Loading [Contrib]/a11y/accessibility-menu.js
SurgiColl
SurgiColl
1.
Salman AA, Jafarnia JL, Ring DC, et al. Are Patient Demographics, Linguistic Tones and Patient Reported Outcome Measures Associated with Health Literacy? SurgiColl. 2025;3(4). doi:10.58616/​001c.143203

View more stats

Abstract

Objectives

Limited ability to obtain, process, and understand health information that enables patients to make health decisions (low health literacy) is associated with worse health and an increased risk of hospitalization. There is evidence that patients’ word choice can reflect illness behavior and care experience.

Correlation of linguistic tones and health literacy could help identify opportunities to ensure patient understanding and participation in decision-making during outpatient visits among patients with musculoskeletal illness.

Methods

A secondary analysis of transcripts of video and audio recordings of 65 adult patients seeking musculoskeletal specialty care was performed. Patients also completed questionnaires quantifying symptoms of depression (PROMIS [Patient-Reported Outcomes Measurement Information System] Depression computerized adaptive test [CAT]), PROMIS Pain Interference CAT (PI), PROMIS Upper Extremity CAT (UE), the Newest Vital Sign (NVS) health literacy questionnaire, and a basic demographics survey. Linguistic Inquiry and Word Count ( LIWC) was used to detect the relative strength of various emotional tones, cognitive processes, and core drives and needs. We tested for associations between health literacy and patient demographics, each of the LIWC domains, and PROMIS Depression, PROMIS PI, and PROMIS UE.

Results

Accounting for potential confounding in multivariable analysis, higher health literacy was associated with greater years of education, greater tones reflecting anxiety, and greater tones describing risk. There were correlations between more limited health literacy and greater pain interference and greater symptoms of depression, but not with upper extremity-specific capability.

Conclusion

The observation that patient linguistic tones are associated with health literacy can be used to develop effective health strategies consistent with what matters most to patients.

Health literacy is the ability to receive, process, and comprehend basic information required to make appropriate health decisions.1 Previous studies have shown that over a third of the United States (US) population has limited health literacy.2 Factors that correlate with lower health literacy include greater body mass index (BMI) and lower socioeconomic status, including greater distance between residence and nearest medical institution, lower monthly income, and lower education level.3

Natural language processing can analyze the content (themes) and tones in written language. It is possible that the language and tones a patient uses in a specialty visit correlate with levels of health literacy. If so, it might be possible to detect patient health literacy in the words and tone of their language, perhaps through computer analysis of their dialogue, and better strategize efforts to ensure understanding of their condition and the test and treatment options most consistent with their values.

Study Questions

Using a dataset of transcribed specialty care visits with corresponding measurements of patient health literacy, we asked: 1) Are there any patient linguistic tones or demographic factors associated with health literacy? 2) Are symptoms of depression, upper extremity magnitude of capability, and the degree to which pain interferes with capability associated with limited health literacy?

Methods

Study Design and Setting

With approval from our Institutional Review Board, we performed a secondary analysis of transcripts of audio or video recordings of 65 adult patients seeking musculoskeletal specialty care between November 2015 and March 2016, collected for other studies. In addition, participants completed a series of questionnaires after the visit.

All transcripts were separated into clinician and patient dialogue. Subsequently, Linguistic Inquiry and Word Count4,5 (LIWC, Pennebaker, University of Texas at Austin, USA) –an automated text analysis program— was used to detect the relative strength of various emotional tones, cognitive processes, and core drives and needs.6 LIWC can analyze written or transcribed verbal texts and compare each word against a user-defined dictionary. The percentage of words in each of the dictionary categories is calculated and compared to the total number of words in a text. For example, if LIWC identifies 15 words in the category ‘family’ (examples: aunt, uncle, nephew, niece) in a text with 100 words in total, the relative strength of this category would be 15%. For patient dialogue, we were interested in the following LIWC domains: the number of words, analytic thinking, confidence in understanding (clout), and authenticity, as well as the relative strength of tones consistent with positive emotion, negative emotion, anxiety, sadness, anger, social, family, cognitive process, insight, cause, tenacity, certainty, reward, and risk.

