0 Boeske, N.T. (Niels)
AMC UVA
May 2019 – November 2019
The impact of eye contact on the
patient-physician relationship
1
The impact of eye contact on the
patient-physician relationship
Student N.T. Boeske Student Number: 11931205 Email: n.t.boeske@amsterdamumc.nl Mentor C. JongeriusDepartment of Medical Psychology Academic Medical Center
Email: c.jongerius@amsterdamumc.nl
Tutor
Dr. E. Joukes
Department of Medical Informatics Academic Medical Center
University of Amsterdam
Email: e.joukes@amsterdamumc.nl
Location of Scientific Research Project
Medical Psychology department Academic Medical Center Meibergdreef 9
1105 AZ Amsterdam The Netherlands
2
Contents
Abstract ...3
1 Introduction ...5
2 Method ...8
2.1 Sample Selection and recruitment ...8
2.2 Procedure ...8
2.3 Instruments ...8
2.3.1 Socio-demographic questionnaire ...9
2.3.2 Tobii eye tracking glasses ...9
2.3.3 Wake Forest Physician Trust Scale questionnaire ...9
2.4 Data preparation ...9 2.5 Statistical Analysis ... 10 3 Results ... 10 3.1 Sample Description ... 10 3.2 Descriptive analyses ... 11 3.3 Assumptions ... 11
3.4 Correlation between physician gaze and patient trust ... 12
3.5 Prediction of average gaze time in first minute for total period ... 12
4 Discussion ... 12 4.1 Principal findings ... 12 4.2 Limitations ... 13 4.3 Future research ... 14 4.4 Strengths ... 14 4.5 Implications ... 15 5 Conclusion ... 15 6 References ... 16
A. Appendix A – Wake Forest Physician Trust Scale ... 18
3
Abstract
Background: Communication is vital for a good patient-physician relationship. Non-verbal
communication such as eye contact between physician and patient is presumed to benefit patients’ trust. The introduction of Electronic Health Records (EHRs) can be expected to influence the amount of eye contact between physicians and patients. Consequently, it could have an effect on the level of the patient’s trust in the physician. Previous studies, because of their technological limitations, lacked precision in measuring this effect. Our study, making use of the latest technology, is the first to examine exactly how much time the physician looks at the patient and in how far that influences the patients’ trust.
Objectives: The aim of this study was to examine whether the amount of time the physician looks at the patient correlates with the patients’ trust. A secondary aim was to see whether the physicians’ gaze time during the first minute of the consultation predicted the physicians’ total gaze time during the consultation.
Study Design: An observational, cross-sectional study design using eye tracking observations and questionnaire data of physician-patient consultations.
Methods: Consultations between 90 patients and their 16 physicians in a Dutch internal medicine outpatient clinic were recorded. The eye gaze by the physician was recorded with eye tracking glasses and the eye tracking output was registered with the use of facial recognition software. Two questionnaires were completed by patients and physicians, the Wake Forest Physician Trust Scale score (WFPTS score) questionnaire, to assess the patients’ trust and a questionnaire on socio-demographic information.
Results: No significant correlation was found between the physicians’ amount of gaze during the consultation and patients’ trust in the physician (r = -.032, p = .765). A significant correlation was found between physicians’ average gaze time during the first minute of the consultation and their total amount of gaze time during the consultation (r = .614, p= .01).
Conclusion: Physicians’ eye gaze towards the patient during a consultation, according to this study, is not a significantly important factor in gaining trust from the patient. Efforts, if any, to influence physicians’ behavior during consultations in the sense that they increase their amount of gaze towards patients, possibly at the expense of their attention for the EHR, therefore need to be
reconsidered. Using gaze tracking during the first minute of a consultation might be enough to gather data for future similar studies. Accurately predicting the amount of gaze during the whole
consultation by extrapolating the information gathered during the first minute of a consultation, as opposed to measuring the amount of gaze during the full consultation, will provide researchers with a possibility to save time collecting data.
