Psychosocial Factors Are Associated With Risk Acceptance in Upper Extremity Patients

Abstract

Background

Patients who help choose their health strategies are more adherent and achieve better health. An important role of the clinician is to verify that a patient’s expressed preferences are consistent with what matters most to the patient and not muddled by common misconceptions about symptoms or conditions. Patient choices are influenced by estimation of the potential benefits and potential harms of a given intervention. One method for quantifying these estimations is the concept of maximum acceptable risk (MAR), or the maximum risk that subjects are willing to accept in exchange for a given therapeutic benefit. This study addressed the hypothesis that misconceptions due to unhelpful cognitive bias regarding pain are associated with risk acceptance among people seeking care for an upper extremity condition.

Methods

We invited 140 new adult patients visiting an upper extremity specialist to complete a survey including demographics, pain intensity, depression and anxiety symptoms, catastrophic thinking, activity limitations, and MAR. Trauma or nontrauma diagnosis was obtained from the treating clinician and recorded by the research assistant. We used bivariate and linear regression analyses to identify factors associated with MAR among this population.

Results

Accounting for potential confounding in multivariable analysis, higher MAR was associated with older age and greater catastrophic thinking.

Conclusions

Specialists can be aware that people with more unhelpful cognitive biases may be willing to take more risk. Vigilance for common misconceptions and gentle, incremental reorientation of those misconceptions can increase the probability that people will choose options consistent with what matters most to them.

Introduction

When patients help choose and develop their health strategies, they are more adherent to their care and achieve better health.^1,2 To help guide patients, clinicians can make sure that a patients’ expressed preferences are consistent with what matters most to them (their values) and not muddled by common misconceptions about symptoms or conditions. One important factor that influences patient choices is a patient’s estimation of the potential benefits and potential harms of a given test or treatment. One method for quantifying this estimation is the concept of maximum acceptable risk (MAR),^3,4 defined as the maximum treatment-related risk that subjects are willing to accept in exchange for a given therapeutic benefit. ⁵

The concept of MAR is used to assess the probability of an adverse event that offsets treatment benefits.^6-9 In 1 study, women with irritable bowel syndrome were willing to tolerate a 2.6% increase in risk of bowel impaction, but only a 1.3% increase in risk of bowel perforation in exchange for alleviation of symptoms. ⁸ Another found out that psychiatrists were willing to accept a 4% increase in the risk of weight gain to obtain a 1-unit improvement of schizophrenia symptoms based on the positive and negative syndrome scale (PANSS). ⁹ Other studies in spine surgery, neurology, and rheumatology have assessed patient risk acceptance with respect to various treatments.^10-12 The U.S. Food and Drug Administration mentions MAR as 1 way of incorporating patient risk tolerance in the analysis of potential benefits and potential harms.^13-15

It’s possible that the estimation and understanding of potential harms and potential benefits are influenced by cognitive bias and emotions. This study addressed these relationships among people seeking care for an upper extremity condition.

Methods

This study was approved by our institutional review board. One hundred and forty new adult patients complete a survey on web-based Health Insurance Portability and Accountability Act of 1996 (Research Electronic Data Capture) system at the end of a visit with an upper extremity specialist in an urban U.S. city. The inclusion criteria were as follows: (1) age 18 years or greater; and (2) English fluency and literacy. Exclusion criteria were illiteracy and return patients. The diagnosis (trauma versus nontrauma) was recorded by the research assistant after asking the treating clinician (Table 1).

Table 1.

Patient and Clinical Characteristics.

Variables	mean plus/minus SD (range) N = 140
Age in years	50 ± 16 (18-86)
Woman, percentage (N)	54 (76)
Race/ethnicity, percentage (N)
White	67 (94)
Hispanic	13 (18)
Other	20 (28)
Marital status, percentage (N)
Married	55 (77)
Single	24 (33)
Divorced/separated/widowed	21 (30)
Level of education, percentage (N)
High school	18 (26)
2-year college	16 (22)
4-year college	41 (57)
Postcollege degree	25 (35)
Work status, percentage (N)
Employed	71 (99)
Other (student, retired, homemaker, unable to work)	29 (41)
Income, percentage (N)
Less than 30K	16 (22)
30-100K	39 (55)
Higher than 100K	45 (63)
Diagnosis, percentage (N)
Trauma	29 (41)
Nontrauma	71 (99)
Pain intensity, mean ± SD (range)	5.1 ± 2.5 (1-11)
PHQ-2 total, median (IQR)	2 (2-3)
Maximum acceptable risk, median (IQR)	21 (10-39)
PCS-4, median (IQR)	3 (1-5)
QuickDASH, median (IQR)	28 (14-45)