Measures

All patients were asked to complete the Patient Reported Outcome Measurement Information System (PROMIS) Depression computerized adaptive test (CAT), the PROMIS Pain Interference CAT (PI), the PROMIS Upper Extremity CAT (UE), the Newest Vital Sign (NVS) health literacy questionnaire, and a basic demographics survey (e.g., age, gender, marital status, work status, and years of education, etc.). PROMIS CATs are valid and reliable tools to assess a wide array of symptoms or limitations. People answer an average of 5 questions, with each new question based on the previously given answer, to arrive at a final score. A score of 50 represents the mean of the US general population, and every 10 points represents one standard deviation (SD).7

The PROMIS Depression CAT measures symptoms of depression on a continuum and is expressed as a t-score that is standardized to the US general population.8 A t-score of 50 represents the mean, and 10 above or below the mean represents one standard deviation. Higher scores indicate greater symptoms of depression. PROMIS questionnaires address illness on the continuum and in any context.9 The diagnosis of major depressive disorder can only be estimated with a questionnaire and is not a consideration in this analysis.

The PROMIS PI CAT (pain interference) instrument measures the effect of pain on all aspects of life: physical, mental, and social.10 In essence, it measures the extent to which pain hinders an individual’s engagement with physical, mental, cognitive, emotional, recreational, and social activities. The PROMIS UE CAT instrument measures activity intolerance of the upper extremity.11 To measure health literacy, we used the NVS health literacy test. This test is based on an ice cream container nutrition label. Patients can achieve a score ranging from 0 to 6, where higher scores indicate greater health literacy. For this study, we categorized health literacy into limited understanding (0-3) and adequate understanding (4-6). We used the same threshold as in the original study of Weiss et al and four other recent studies.12–16 An NVS score less than 4 has a sensitivity of 100% and a specificity of 64% for predicting limited health literacy.12

Patient characteristics

Sixty-five patients were included with an average age of 52 years (interquartile range (IQR): 39-62) [Table 1].

Table 1.Patient Demographics (N = 65)
Variables N (%)
N65
Gender
Women 24 (37)
Men 41 (63)
Marital
Single 22 (34)
Living with Partner3 (5)
Married 27 (42)
Separated or Divorced12 (18)
Widowed 1 (1)
Work Status
Working 38 (58)
Retired 11 (17)
Disabled/Unemployed 12 (18)
College4 (6)
Variable Median (IQR)
Age52 (39-62)
Years of Education16 (12-18)
PROMIS PI60 (54-64)
PROMIS Dep48 (45-53)
PROMIS UE 35 (30-40)

Continuous variables as median (interquartile range); discrete variables as percentage (number). PROMIS = Patient Reported Outcome Measurement Information System, UE = Upper Extremity, Dep= Depression, PI= pain interference

Most patients were employed (N=38, 58%), and the majority were married (N=27, 42%). The average score for symptoms of depression was 48, which is close to the average population of the USA.8,17 The median years of education in this study group were 16 (IQR 12-18). The average score on the PROMIS Pain Interference CAT was 60 (1 standard deviation above the US population), and the average score of the PROMIS Upper Extremity Physical Function score was 35 (1.5 SD below the US population).

Statistical Analysis

Data were reported as median and interquartile range (IQR) for continuous variables and numbers (percentages) for categorical variables. We sought bivariate associations between NVS health literacy scores and patient demographics, Health literacy scores and each of the LIWC domains, and health literacy scores and PROMIS Depression, PROMIS PI, and PROMIS UE using Spearman rank order tests for all non-parametric continuous variables, Kruskal-Wallis tests for all multi-group categorical variables, and Mann-Whitney U tests for categorical variables consisting of two groups.

A post hoc power analysis demonstrated that a sample of 61 patients provides 80% statistical power to detect a Pearson correlation of 0.35 or higher with alpha set at 0.05.