Keywords: Patient, Physician, gaze, trust, eye-tracking Nederlandse samenvatting
Achtergrond: Communicatie is van groot belang voor een goede relatie tussen patiënt en arts. Verondersteld wordt dat non-verbale communicatie, zoals oogcontact tussen arts en patiënt, het vertrouwen van de patiënt gunstig beïnvloedt. De ingebruikneming van Elektronische
Patientendossiers (EPD’s) heeft, naar verwachting, invloed op de hoeveelheid oogcontact tussen artsen en patiënten. Dientengevolge zou zij een effect kunnen hebben voor de mate van het vertrouwen dat de patiënt in de arts heeft. Technologische beperkingen van voorgaande
onderzoeken misten precisie bij het meten van dit oogcontact. Ons onderzoek, waarvoor gebruik is gemaakt van de nieuwste technologie: eye-tracking, is het eerste waarin is onderzocht hoe lang de arts precies naar de patiënt kijkt en in hoeverre dit het vertrouwen van patiënten beïnvloedt.
4 Doelstellingen: Dit onderzoek was erop gericht te onderzoeken of de hoeveelheid tijd gedurende welke de arts naar de patiënt kijkt, een correlatie had met het vertrouwen van de patiënt. Een tweede doel was om vast te stellen of de aankijktijd gedurende de eerste minuut van een consult, voorspellende waarde had ten aanzien van de aankijktijd gedurende het gehele consult.
Design: Een observationeel onderzoek, waarbij we gebruik maakten van eye-tracking om de
kijkrichting van de arts vast te leggen. Ook gebruikten we informatie uit vragenlijsten die betrekking hadden op de consulten van artsen en patiënten.
Methoden: Negentig consulten tussen arts en patiënt werden opgenomen in een Nederlandse polikliniek inwendige geneeskunde. Oogcontact tussen arts en patiënt werd vastgelegd door middel van een bril die de oogbeweging registreerden middels eye-tracking. De informatie over
oogbewegingen werd vervolgens verwerkt middels gezichtsherkenningssoftware. Twee vragenlijsten werden ingevuld door patiënten, Het wake Forest Physician Trust Scale (WFPTS), ter vaststelling van het vertrouwen van de patiënt, en een vragenlijst met betrekking tot sociaal-demografische
informatie.
Resultaten: We vonden geen significante correlatie tussen de mate van oogcontact gedurende het consult en het vertrouwen van de patiënt in de arts na het consult (r = -.032, p = .765). Wel werd een significante correlatie gevonden tussen de mate van oogcontact van de arts gedurende de eerste minuut van het consult en hun totale mate van oogcontact gedurende het gehele consult (r = .614, p= .01).
Conclusie: het aankijken van een patiënt door artsen gedurende een consult lijkt geen significant relevante factor voor het verkrijgen van het vertrouwen van de patiënt. Als al inspanningen worden gedaan om het gedrag van artsen tijdens consulten te beïnvloeden, in die zin dat zij de tijd van het aankijken van patiënten vermeerderen, zulks wellicht ten koste van hun aandacht voor het EPD, moeten deze inspanningen worden heroverwogen. De registratie van
oogbewegingen gedurende de eerste minuut van het consult volstaat wellicht bij het verzamelen van gegevens voor toekomstig vergelijkbaar onderzoek. Het accuraat voorspellen van de
aankijktijd gedurende het gehele consult door extrapolatie van de tijdens de eerste minuut van het consult verzamelde informatie, zal, zulks in tegenstelling tot het meten van de aankijktijd gedurende het gehele consult, voor onderzoekers in een tijdsbesparing bij het verzamelen van gegevens resulteren.
