patient visits, which was anticipated to yield 250 patients (2.5%) with PROMIS anxiety scores ≥2SD above the mean. Presuming 250 patients with HAB, we estimated that approximately one-quarter of them (n = 63) would have presented a common diagnosis suitable for matching. This statistical analysis was planned presuming that we would be combining all patients with one of several common diagnoses used in matching. We anticipated qualitative-only presentation of data from smaller individual diagnosis groups.

RESULTS

Among 9,344 new adult patients, 211 (2.3%) of patients presented with PROMIS anxiety scores ≥70. Compared with unaffected new patients seen during the study period, the high anxiety group comprised a greater proportion of Black (30% compared with 13% of control patients, P < .05) and women (66% compared with 58%, P < .05). High anxiety patients averaged baseline PROMIS scores in every assessed domain that were ≥1 SD (≥10 points) worse than unaffected patients despite being similar in age and distribution of common diagnoses (Appendix 1, available online on the Journal’s website at www.jhandsurg.org).

Forty-five patients with HAB with CTS, trigger finger, and distal radius fracture were matched with 135 controls (Table 1). Initial visit PROMIS scores were substantially worse for patients with HAB in all diagnostic subgroups, but there was no consistent evidence of differential condition severity or duration (Table 2). Patients in both groups were offered surgery at a comparable rate (58% vs 47%, P = .23). Among those offered surgery, patients with HAB were less likely to undergo surgery (73% vs 92%, P < .05). Complications following surgery occurred more frequently in the HAB group (32% vs 8%, P < .05). Sentiment analysis of office notes revealed that patients with HAB were significantly more likely to have a surgeon implicitly or explicitly acknowledge a patient’s emotional state (40% vs 24%, P < .05). In these notes, patients with HAB averaged 2.3 negative and 0.6 positive adjectives/adverbs while control patients’ notes averaged 1.4 positive and 1.3 negative adjectives/adverbs.

TABLE 1.

Demographic Data From Patients With High Anxiety Burden and Matched Controls

Variable	HAB (n = 45)	Control (n = 135)
Age (y), mean (SD)	55 (14)	55 (14)
Female (%)	27 (60%)	84 (62%)
Non-Hispanic White (%)	28 (62%)	84 (62%)
Black (%)	16 (36%)	48 (36%)
Currently smoking (%)	15 (33%)	22 (16%)
Currently using alcohol (%)	12 (27%)	76 (56%)
Initial PROMIS scores: mean (SD)
Anxiety	74 (4)	52 (10)
Depression	65 (8)	47 (9)
Pain interference	70 (7)	61 (7)
Upper-extremity function	25 (9)	42 (11)
Offered surgery (%)	26 (58%)	64 (47%)
Underwent surgery offered (%)	19 (73%)	59 (92%)
Postoperative complication (%)	6 (32%)	5 (8%)

TABLE 2.

Grading of Diagnosis Severity According to Patient Group

Diagnosis Grading and Duration	HAB	Controls
Carpal tunnel stage11	n = 18	n = 54
1—Predominantly night symptoms	3 (17%)	19 (35%)
2—Daytime manifestations and weakness	12 (67%)	20 (37%)
3—Atrophy and sensory loss	3 (17%)	15 (28%)
Symptoms present ≥ 1 y	14 (78%)	20 (37%)
Trigger finger Green classification 12	n = 13	n = 39
1—Tenderness	2 (15%)	19 (49%)
2—Catching not locking	5 (38%)	14 (36%)
3—Locking correctable	4 (31%)	4 (10%)
4—Fixed locked	2 (15%)	2 (5%)
Symptoms present ≥ 1 y	1 (8%)	15 (38%)
Distal radius fracture description	n = 14	n = 42
Intra-articular	10 (71%)	22 (52%)
Displaced	11 (79%)	33 (79%)