Results

Factors Associated with NVS Scores

Accounting for potential confounding between variables with P < 0.10 in bivariate analysis–including greater number of years of education [Appendix 1, Table 2], greater number of words used by patients, greater anxiety tones, and greater tones describing risk [Appendix 2, Table 3]–both greater anxiety and greater tones describing risk were independently associated with higher health literacy scores in multivariable analysis.

Table 2.Multivariable analysis of NVS scores and Years of education
Variable β Standard Error t P
Years of education 0.32 0.085 3.7 <0.001

Continuous variables as median (interquartile range); discrete variables as numbers. All significant correlations in bold. NVS= Newest Vital Sign

Table 3.Multivariable analysis of NVS scores and LIWC Dimensions
Variable β Standard Error t P
Word count 0.0076 0.00047 1.6 0.11
Anxiety 2.8 0.87 3.21 0.002
Risk 1.38 0.56 2.5 0.017

Continuous variables as median (interquartile range); discrete variables as numbers. All significant correlations in bold. NVS= Newest Vital Sign, LIWC= Linguistic Inquiry and Word Count

Relationships between Health Literacy and Capability and Symptoms of Depression

More limited health literacy was associated with greater pain interference (ρ -0.39) and greater symptoms of depression (ρ -0.37), but was not associated with upper extremity specific capability [Table 4].

Table 4.Correlations between Patient Reported Outcome Measures and NVS scores
PROMIS PI PROMIS Depression PROMIS UE
Median P Median P Median P
NVS Score 0.0013 0,0024 0,19
Limited 62 (56-68) 53 (48-58) 33 (26-38)
Adequate 59 (53-64) 46 (39-52) 36 (32-41)
ρ -0.39 -0.37 0.17

Continuous variables are presented as median (interquartile range [IQR]). PROMIS = Patient Reported Outcome Measurement Information System. PI = Pain Interference. Dep = Depression. UE = Upper Extremity, NVS=Newest Vital Sign, ρ = correlation coefficient. All significant correlations in bold.

Discussion

Limited ability to obtain, process, and understand health information, which enables patients to make health decisions (health literacy), is associated with worse health and an increased risk of hospitalization.18–21 There is evidence that a patient’s word choice can reflect illness behavior and care experience.22,23 For instance, certain words and concepts are associated with feelings of worry or despair and unhelpful thinking.24 Correlation of linguistic tones and Health Literacy could help identify opportunities to ensure patient understanding and participation in decision-making during outpatient visits among patients with musculoskeletal illness. We performed a linguistic analysis of transcripts of recordings of patients presenting for musculoskeletal specialty care. We found that a greater number of years of education, a greater number of words used by patients, higher levels of anxiety, and higher levels of risk perception were associated with greater health literacy scores.

This study has several limitations. All patients were enrolled from a single large urban area, which may limit generalizability, though our demographic data suggests that there was adequate diversity to identify statistical associations. Second, idiosyncratic elements of speech, such as stutters, repetition, pauses, nonverbals, and involuntary vocalizations, do not contribute to the measured linguistic tones. The verbal and nonverbal cues in combination with the actual words spoken — known to convey patients’ mood and coping strategies—might also better represent health literacy but are missed in this analysis. LIWC also does not pick up on sarcasm, idioms, or jokes, which may lead to misclassification of certain dialogue. This can be considered a first step in the attempt to use NLP to assess health literacy and its effect on care. Third, people behave differently under observation (Hawthorne effect),25 which could affect the topics that patients feel comfortable discussing with their physician,26 perhaps this is particularly true when people are being recorded. Although there is evidence that Hawthorne effects have a limited influence on experiments.27