5
1 Introduction
Patient-physician communication is widely considered to be a vital part of a consultation (Duke et al., 2013). It has been established to affect outcomes such as patient trust, satisfaction, adherence to treatment advice and other clinical outcomes, and to maximize the success rate of efforts to resolve conflicts between physicians and patients, should they arise (Orom et al., 2018, Stewart, 1995, Henry et al., 2012, Hillen et al., 2015, Stewart et al., 2000, Halpern, 2007). Suboptimal communication between physicians and patients has been associated with increased complaints by the patient (Moore et al., 2011). Studies even show a relationship between physicians’ communication behavior and malpractice claims (Levinson et al., 1997, Stewart et al., 1999).
Communication contains both verbal and non-verbal components. Verbal communication contains linguistics as opposed to non-verbal communication, which does not contain linguistics (Knapp et al., 2013). More studies have been conducted on verbal communication between patient and physician than on non-verbal communication (Mast, 2007). Non-verbal communication can be speech related: it includes, for example, changing the tone of voice. Other examples of non-verbal communication are non-speech-related, e.g. nodding, body posture and eye contact or gaze (Knapp et al., 2013). Eye contact being a highly noticeable and possibly influential form of non-verbal behavior, was the basis for focusing in this study on gaze instead of on other non-verbal behavior. Gaze is defined as the time the physician looks directly at the patient’s face (Bensing et al., 1995). Several studies on communication between patient and physician addressed gaze. One study, for example, examined 86 audio- and videotaped patient-physician consultations, and showed that gaze was highly
consequential of behavior during certain activities, such as history taking, physical examination, diagnosis, treatment and closing. It showed that the tasks performed during the consultation by the physician, such as the start of the consultation, or taking history, influenced the amount of gaze. (Robinson, 1998). Furthermore, a review researching non-verbal aspects in patient-physician interactions, showed that if the physician made less eye contact, patients were less satisfied. (Shachak and Reis, 2009).
This present study focuses on the effect of gaze of the physician on the reported trust of a patient in his or her physician. Patients’ trust in their physician is an important part of the patient-physician relationship. A high level of patients’ trust in their physician corresponds with increased patients’ satisfaction, treatment adherence, improved health status, along with a diminished likelihood of leaving the physician’s practice or withdrawing from a health plan (Pearson and Raeke, 2000). A study in adult patients (n=414) showed that trust of a patient in his or her physician is correlated with the physician’s satisfaction, continuity of care, adherence to medication and general health
outcomes (Thom et al., 1999). Trust in general is also referred to with terms like ‘confidence’ or ‘faith’ (Hupcey et al., 2001). Within a medical situation social and interpersonal trust can be distinguished (Pearson and Raeke, 2000), social trust meaning trust in the health care system as a whole, whereas interpersonal trust refers to the trust of a patient in his or her health care worker. Interpersonal trust has been defined as “the optimistic acceptance of a vulnerable situation in which the patient believes the physician will care for the patient’s interest” (Hall et al., 2001). Trust can be broken down in the following dimensions:
(1) Fidelity: pursuing a patient’s best interests and not taking advantage of his/her vulnerability (2) Competency: avoiding mistakes and producing the best achievable results
(3) Honesty: telling the truth and avoid intentional falsehoods
6 (5) Global trust: anything related to trust but not exclusively to one of the above-mentioned
dimensions or with a significant component that is not subject to dissection (Hall et al., 2001).
Ways of measuring a patient’s trust in his or her physician were published in 1990 in the “Trust in Physician Scale” (Anderson and Dedrick, 1990). In order to improve the above-mentioned Trust in Physician scale, the WFPTS was developed, based on all of the before-mentioned dimensions with the exception of the confidentiality dimension (Hall et al., 2002). A Dutch version of the WFPTS was developed, and its psychometric properties were tested and found a sound instrument to assess patients’ trust (Bachinger et al., 2009).