Note. Continuous variables as mean ± standard deviation (range) or median (IQR); Discrete variables as percentage (number). SD = Standard Deviation; IQR = interquartile range; PHQ-2 = Patient Health Questionnaire; PCS-4 = Pain Catastrophizing Scale; QuickDASH = Quick Disabilities of the Arm, Shoulder, and Hand.

Reading the recruitment letter and completing the questionnaires represented informed consent. Patients answered questions about their demographics (age, sex, race, marital status, level of education, work status, and income); a pain intensity numerical scale (with a range of 0 representing no pain up to 10 representing the worst imaginable pain); The Patient Health Questionnaire (PHQ-2) which is a 2-item measure to assess symptoms of depression in the past 2 weeks with a range of 0 to 6; the Pain Catastrophizing Scale (PCS-4) is a 4-item measure of worst-case or catastrophic thinking in response to pain; and the QuickDASH is an abbreviated form of original Disabilities of the Arm, Shoulder, and Hand (DASH) which measures upper extremity specific magnitude of limitations.

The MAR questionnaire asks patients to rate the maximum risk they are willing to accept for 1 unit improvement in their symptoms or function with a range of 0% to 100%. The median MAR in our population was 21%, which is similar to another study on rheumatoid arthritis patients that showed patients are willing to accept an 18% risk increase in serious infection to achieve a 50% improvement in their physical function. ¹⁶

Statistical Analysis

An a priori power analysis indicated that a sample of 136 subjects would provide 80% statistical power, with alpha set at 0.05, for a regression with 6 predictors if 1 of the factors would account for 5% or more of the variability in pain intensity and our complete model would account for 15% of the overall variability.

The descriptive analysis of the demographics, pain intensity, PCS-4, PHQ-2, MAR, and QuickDASH and their mean and median, range, standard deviation (SD), interquartile range (IQR), and percentages was performed based on distribution pattern; for normally distributed variables mean and SD and for nonnormally distributed variables, median, and IQR reported.

Associations between MAR (nonnormal continuous) with dichotomous variables were tested using the Mann-Whitney Test. The association between MAR and continuous variables were tested by Spearman’s rank correlation coefficient test. For the association between MAR and nominal variables, we used Kruskal-Wallis test. An alpha level of 0.05 was used to determine statistical significance. All variables with P < .1 were included in a linear regression model. We omitted pain intensity and self-reported activity limitation in multivariable linear regression model because we consider them response variables, and we know from prior work that they are colinear.

Results

In bivariate analysis, there were small correlations between greater risk acceptance (MAR) and older age, higher QuickDASH scores (greater magnitude of self-reported activity limitations), higher pain intensity, and greater worst-case thinking (higher PCS-4 scores; Table 2). Accounting for potential confounding in multivariable analysis, older age (coefficient: 0.27; P-value = .02), and greater worst-case thinking (coefficient: 1.7; P-value = .003) were associated with greater risk acceptance (Table 3).

Table 2.

Bivariate Analysis of Factors Associated With Maximum Acceptable Risk.

Variable		P value
Age in years, Spearman’s Rho	0.17	.04
Sex, mean± SD		.8
Woman	27 ± 23
Man	27 ± 23
Race/ethnicity
White	28 ± 24
Hispanic	27 ± 25
Other	26 ± 18	.99
Marital status, mean± SD
Married	29 ± 23
Single	20 ± 20
Divorced/separated/widowed	32 ± 26	.06
Level of education, mean± SD
High school or less	29 ± 18	.10
2-year college	37 ± 29
4-year college	23 ± 23
Postcollege degree	27 ± 23
Work status, mean± SD
Employed	26 ± 21
Other (student, retired, homemaker, unable to work)	30 ± 27	.82
Income, mean± SD
Less than 30K	28 ± 27
30-100K	30 ± 24	.57
More than 100K	25 ± 21
Diagnosis, mean± SD
Trauma	28 ± 24	.82
Nontrauma	26 ± 21
PHQ-2 total, Spearman’s Rho	0.007	.94
QuickDASH, Spearman’s Rho	0.23	<.01
Pain intensity, Spearman’s Rho	0.20	.02
PCS4-total, Spearman’s Rho	0.18	.03

Note. Bold values show significance. SD = Standard Deviation; PHQ-2 = Patient Health Questionnaire; PCS-4 = Pain Catastrophizing Scale; QuickDASH = Quick Disabilities of the Arm, Shoulder, and Hand.