The observation that greater health literacy was independently associated with greater tones of anxiety and greater tones describing risk may at first seem to contradict the evidence of the correlation of fewer symptoms of anxiety with greater health literacy.28–30 Perhaps people experiencing greater feelings of worry or despair are less able to articulate their feelings.31 This is consistent with evidence that people with more symptoms of depression tend to avoid expressing negative emotions.32 Not acknowledging one’s emotions and emotional avoidance strategies, such as emotional suppression, are mediators of a variety of psychological illnesses, including depression.33 Furthermore, ineffective cognitive coping strategies and the inability to regulate negative or aversive emotions are associated with symptoms of depression,34,35, and there is evidence that the lack of emotional clarity (i.e., the extent to which individuals can identify, label, and express emotions) can worsen depressive symptoms.34 It may sound counterintuitive that patients with greater symptoms of depression express less emotion. Still, it is in agreement with previous studies that indicated that emotional numbness and less expression of emotion in general can be indicative of major depressive disorder35—especially in men.36 One study found that among both undergraduate students and patients being treated for mood disorders, both inhibition of emotional expression and suppression of unwanted thoughts were associated with symptoms of depression and hopelessness.37 Or perhaps greater tones describing risk and anxiety reflect considerations specific to the symptoms, unrelated to general feelings of anxiety. The fact that these tones were picked up may relate to the fact that patients more involved with their health ask more questions13,16,38,39 and have longer visits16 in which more details are discussed.

The observation that more limited health literacy was associated with greater symptoms of depression and greater pain interference is consistent with previous research demonstrating that worse mental health is associated with more limited health literacy and worse health status.40 One key aspect of health literacy may be an understanding of the normal functioning of the human mind, mental short-cuts in particular. An awareness that the mind’s guessing apparatus—designed for quick, pressing decisions—may, at times, need to be overruled and ideas rethought (analytical/critical thinking) seems to be a key aspect of effective accommodation and good health. The construct of pain interference is strongly associated with unhelpful thinking patterns such as catastrophic thinking, kinesiophobia, and low self-efficacy.41 The ability to ameliorate unhelpful thinking may be a key aspect of health literacy.

The observation that a person’s language can reflect their health literacy may point to new approaches to health and health strategies. If clinicians or clinical tools, such as computer analysis of spoken language, facial expression, and other aspects of speech (tone, cadence, etc.), identify evidence of limited health literacy, the next thought might be about a person’s critical thinking skills and ability to move from their default mindset to a healthier mindset.

Conclusion

We can build on the evidence that unhelpful thoughts (misconceptions) about symptoms are associated with greater discomfort and incapability, understand that patient verbal and non-verbal cues can signal unhelpful thinking, and cultivate in all clinicians the ability to recognize unhelpful thinking and used crafted strategies for gently and incrementally reorienting important misconceptions, always mindful that suggesting new ways of thinking can be uncomfortable and are a risk to a healthy patient-clinician relationship.

Declaration of conflict of interest

The authors do NOT have any potential conflicts of interest for this manuscript.

Declaration of funding

The authors received NO financial support for the preparation, research, authorship, and publication of this manuscript.

Declaration of ethical approval for study

IRB Approval Number: 2019-03-0118

There is no information in the submitted manuscript that can be used to identify patients.