With the increased use of EHRs in the consulting room, physician-patient eye contact may be affected. In the past two decades, studies were performed to examine the impact of such EHRs on clinical consultations. The positive effects of EHR use were shown in a systematic review of 257 papers. This review found EHR systems to improve quality of care by increasing care providers’ adherence to guidelines, enhancing disease surveillance and decreasing medication errors (Chaudhry et al., 2006). While the use of electronic EHRs thus appears to have a positive effect on the quality of care, there are also concerns: some fear that their use may impede the level of connectedness between physicians and patients, especially on a non-verbal level. For example, a study examining gaze patterns between patients and physicians while the physician used the EHR, found that EHR use reduced patient-physician eye contact, possibly resulting in a diminished trust by the patient in the physician (Montague and Asan, 2014). Furthermore, another study addressing primary complaints by patients with regard to the orientation of the physician towards the patient or to the EHR, suggests that, as physicians disengage from interaction with the patient and shift their attention to the medical record, this could leave the patient in doubt as to whether the physician is listening to what the patient is saying (Ruusuvuori, 2001). These findings show that when physicians spend less time looking at the patient, this may have implications for communication. If non-verbal communication is affected, this may be detrimental for patients’ trust.
As the use of EHRs is likely to diminish the amount of gaze, patients may perceive that physicians using electronic EHRs are paying less attention to them, which might have an effect on the patients’ trust (Asan et al., 2014). Patients’ trust in their physician is an important part of the patient-physician relationship. A high level of patients’ trust in their physician corresponds with increased patients’ satisfaction, treatment adherence, improved health status, along with a diminished likelihood of leaving the physician’s practice or withdrawing from a health plan (Pearson and Raeke, 2000). Eye contact could be an important facilitator in trust; a study with breast cancer patient showed the importance of maintaining eye contact and patients trust (Hillen, 2015).
The aforementioned studies were all observational using video recordings of consultations that were manually coded for gaze patterns. When using these observational techniques, the limitations in the human capacity to objectively assess complex interactions, make it difficult to actually observe physicians’ focus of gaze. As of late so-called electronic eye tracking devices have been developed, which can measure the gaze patterns accurately. Eye tracking as a research tool has become more widely used in a range of disciplines, such as entertainment, marketing, advertising, sociology and neuroscience (Holmqvist et al., 2011). Eye tracking has been used for a variety of purposes, e.g. measuring what a subject is focusing on in a store or on a computer screen, or during other information acquiring or processing tasks, such as pilots interacting with a flight simulator (Duchowski, 2002). We did not find any literature on patient-physician communication using eye tracking technology, whereas the use of such technology could very well enhance the accuracy of eye gaze measurement, compared to manual coding of gaze patterns.
7 The first aim of the present study is to examine the effect of gaze of the physician, in situations where electronic EHRs are used, on the reported trust of a patient in his or her physician. The hypothesis was that the use of EHRs could diminish the amount of the physician’s gaze towards the patient, which subsequently could result in a diminished trust of the patient in the physician. We made use of eye tracking technology, which previously had not or scarcely been available, and more accurately measures the amount of gaze. Besides correlating between average gaze and trust we also explored some known correlations to patient trust such as age, gender and educational level.
Our secondary research goal was to establish whether the physician’s amount of gaze towards the patient during the first minute of a consultation can serve as an indicator of the total amount of gaze time during the whole consultation. This builds on the thin slice theory, a term used in psychology to describe the ability to find patterns in narrow windows of observations. This means that it is possible to make very quick inferences about characteristics of an individual or situation with small amounts of information (Carney et al., 2007). The type of research conducted within this present study is often very time consuming. If eye contact during the first minute of a consultation offers an adequate approximation of the amount of gaze during the whole consultation, future researchers could under certain circumstances limit themselves to using the results during that first minute, which would save them time in processing the collected data.
8
2 Method
2.1 Sample Selection and recruitment
Patients and residents of the department of internal medicine at a Dutch academic hospital were invited to take part in the study. Eligible patients had to be over 18 years of age, fluent in the Dutch language and had to agree voluntarily to participate in the study. Patients were selected by the researchers through Epic, the EHR system used in the academic hospital (Epic, 2019). Only patients who were meeting with this particular resident for the first time, were eligible. Patients who suffered from a major psychological or psychiatric disorder, as judged by their physician based on their EHR, were excluded. Informed consent was signed by all participants prior to data collection. The study was judged to be exempt from the Medical Research Involving Human Subjects Act by the Medical Ethics Committee of the Amsterdam University Medical Center.
