To the Editor
Safety net clinics serve limited English proficiency (LEP) and limited health literacy (LHL) populations who experience communication barriers that contribute to disparities in care and health.1 Safety net electronic health record (EHR) implementation may affect patient-provider communication.2 We studied associations between clinician computer use and safety net communication with diverse chronic disease patients.
Methods
This IRB-approved observational study occurred 2011-2013 at an academically-affiliated public hospital with a basic EHR for reviewing results, tracking health care maintenance, prescribing, and referring. Some clinics (internal medicine and diabetes) required typed visit documentation, which was optional in others (family medicine, cardiology, and rheumatology).
Eligible English-/Spanish-speaking adults had specific chronic conditions and received primary AND subspecialty care (Table 1). Physicians, nurse practitioners, fellows, and residents could decline participation or designate ineligible patients. Research assistants enrolled and interviewed patients by phone before appointments, videotaped the subsequent visit, and interviewed patients post-visit. Clinician participants completed paper or online questionnaires.
Table 1. Patient and Clinicians in a Study of Communication Behaviors by Clinician Computer Use in Safety Net Encounters.
| Patients (n=47) | |
| Mean age, years (SD) | 56.5 (11.4) |
| Women, n (%) | 26 (55.3) |
| Self-reported race/ethnicity | |
| Hispanic, n (%) | 27 (57.5) |
| African-American, n (%) | 8 (17.0) |
| Caucasian, n (%) | 3 (6.4) |
| Asian, n (%) | 7 (14.9) |
| Multiethnic, n (%) | 2 (4.3) |
| Primary Language Spanish, n (%) | 26 (55.3) |
| Limited English proficiency* | 13 (27.7) |
| Education, n (%) | 12 (25.5) |
| ≤ 8th grade | 13 (27.6) |
| Some high school or graduate/GED | 22 (46.8) |
| Some college or college graduate | |
| Inadequate health literacy† | 14 (29.8) |
| Income ≤ $20,000 / year, n (%) | 43 (91.5) |
| Primary recruitment condition | |
| Diabetes | 17 (36.2) |
| Rheumatoid arthritis | 15 (31.9) |
| Congestive heart failure | 15 (31.9) |
| Quality of life | |
| Excellent | 1 (2.1) |
| Very good | 6 (12.8) |
| Good | 6 (12.8) |
| Fair | 19 (40.4) |
| Poor | 15 (31.9) |
| Clinicians (n=39) | |
| Age, years (SD) | 43.7 (11.3) |
| Women, n (%) | 25 (61.5) |
| Primary care*, n (%) | 28 (71.8) |
| Specialty*, n (%) | 11 (28.2) |
| Diabetes | 5 (12.8) |
| Cardiology | 2 (5.1) |
| Rheumatology | 3 (10.3) |
| Degree, n (%) | 27 (71.1) |
| Physician | 11 (28.9) |
| Nurse practitioner or physician assistant | |
| Resident, n (%) | 8 (20.5) |
| Years since professional degrese, mean (SD) | 13.9 (10.0) |
| Spoke Spanish during encounter, n (%) | 16 (48.7) |
| Encounters (n=71) | |
| Relationship length years, n (%)‡ | |
| < 1 year | 11 (15.9) |
| 1-5 years | 37 (53.6) |
| >5 years | 21 (30.4) |
| Mean visit length, minutes (SD) | 24.6 (10.0) |
| Language concordant, n (%) | |
| English | 42 (59.2) |
| Spanish | 25 (35.2) |
| Interpreter | 4 (5.6) |
| Clinician computer use, n (%) | 19 (26.8) |
| Low (score 0-4) | 27 (38.0) |
| Moderate (score 5-7) | 25 (35.2) |
| High (score 8-12) |
Spanish-speaking patients who reported English proficiency less than “very well”
“Somewhat,” “a little bit” or “not at all” “confident filling out medical forms by yourself”
69 responses
The clinician computer use score summed 4 coder ratings (Cronbach's alpha 0.67): amount of computer data review, typing/clicking, eye contact with patients, and non-interactive pauses.2-4. With “eye contact” reversed, high total scores (range 0-12) indicated more computer use. Inter-rater reliability was 0.90 (4 videos), and we validated the score calculating its correlation (0.66) with clinician/patient statements occurring during computer use (33 encounters).
After visits, patients rated the quality of medical care received in the past 6 months (poor to excellent).
We analyzed communication using the Roter Interaction Analysis System.5 Statements were assigned one of 37 codes (average inter-rater reliability 0.74), which were summed in categories (Table 2). Rapport-building included: positive (e.g., laughter or agreement); negative (e.g., criticism or disagreement); emotional (e.g., empathy or partnership,); and social (“chit-chat”). Positive affect sums ratings for emotional tone.