Submitted: May 05, 2025 EDT

Accepted: August 10, 2025 EDT

References

1.
Shah LC, West P, Bremmeyr K, Savoy-Moore RT. Health Literacy Instrument in Family Medicine: The “Newest Vital Sign” Ease of Use and Correlates. Published online 2005. doi:10.3122/​jabfm.2010.02.070278
Google Scholar
2.
Kutner, Greenberg, Jin, Paulsen. The Health Literacy of America’s Adults: Results From the 2003 National Assessment of Adult Literacy.; 2003.
3.
Xie Y, Ma M, Zhang Y, Tan X. Factors associated with health literacy in rural areas of Central China: Structural equation model. BMC Health Serv Res. 2019;19(1):300. doi:10.1186/​s12913-019-4094-1
Google Scholar
4.
Pennebaker JW, Booth RJ, Francis ME. Operator’s Manual: Linguistic Inquiry and Word Count - LIWC2007. DeptsTtuEdu; 2007.
Google Scholar
5.
Chung CK, Pennebaker JW. Linguistic Inquiry and Word Count (LIWC). In: Applied Natural Language Processing. ; 2013. doi:10.4018/​978-1-60960-741-8.ch012
Google Scholar
6.
Kahn JH, Tobin RM, Massey AE, Anderson JA. Measuring emotional expression with the Linguistic Inquiry and Word Count. American Journal of Psychology. Published online 2007. doi:10.2307/​20445398
Google Scholar
7.
Menendez ME, Bot AGJ, Hageman MGJS, Neuhaus V, Mudgal CS, Ring D. Computerized adaptive testing of psychological factors: Relation to upper-extremity disability. Journal of Bone and Joint Surgery - Series A. Published online 2013. doi:10.2106/​JBJS.L.01614
Google Scholar
8.
Pilkonis PA, Yu L, Dodds NE, Johnston KL, Maihoefer CC, Lawrence SM. Validation of the depression item bank from the Patient-Reported Outcomes Measurement Information System (PROMIS®) in a three-month observational study. J Psychiatr Res. 2014;56(1). doi:10.1016/​j.jpsychires.2014.05.010
Google Scholar
9.
Bernstein DN, Greenstein AS, D’Amore T, Mesfin A. Do PROMIS Physical Function, Pain Interference, and Depression Correlate to the Oswestry Disability Index and Neck Disability Index in Spine Trauma Patients? Spine (Phila Pa 1976). Published online 2020. doi:10.1097/​BRS.0000000000003376
Google Scholar
10.
Amtmann D, Cook KF, Jensen MP, et al. Development of a PROMIS item bank to measure pain interference. Pain. 2010;150(1). doi:10.1016/​j.pain.2010.04.025
Google Scholar
11.
Kaat AJ, Buckenmaier C “Trip,” Cook KF, et al. The expansion and validation of a new upper extremity item bank for the Patient-Reported Outcomes Measurement Information System® (PROMIS). J Patient Rep Outcomes. 2019;3(1). doi:10.1186/​s41687-019-0158-6
Google Scholar
12.
Weiss BD, Mays MZ, Martz W, et al. Quick assessment of literacy in primary care: The newest vital sign. Ann Fam Med. 2005;3(6). doi:10.1370/​afm.405
Google Scholar
13.
Menendez ME, van Hoorn BT, Mackert M, Donovan EE, Chen NC, Ring D. Patients With Limited Health Literacy Ask Fewer Questions During Office Visits With Hand Surgeons. Clin Orthop Relat Res. 2017;475(5). doi:10.1007/​s11999-016-5140-5
Google Scholar
14.
van Hoorn BT, Menendez ME, Mackert M, Donovan EE, van Heijl M, Ring D. Missed Empathic Opportunities During Hand Surgery Office Visits. Hand. 2021;16(5). doi:10.1177/​1558944719873395
Google Scholar
15.
Menendez ME, Mudgal CS, Jupiter JB, Ring D. Health literacy in hand surgery patients: A cross-sectional survey. Journal of Hand Surgery. 2015;40(4). doi:10.1016/​j.jhsa.2015.01.010
Google Scholar
16.
Menendez ME, Parrish RC, Ring D. Health Literacy and Time Spent with a Hand Surgeon. Journal of Hand Surgery. 2016;41(4). doi:10.1016/​j.jhsa.2015.12.031
Google Scholar
17.
Gershon R, Rothrock NE, Hanrahan RT, Jansky LJ, Harniss M, Riley W. The development of a clinical outcomes survey research application: Assessment centerSM. Quality of Life Research. Published online 2010. doi:10.1007/​s11136-010-9634-4
Google Scholar
18.
Koay K, Schofield P, Jefford M. Importance of health literacy in oncology. Asia Pac J Clin Oncol. 2012;8(1). doi:10.1111/​j.1743-7563.2012.01522.x
Google Scholar
19.
Balakrishnan MP, Herndon JB, Zhang J, Payton T, Shuster J, Carden DL. The Association of Health Literacy With Preventable Emergency Department Visits: A Cross-sectional Study. Academic Emergency Medicine. 2017;24(9). doi:10.1111/​acem.13244
Google Scholar
20.
Jansen T, Rademakers J, Waverijn G, Verheij R, Osborne R, Heijmans M. The role of health literacy in explaining the association between educational attainment and the use of out-of-hours primary care services in chronically ill people: A survey study. BMC Health Serv Res. 2018;18(1). doi:10.1186/​s12913-018-3197-4
Google Scholar
21.