2.2 Procedure
This study is part of a larger study, investigating the impact of EHR use on communication between patient and physician. See Figure 1. Procedure of the observational study. The process started with the selection of the subjects. Questionnaires for both patient and physician were completed at four points in time (T0 to T3). Questionnaires were provided online, through a web-based solution called Qualtrics (Qualtrics, 2019). Paper questionnaires were available as backup. Furthermore,
measurements, consisting of video recording and eye tracking, were performed of a consultation of the patient with the physician.
At T0, immediately after the selection of the subjects, a link to an online questionnaire was sent to physicians with questions about their socio-demographic characteristics and attitude towards the use of EHRs. At T1, approximately 2 weeks prior to the consultation, a link to an online questionnaire was sent to the participating patient, containing questions on socio-demographic characteristics, attachment style and social anxiety. Directly before the consultation at T2, the patient completed a questionnaire assessing their state anxiety (Bekker et al., 2003, van der Bij et al., 2003). After the consultation at T3 the patient again reported state anxiety, satisfaction, trust and perceived empathy. Immediately after the consultation at T3 residents completed questions about their satisfaction about their own communication and the consultation as a whole.
2.3 Instruments
For this study we analyzed a selection of the abovementioned instruments. We used the following measurements and questionnaires:
Physician Patient
T0
Consult: eye tracking and video recording +/- 2 weeks prior to consult Prior to consult After consult Start T1 T2 T3 End
9
2.3.1 Socio-demographic questionnaire
Socio-demographics assessed for the resident included sex, birth date, nationality, phase of professional education and years of experience. Socio-demographic characteristics assessed for patients were sex, age, nationality, relationship status, employment status and level of education.
2.3.2 Tobii eye tracking glasses
During the consultation Tobii Pro 2 eye tracking glasses were used. Tobii Pro 2 eye tracking glasses show exactly what a person is looking at in real time while the wearer can move freely. The product consists of a head unit, a recording unit and controller software. The glasses consist of the following parts: a high definition scene camera that captures a full HD video of what is in front of the wearer, a microphone and eye tracking sensors that record eye orientation, i.e. the direction of the eye gaze. The eye tracking glasses are connected to a recording unit. This recording unit uses a removable SD memory card to record the eye tracking data, sounds and scene camera video. The recording unit also carries a replaceable and rechargeable LI-ion battery that supplies power to both the recording unit and the glasses (Tobii, 2019a)
2.3.3 Wake Forest Physician Trust Scale questionnaire
The Wake Forest Physician Trust Scale asks patients to indicate their trust in their physician by answering 10 questions on a 5 point Likert scale with response categories ‘Strongly Disagree’ = 1, ‘Disagree’ = 2, ‘Neutral’ = 3, ‘Agree’ = 4, ‘Strongly Agree’ = 5. One example of a question is: ‘Your doctor will do whatever it takes to get you all the care you need’. See Appendix A for the full Wake Forest Physician Trust Scale questionnaire. Final scores, by using the mean of the answers, can range from 1 to 5 with a higher score indicating more trust. This study used the Dutch version of the Wake Forest Physician Trust Scale, which was previously tested for validity and reliability (Bachinger et al., 2009). Internal consistency on the Wake Forest Physician Trust Score scale was tested using
Cronbach’s alpha. Internal consistency in our sample was adequate (Cronbach’s α= .651).