Table 2. Differences in Communication Outcomes by Degree of Clinician Computer Use in Safety Net Encounters.
| Low computer use | Moderate computer use | High computer use | |||||
|---|---|---|---|---|---|---|---|
| Patient | Mean | Mean | Adj Diff | Adj p-value* | Mean | Adj Diff | Adj p-value* |
| Rapport-building | |||||||
| Positive | 43.7 | 33.1 | -18.3 | <0.01 | 36.6 | -9.5 | 0.16 |
| Negative | 1.7 | 3.1 | 1.9 | 0.10 | 1.3 | -0.4 | 0.96 |
| Emotional | 10.3 | 17.1 | 6.4 | 0.40 | 11.7 | 2.1 | 0.75 |
| Social | 5.5 | 4.4 | 3.4 | 0.28 | 10.9 | 9.6 | 0.04 |
| Biomedical information | 114.4 | 119.7 | -3.6 | 0.89 | 146.8 | 8.6 | 0.77 |
| Psychosocial information | 10.7 | 11.3 | -8.0 | 0.34 | 7.6 | -11.1 | 0.13 |
| Activation | 3.8 | 2.3 | -1.2 | 0.37 | 3.0 | -0.6 | 0.68 |
| Positive affect score | 18.2 | 19.9 | 2.4 | 0.02 | 18.0 | 0.4 | 0.55 |
| Clinician | Mean | Mean | Adjusted Difference | p-value* | Mean | Adjusted Difference | p-value* |
| Rapport-building | |||||||
| Positive | 32.9 | 26.0 | -9.7 | 0.15 | 36.6 | -8.9 | 0.15 |
| Negative | 0.2 | 0.7 | 1.7 | 0.30 | 1.3 | 2.7 | <0.01 |
| Emotional | 13.3 | 13.6 | -0.3 | 0.95 | 11.7 | 0.68 | 0.89 |
| Social | 4.2 | 4.2 | 2.7 | 0.60 | 10.9 | 9.7 | <0.01 |
| Biomedical information | 110.9 | 126.5 | -23.5 | 0.35 | 157.5 | 18.1 | 0.56 |
| Psychosocial information | 12.9 | 51.2 | 23.0 | 0.07 | 11.1 | 4.4 | 0.71 |
| Activation | 20.4 | 27.7 | 4.9 | 0.37 | 26.6 | -0.6 | 0.88 |
| Positive affect score | 24.6 | 24.7 | -1.5 | 0.15 | 21.2 | -4.1 | <0.01 |
| Encounter | Mean | Mean | Adjusted Difference | p-value* | Mean | Adjusted Difference | Adj p-value* |
| Verbal dominance | 1.24 | 1.60 | 0.18 | 0.29 | 1.65 | 0.23 | 0.12 |
| Patient-centeredness score | 0.75 | 1.14 | 0.22 | 0.31 | 0.69 | -0.1 | 0.50 |
Analyses used “low computer use” as the reference and were adjusted for patient educational attainment and quality of life, clinician years in practice, clinician type (physician, nurse practitioner, physician assistant), clinic, and visit length.
We categorized computer use scores into tertiles (Table 1). Multivariate analyses controlled for visit length and variables with bivariate associations (p<0.10) with higher computer use (lower patient education, poorer quality of life, nurse practitioners, fewer clinician practice years, and general medicine, family medicine, and diabetes clinics). We performed generalized estimating equations regression for within-clinician correlations (Stata/SE 12.1), after multilevel regression showed minimal within-patient correlation.
Results
We recorded 71 encounters among 47 patients and 39 clinicians (38% and 83% participation) (Table 1).
Compared with patients in low computer use encounters, patients in high computer use encounters were less likely to rate care as “excellent” (48% vs. 83%, p=0.04) and used more social rapport-building (+9.6, p=0.04) (“You like wearing your hair that way …”)
Clinicians in high computer use encounters (Table 2):
Used more negative rapport-building (+2.7, p<0.01). (“No, it looks like [your specialist] filled that medication for you. It has a refill.”)
Used more social rapport-building (+9.7, p<0.01). (“I'm looking at a few different jobs.”)
Demonstrated less positive affect (-4.1, p<0.01).
Discussion
High computer use by safety net clinicians was associated with lower patient satisfaction and observable communication differences. Although social rapport-building can build trust and satisfaction,6 concurrent computer use may inhibit authentic engagement, and multi-tasking clinicians may miss openings for deeper connection. Disagreement may arise when clinicians educate patients using information learned through the EHR, particularly if clarifying misunderstandings resulting from communication barriers in different clinical settings. Disagreements build rapport by signaling sufficient trust to disagree honestly, but if the overall affective tones are less positive, this could ultimately inhibit patient engagement. These factors may affect patients' overall perceptions of care.