Wright JP, Edwards GC, Goggins K, et al. Association of health literacy with postoperative outcomes in patients undergoing major abdominal surgery. JAMA Surg. 2018;153(2). doi:10.1001/​jamasurg.2017.3832
Google Scholar
22.
Ong LML, Visser MRM, Lammes FB, De Haes JCJM. Doctor-patient communication and cancer patients’ quality of life and satisfaction. Patient Educ Couns. Published online 2000. doi:10.1016/​S0738-3991(99)00108-1
Google Scholar
23.
Sloan TJ, Walsh DA. Explanatory and diagnostic labels and perceived prognosis in chronic low back pain. Spine (Phila Pa 1976). Published online 2010. doi:10.1097/​BRS.0b013e3181e089a9
Google Scholar
24.
Bot AGJ, Vranceanu AM, Herndon JH, Ring DC. Correspondence of patient word choice with psychologic factors in patients with upper extremity illness. Clinical Orthopaedics and Related Research. Published online 2012. doi:10.1007/​s11999-012-2436-y
Google Scholar
25.
Willemain TR. Model Formulation: What Experts Think About and When. Oper Res. Published online 1995. doi:10.1287/​opre.43.6.916
Google Scholar
26.
McHaney R, Tako A, Robinson S. Using LIWC to choose simulation approaches: A feasibility study. Decis Support Syst. Published online 2018. doi:10.1016/​j.dss.2018.04.002
Google Scholar
27.
Chiesa M, Hobbs S. Making sense of social research: How useful is the Hawthorne Effect? Eur J Soc Psychol. 2008;38(1). doi:10.1002/​ejsp.401
Google Scholar
28.
Kampouroglou G, Velonaki VS, Pavlopoulou I, et al. Parental anxiety in pediatric surgery consultations: the role of health literacy and need for information. J Pediatr Surg. 2020;55(4). doi:10.1016/​j.jpedsurg.2019.07.016
Google Scholar
29.
Kugbey N, Meyer-Weitz A, Oppong Asante K. Access to health information, health literacy and health-related quality of life among women living with breast cancer: Depression and anxiety as mediators. Patient Educ Couns. 2019;102(7). doi:10.1016/​j.pec.2019.02.014
Google Scholar
30.
Duplaga M, Grysztar M. The association between future anxiety, health literacy and the perception of the covid-19 pandemic: A cross-sectional study. Healthcare (Switzerland). 2021;9(1). doi:10.3390/​healthcare9010043
Google Scholar
31.
Al Salman A, Kim A, Mercado A, et al. Are Patient Linguistic Tones Associated with Mental Health and Perceived Clinician Empathy? Journal of Bone and Joint Surgery. 2021;103(23). doi:10.2106/​jbjs.21.00124
Google Scholar
32.
Campbell-Sills L, Barlow DH, Brown TA, Hofmann SG. Acceptability and suppression of negative emotion in anxiety and mood disorders. Emotion. Published online 2006. doi:10.1037/​1528-3542.6.4.587
Google Scholar
33.
Sydenham M, Beardwood J, Rimes KA. Beliefs about Emotions, Depression, Anxiety and Fatigue: A Mediational Analysis. Behavioural and Cognitive Psychotherapy. Published online 2017. doi:10.1017/​S1352465816000199
Google Scholar
34.
Thompson RJ, Kuppens P, Mata J, et al. Emotional clarity as a function of neuroticism and major depressive disorder. Emotion. Published online 2015. doi:10.1037/​emo0000067
Google Scholar
35.
Sydenham M, Beardwood J, Rimes KA. Beliefs about Emotions, Depression, Anxiety and Fatigue: A Mediational Analysis. Behavioural and Cognitive Psychotherapy. Published online 2017. doi:10.1017/​S1352465816000199
Google Scholar
36.
Oliffe JL, Phillips MJ. Men, depression and masculinities: A review and recommendations. J Mens Health. Published online 2008. doi:10.1016/​j.jomh.2008.03.016
Google Scholar
37.
Lynch TR, Robins CJ, Morse JQ, Krause ED. A mediational model relating affect intensity, emotion inhibition, and psychological distress. Behav Ther. Published online 2001. doi:10.1016/​S0005-7894(01)80034-4
Google Scholar
38.
Katz MG, Jacobson TA, Veledar E, Kripalani S. Patient literacy and question-asking behavior during the medical encounter: A mixed-methods analysis. J Gen Intern Med. 2007;22(6). doi:10.1007/​s11606-007-0184-6
Google Scholar
39.
Parikh NS, Parker RM, Nurss JR, Baker DW, Williams MV. Shame and health literacy: The unspoken connection. Patient Educ Couns. 1996;27(1). doi:10.1016/​0738-3991(95)00787-3
Google Scholar
40.
Gazmararian J, Baker D, Parker R, Blazer DG. A multivariate analysis of factors associated with depression: Evaluating the role of health literacy as a potential contributor. Arch Intern Med. 2000;160(21). doi:10.1001/​archinte.160.21.3307
Google Scholar
41.
Kortlever JTP, Janssen SJ, van Berckel MMG, Ring D, Vranceanu AM. What Is the Most Useful Questionnaire for Measurement of Coping Strategies in Response to Nociception? Clin Orthop Relat Res. 2015;473(11). doi:10.1007/​s11999-015-4419-2
Google Scholar