2.4 Data preparation
This study used automated gaze measurements by means of the Tobii eye tracking glasses and two software programs, a facial recognition program, based on Openpose (Callemein et al., 2018), and Head Analyzer, both of which were developed specifically for this study. Several preparatory steps were needed before the data could be analyzed. See Figure 2 for the flow of the data preparation. First, the eye tracking glasses recordings were viewed in the Tobii pro lab software (Tobii, 2019b). Using the Tobii pro lab software, a video from the physician's perspective with a small circle indicating the gaze of the physician was created; no additional filtering was performed on the gaze data with the use of the Tobii pro lab software. The raw data were used. Second, the video with the small gaze circle was analyzed using facial recognition software based on the aforementioned Openpose to identify faces on the video. This resulted in a file with exact coordinates of a rectangle bordering the patients’ face, or the face of any other person accompanying the patient, per frame of the video. Additionally, using the Tobii pro lab software, a second file was exported containing per frame the coordinates in pixels of the gaze point, using an x-axis horizontally and a y-axis vertically of the frontally recorded video. These x and y coordinate files were then processed in Rstudio to create a correctly formatted file for the Head Analyzer software program (RStudio, 2019). The resulting three files (video, rectangle coordinates of a person’s head in the video and x and y coordinates of the focal point of the physician’s gaze) were subsequently combined using Head Analyzer software to code whether the gaze of the physician fell within the pre-specified frame of the patient’s or
10 physician’s gaze was on the patient/caregiver’s face (1) or not (0). The videos were examined and times were recorded when the examination room was opened, the moment at which the patient sat down, when the patient stood up and when the patient left the examination room. Also, the number of people accompanying the patient, if any, were noted. If the physician did a physical examination, the start and end times of such an examination were noted as well. The physician was instructed to remove the eye tracking glasses during the examination. These periods were excluded from the examination.
2.5 Statistical Analysis
The statistical analyses were carried out using SPSS Statistics, version 25. There were no missing values. The average gaze time, for the gaze-trust analyses and for the first minute correlation, was calculated from the first time the patient sat down until the moment when the patient stood up to leave the room, excluding the physical examination time if applicable. If a physical examination took place, that period of time was removed. Necessary assumptions were checked i.e. outliers (using boxplot), normality (using Shapiro Wilk test), and skew (using descriptive analysis). These assumption were checked on all variables. Correlations were calculated between trust (WFPTS score) and
average gaze, the patient’s age, gender and education level.Because parametric assumptions for the WFPTS score, particularly the normal distribution, were not met, Spearman’s rho correlation was used with a two-tailed significance level of p< 0.05. For the correlation between the first minute of gaze average and the total amount of gaze average we used Pearson, since both variables had a normal distribution.
3 Results
3.1 Sample Description
The collection of data started in February 2018 and was finalized in May 2019. In total 208 patients were asked to participate in the study. The response rate was 56%. 93 patients refused to
participate, citing the following reasons: time constraints, no interest or privacy issues.
Consequently, 115 patients participated in the analysis of the study. Three patients cancelled their appointments with the physician, one did not speak sufficient Dutch. Data from the Tobii eye tracking glasses were missing on four patients and one had only three minutes of recording. During data preparation we encountered 16 cases with synchronization issues of the various data files. The patient data related to these instances were omitted from the analysis, the result of which was that 90 patients were included in our analyses. Of the patients used, there were no missing values from the questionnaires. Table 1 shows socio-demographic characteristics of the sample. The sample
Tobii Project file X and Y coordinates
Gaze file
Head Analyzer Facial Recognition Video with gaze RStudio Output file Head 1/011 shows a diverse group of patients. The patients were consulted by 16 different resident physicians. Fifty percent of consultations were performed by female physicians and fifty percent by male physicians.