Table 3.

Linear Regression Analysis of Factors Associated With Maximum Acceptable Risk.

Dependent variable	Retained variable	Coefficient (95% interval)	Standard error	P value	Semipartial R2	R 2
Maximum acceptable risk	Age in years	0.27 (0.03-0.5)	0.12	.02	0.03	0.09
	PCS4-total	1.7 (0.57-2.7)	0.55	.003	0.06

Note. PCS-4 = Pain Catastrophizing Scale.

Bold values show significance.

Discussion

Prior studies of factors associated with risk acceptance have mostly focused on symptoms rather than psychological factors. We studied new patients after a single musculoskeletal specialty visit to assess psychological factors related to risk acceptance in upper extremity care. We found that greater risk acceptance was related to greater cognitive bias regarding pain (catastrophic or worst-case thinking).

The findings of this study can be interpreted in light of some limitations. The small number of people (approximately 10) that declined participating in this study might vary with respect to pain intensity, greater self-reported activity limitations, and catastrophic thinking, which could skew the results. We think this is unlikely given the small number of declines. Since our instruments were in English, the lack of non-English-speaking patients limits generalizability. We used a general question rather than a scenario specific to the patient’s problem, and served as hypotheticals. It might be better applied to actual rather than fictive scenarios. We studied risk across various diagnoses as an initial, practical step to learn more about factors associated with risk acceptance, and the results might be different in a study of a specific disease. Another limitation is that most of our patients had a condition and choices that involve relatively little risk. This is typical for musculoskeletal concerns and therefore of interest to us. A study of the relationship between cognitive bias and risk acceptance might have different results in settings where patients face greater risks.

The finding that greater tendency toward the cognitive bias of worst-case thinking is associated with greater risk acceptance among people seeking care for hand and upper extremity conditions is consistent with what one might expect: people who are feeling less hopeful and in control are more willing to take risks. There is relatively little study of relationship of cognitive bias and risk acceptance, and it might merit further investigation. Worst case thoughts in response to pain are associated with greater pain intensity and greater magnitude of limitations^17,18 which are associated with increased acceptance of risk in prior studies. A prior study among people with low back pain found greater pain intensity correlated with greater acceptance of operative risks. ¹⁰ Prior studies among people with Crohn’s disease, multiple sclerosis, and rheumatoid arthritis found greater risk acceptance among people that reported greater limitations.^11,12,19 One study in people with peripheral vascular disease found no association between pain or limitations and risk acceptance. ²⁰ The slight correlation between older age and risk acceptance is difficult to interpret, was inconsistent in prior studies, and may be spurious. A study of people with multiple sclerosis found the same small correlation with older age, ²¹ while studies of people considering spine and vascular surgery, and among people seeking care for gastrointestinal symptoms found no relationship to age^10,22,23 and studies of people with rheumatoid arthritis and health risk taking found less risk acceptance with older age.^12,24

Conclusion

Our findings suggest that people with more unhelpful cognitive biases such as worst-case thinking are more willing to accept more potential harms to try to improve their health. When a surgeon’s offer of discretionary surgery is accepted by a patient, the surgeon tends to consider that decision rational and informed, but this line of evidence suggests such decisions are based to some degree on misconceptions, and potentially rooted in elements of despair rather than measured consideration of the potential harms and potential benefits. People who are willing to accept more potential harms might benefit from greater attention to ensuring that they understand the chances of benefit and the chances of harm accurately. The accumulating evidence suggests that what we should be aiming for is:

The patient can clearly articulate what matters most to him/her and express a good understanding of the potential benefits and the potential harms, and the patient’s understanding is consistent with the available evidence and is not distorted by cognitive bias, stress, or psychological distress.

A longitudinal study regarding how people reflect on worst- case thoughts 1 year after specialty consultation would be interesting.