Appendices

Appendix 1.Bivariate Analysis between Patient Demographics and NVS Scores
Limited Health Literacy Adequate Health Literacy P
Gender 0.35
Women 5 19
Men 13 28
Marital 0.69
Single 7 15
Living with Partner 0 3
Married 8 19
Separated or Divorced 3 9
Widowed 0 1
Work Status 0.56
Working 9 29
Retired 4 7
Disabled/Unemployed 5 7
Collage 0 4
Age 49 (39-64) 52 (37-59) 0.84
Years of Education 12 (12-16) 16 (15-18) <0.001

Continuous variables as median (interquartile range); discrete variables as numbers. All significant correlations in bold.

Appendix 2.Correlations between LIWC Dimensions and NVS scores
LIWC Dimension Example Language Limited Health Literacy Adequate Health Literacy P
Word Count 302 (153-411) 420 (239-753) 0.047
Analytic 16 (10-27) 11 (8-22) 0.14
Clout 23 (14-37) 18 (11-29) 0.17
Authentic 70 (59-92) 82 (71-92) 0.26
Posemo love, nice sweet 5 (3-9) 4 (3-7) 0.92
Negemo hurt, ugly, nasty 1.3 (0.92-2) 1.7 (1.2-2.6) 0.13
Anxiety worried, fearful 0 (0-0) 0.24 (0-0.47) <0.001
Anger hate, kill, annoyed 0 (0-0.55) 0.1 (0-0.3) 0.72
Sad crying, grief, sad 0.3 (0-0.7) 0.42 (0-0.85) 0.49
Social mate, talk, they 5.5 (3.5-7.6) 4.9 (3.8-6.7) 0.62
Family daughter, dad, aunt 0 (0-0) 0 (0-0.16) 0.095
Cogproc cause, know, ought 10 (8-12) 13 (10-14) 0.006
Insight think, know 1.4 (0.7-3.3) 2.6 (1.8-3.2) 0.05
Cause because, effect 1.4 (0.65-2.4) 1.3 (0.61-2) 0.75
Tentat maybe, perhaps 2.4 (1.8-2.7) 2.9 (2.2-3.9) 0.062
Certain always, never 1.2 (0.32-1.7) 1.1 (0.6-1.8) 0.64
Reward take, prize, benefit 1.3 (0.65-1.9) 1.5 (0.93-2.3) 0.19
Risk danger, doubt 0.17 (0-0.37) 0.53 (0.24-0.78) 0.003

Continuous variables are presented as median (interquartile range [IQR]). LIWC = Linguistic Inquiry Word Count. Analytic = summary variable for analytic thinking. Clout = a summary variable that reflects the confidence in the text. Authentic = summary variable to understand the tone of discourse. All significant correlations in bold.