This study used a validated coding method and a linguistically diverse population. Limitations include possible volunteer bias; recall bias with the satisfaction measure; confounding, and effects on eye contact ratings by non-computer tasks.
Software, structural, and curricular interventions7 should support clinicians' EHR use in ways that enhance their capacity to communicate with and care for diverse patients.
Acknowledgments
Funding/Support: Research reported in this publication was supported by AHRQ Grants 1K08HS022561 and K99HS022408 and the National Center for Advancing Translational Sciences of the NIH under Award Number KL2TR000143. Dr. Schillinger is supported by the Health Delivery Systems Center for Diabetes Translational Research (CDTR) funded through NIDDK grant 1P30-DK092924. Drs. Ratanawongsa and Barton were fellows supported by the Pfizer Medical Academic Partnership Fellowship in Health Literacy, under the mentorship of Drs. Schillinger and Yelin.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of AHRQ or the NIH.
Footnotes
Presentations: Preliminary data from this manuscript was presented at the International Conference on Communication in Healthcare, Montreal, Quebec, Canada, September 30, 2013
Author Contributions: Dr. Ratanawongsa had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Ratanawongsa, Yelin, Schillinger.
Acquisition, analysis, or interpretation of data: Ratanawongsa, Barton, Lyles, Wu, Martinez.
Drafting of the manuscript: Ratanawongsa, Yelin.
Critical revision of the manuscript for important intellectual content: Barton, Lyles,Wu, Yelin, Martinez, Schillinger.
Statistical analysis: Ratanawongsa, Yelin.
Study supervision: Ratanawongsa, Yelin, Schillinger.
Conflict of Interest Disclosures: No disclosures were reported.
Obtained funding: Ratanawongsa, Schillinger.
Administrative, technical, or material support: Barton, Lyles,Wu, Martinez, Schillinger.
Contributor Information
Neda Ratanawongsa, Email: neda.ratanawongsa@ucsf.edu.
Jennifer L. Barton, Email: bartoje@ohsu.edu.
Courtney R. Lyles, Email: courtney.lyles@ucsf.edu.
Michael Wu, Email: MichaelWu0322@gmail.com.
Edward H. Yelin, Email: ed.yelin@ucsf.ed.
Diana Martinez, Email: diana.martinez@ucsf.edu.
Dean Schillinger, Email: dean.schillinger@ucsf.edu.
References
- 1.Berkman ND, Sheridan SL, Donahue KE, Halpern DJ, Crotty K. Low health literacy and health outcomes: an updated systematic review. Ann Intern Med. 2011;155(2):97–107. doi: 10.7326/0003-4819-155-2-201107190-00005. [DOI] [PubMed] [Google Scholar]
- 2.Frankel R, Altschuler A, George S, et al. Effects of exam-room computing on clinician-patient communication: a longitudinal qualitative study. J Gen Intern Med. 2005;20(8):677–682. doi: 10.1111/j.1525-1497.2005.0163.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rouf E, Whittle J, Lu N, Schwartz MD. Computers in the exam room: differences in physician-patient interaction may be due to physician experience. J Gen Intern Med. 2007;22(1):43–48. doi: 10.1007/s11606-007-0112-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Margalit RS, Roter D, Dunevant MA, Larson S, Reis S. Electronic medical record use and physician-patient communication: an observational study of Israeli primary care encounters. Patient Educ Couns. 2006;61(1):134–141. doi: 10.1016/j.pec.2005.03.004. [DOI] [PubMed] [Google Scholar]
- 5.Roter D, Larson S. The Roter interaction analysis system (RIAS): utility and flexibility for analysis of medical interactions. Patient Educ Couns. 2002;46(4):243–251. doi: 10.1016/s0738-3991(02)00012-5. [DOI] [PubMed] [Google Scholar]
- 6.Mauksch LB, Dugdale DC, Dodson S, Epstein R. Relationship, communication, and efficiency in the medical encounter: creating a clinical model from a literature review. Arch Intern Med. 2008;168(13):1387–1395. doi: 10.1001/archinte.168.13.1387. [DOI] [PubMed] [Google Scholar]
- 7.Duke P, Frankel RM, Reis S. How to integrate the electronic health record and patient-centered communication into the medical visit: a skills-based approach. Teach Learn Med. 2013;25(4):358–365. doi: 10.1080/10401334.2013.827981. [DOI] [PubMed] [Google Scholar]