Qualitative/exploratory individual diagnoses data

The CTS subgroup comprised 18 patients with HAB and 54 controls. At presentation, there was no substantial difference in the CTS between groups, although 77% of patients with HAB reported greater than 12 months of symptoms vs 37% of controls. Patients with HAB were offered surgery at least as frequently as controls (72% vs 54%). Among patients offered surgery, eight of 13 (62%) patients with HAB underwent surgery versus 24 of 29 (83%) controls. Five of 13 (38%) patients in the HAB group did not undergo surgery offered with two requesting a delay but never returning, and three calling to cancel shortly before surgery. In the control group, five of 29 (17%) patients offered surgery requested delayed treatment with temporizing injection and did not return. After surgery, three of eight patients (38%) in the HAB group developed diffuse pain and hypersensitivity. In the control group, three of 24 (13%) patients developed pillar pain. Sentiment analysis revealed that eight of 18 (44%) patients with HAB had their emotional state acknowledged versus 13 of 54 (24%) controls.

Thirteen patients in the HAB group with trigger fingers matched with 39 controls. At presentation, 11 of 13 (85%) patients with HAB reported some catching or locking as opposed to tenderness alone versus 51% of controls. One patient (8%) with HAB presented with symptoms for at least 1 year versus 47% of controls. Patients with HAB were offered surgery at least as frequently as controls (five of 13 [38%] vs 33% of controls). Two of five (40%) patients with HAB declined surgery in favor of nonsurgical management and did not return. All 13 control patients who were offered surgery chose to proceed with surgical treatment. After surgery, there were no reported complications in either group. Sentiment analysis revealed that six of 13 (46%) of patients with HAB had the patient’s emotional state acknowledged versus 23% of controls.

Fourteen patients with HAB and acute distal radius fractures were matched with 42 control patients. In each group, 79% of fractures were displaced. Patients with HAB were offered surgery at a comparable rate with controls (57% vs 52%). All patients in each group who were recommended to proceed with surgical fixation underwent surgery. Three of 8 (38%) patients with HAB experienced complications. These included one with deep venous thrombosis with subsequent hand hypersensitivity, one with excessive postoperative swelling, and another who developed recurrent malalignment. Two of 8 (25%) patients with HAB underwent eventual hardware removal. In the control group, two of 22 patients (9%) experienced complications. One developed delayed CTS, which was surgically released 5 months after the distal radius fracture while the second developed cellulitis at the incision site, which resolved with antibiotics. Sentiment analysis revealed that four of 14 (29%) of patients with HAB had patient’s emotional state acknowledged versus 24% of controls.

DISCUSSION

In this analysis, patients with HAB were offered hand surgery for common diagnoses at comparable rates with unaffected patients. However, patients with HAB were less likely to undergo surgery and were more likely to experience postoperative complications. Furthermore, surgeons did not explicitly note the emotional state of the majority of patients with HAB. Interestingly, it seemed that surgeons were more likely to appreciate this in patients presenting for elective conditions. This may reflect how patient affect and emotional state are more recognizable when assessing atraumatic conditions as opposed to following acute fracture when most patients are in substantial pain and treatment may depend more on injury characteristics than patient emotional state.

Although not recording the patients’ emotional state may represent an omission, it likely signifies a missed opportunity for potentially beneficial intervention. Empathetic opportunities occur during 70% of hand surgery visits, but surgeons offered empathetic responses in only half of those instances.²⁶ At the same time, providing routine musculoskeletal care for orthopedic patients has neither substantially improved patient mental health nor corrected the gap in patient-reported health associated with patient anxiety/depression.^27–29 In fact, most patients sampled from our department preferred their mental health to be acknowledged, and existing literature supports the association between perceived empathy and clinical outcomes.^28,30–32 However, barriers exist because surgeons face time constraints, limited training for addressing psychosocial issues, and there is a lack of available mental health specialists.²⁸ Furthermore, recording impressions of patients’ emotional state risks patient perception of negative bias in notes. We are aware of one unsuccessful lawsuit related to a patient feeling that a hand surgeon diagnosed them with anxiety in the record. This underscores that surgeons may want to record symptoms and refer them for treatment without indicating a formal anxiety diagnosis. Despite this rare event, we suggest that ideal treatment would incorporate a holistic approach to improve mental and physical health among patients seeking hand care.

Before this study, it was unclear if patients with HAB would be offered surgery more often due to greater reported pain and functional difficulties compared with others or be offered surgery less often due to perceived disproportionate pain or concern over suboptimal outcomes. The finding that patients with HAB and controls were offered surgery at similar rates, although a greater proportion of high anxiety patients never underwent surgery, is consistent with literature reporting that patients with anxiety are more likely to delay seeking and obtaining treatment.³³ This may indicate a missed opportunity to help these patients effectively access hand care. As these patients who did not arrive for surgery never followed up, we cannot provide reasons behind their decisions. However, it would be reasonable to suspect that surgeons who reduce patient anxiety may improve the chances of patients following through with treatment. To that end, a previous study showed that empathetic and optimistic discussions increased trust in clinicians treating osteoarthritis.³⁴ Similarly, warm empathetic communications with positive expectations reduced women’s anxiety in general practice³⁵.

The fact that postoperative complications occurred more frequently in patients with HAB is consistent with documented associations between elevated anxiety and poor surgical outcomes.^36,37 Patients with greater pain catastrophizing have higher rates of kinesiophobia and elevated potential for disability after hand injuries.^38,39 Preoperative anxiety and negative affect have also been associated with worse pain and function after foot and ankle surgery as well as complex regional pain syndrome after knee arthroplasty.^37,40,41 This association has been postulated to occur via central sensitization to pain. Most of the complications that occurred in this study were conceivably related to patient anxiety. These included diffuse pain and hypersensitivity after carpal tunnel release and may also include swelling and hypersensitivity after distal radius fracture, despite this being complicated by diagnosed venous thrombosis in one case. Although we cannot determine the exact cause of thrombosis, in our experience, highly anxious patients are often reluctant to move their injured extremities, and as such, we wonder if anxiety contributed to this rare thrombotic complication. Given the elevated risk of complicated postoperative recovery for patients with high levels of anxiety, surgeons may consider referrals for anxiety treatment or recommend simple nonpharmacological interventions for perioperative anxiety management.⁴²

This study has several limitations. First, this single-center sample may not generalize to all practices. Second, statistical analysis was limited to the entire matched cohorts combining three diagnoses as opposed to an analysis for each individual diagnosis. Third, patient matching did not include medical comorbidities; hence, it is possible that differences in general health could confound the between-group findings. Fourth, although our matched groups were smaller than our estimated sample size goal, we did not expand our data retrieval because the data demonstrated significant differences among outcomes of interest. Fifth, the classification of the surgeon sentiment score was undertaken manually, although personal review has previously been suggested as the reference standard and is the basis for automated methods of sentiment analysis.¹³ Finally, some study data were collected during the COVID pandemic, which influenced the mental health of many individuals. However, among patients accessing our department during this time, there were no systematic alterations in patient-reported mental and physical health.⁴³

Surgeons are likely to miss opportunities to positively influence extreme patient anxiety. We recommend routine anxiety screening as a first step for surgeons to become aware of patients’ anxiety levels. This can be as simple as administering the Generalized Anxiety Disorder-2 Scale, which is a validated two-question survey.⁴⁴ Although self-directed tools for anxiety reduction are available, we most frequently contact patients’ primary doctors to suggest anxiety treatment. We view this as analogous to hand surgeons inquiring about glucose control or nicotine use. Although surgeons do not typically manage such medical and substance comorbidities, surgeon inquiry and facilitation of better glycemic control or nicotine cessation can improve patient health and surgical outcomes. Future investigations targeting preoperative anxiety reduction should be considered because this could improve access to surgery and reduce the incidence of surgical complications in patients with HAB.