Table 1 Socio-demographic and economic variables
Variable label Value labels Frequency %
Patient Age 18-30 years 3 3.3 31-60 years 45 50.0 61-99 years 42 46.7 Gender Male 46 51.1 Female 44 48.9 Nationality Dutch 84 93.3 Other 6 6.7 Partner Yes 64 71.1 No 26 28.9 Children Yes 66 73.3 No 24 26.7
Education Lower education 20 22.2
Average education 44 48.9 Higher education 26 28.9 Employment Yes 32 35.6 No 58 64.4 Physician
Mean Age (SD) 34 (2.3) Range 29-38
Gender Male 8 50
Female 8 50
3.2 Descriptive analyses
Patients’ average trust in the resident physicians (WFTS) was strong, with an average score of M = 4.37, SD = .64, and a range of 2.1-5.0. Total average gaze time was 44%, SD = 16% and a range of 10% – 78%. Average gaze time in the first minute was 61%, SD = 20% and a range of 9% – 96%
3.3 Assumptions
The assumptions were checked for Pearson correlation. Figure 3 in Appendix B represents the distribution of the WFPTS score variable. According to SPSS box plot (1.5 times the interquartile range) the WFPTS had three outliers: patient numbers 94, 119 and 140. The outliers were all reporting lower trust. After examining the videos no immediate reason could be found. The WFPTS showed a right-skewedness of (-1.365 (SE=.254)) and a kurtosis of (2.043 (SE=-.503)). WFPTS was, according to a Shapiro-Wilk test, not normally distributed (p<.001). Removing or substituting the outliers with the mean did not change the non-normality or skewedness. Outliers were kept as reported. Average Gaze variable showed no outliers and had a normal distribution.
12
3.4 Correlation between physician gaze and patient trust
Table 2 represents the Spearman’s rho correlation coefficients for correlation between trust (WFPTS score), average gaze, gender, age and education level.
Table 2 Spearman's rho correlation coefficients for correlation between WFTS trust, Average Gaze, Age, Gender and Education level
N=90
Wake Forest Trust Scale score Average Gaze Patient Gender Patient Age Patient Education level
Wake Forest Trust Scale score 1 -.032 .107 .274* .125
Average Gaze -.032 1 .006 -.108 -.067
Gender .107 .006 1 .046 -.030
Age .274* -.108 .046 1 -.143
Education level .125 -.067 -.030 -.143 1
*. Correlation is significant at the 0.01 level (2-tailed).
No significant correlation was found between the amount of eye contact (average gaze) and the trust level (WFPTS score) about the physician, reported by the patient: r = -.032, p= .765. There was a significant positive relationship between age of the patient and the trust level reported, r = .274, p = .009, indicating that the older the patient the stronger patients’ trust.
3.5 Prediction of average gaze time in first minute for total period
The correlation between the eye contact (average gaze) during the first minute after the patient sat down, and the total average gaze was found positive and significant, r = .614, p = .01.
4 Discussion
4.1 Principal findings
The presence of computers for EHR use and other diagnostic equipment has created an increase of activities by physicians not directly related to their communication with patients, with the effect that the patient-physician communications are diminished (Street Jr et al., 2014). We tested whether the amount of time the physician looks at the patient has an effect on patients’ trust in the physician. This study was the first to examine physician eye contact, the so-called gaze, towards patients using mobile eye tracking technology to accurately measure the gaze time during a consultation. Whereas previously correlations between doctors’ gaze and patients’ trust, for lack of the new technology used in this study, had only been researched with the use of manually coded data sets, we were now able to more accurately establish the amount of gaze of a physician toward a patient during a consultation. Because of the reduced reliance on human judgement on measurements, these new data were probably more accurate and reproducible. They were combined with information about the trust reported by patients, which was expected to create a more accurate view of the influence
13 of the amount of gaze by physicians on the patients’ perception of the physicians’ actions than had previously been possible.
The results showed no significant correlation between the amount of time a physician looks at the patient during a consultation and the trust reported by that patient. As mentioned before, previous research on gaze and trust in a patient-physician setting is scarce. One study testing how oncologists’ amount of eye contact influenced breast cancer patients’ trust, showed an increase in trust when the oncologist maintained more eye contact (Hillen et al., 2015). However, the methodology used was different. In that study re-enacted videos of a consultation between an oncologist and patient were used. These videos were viewed by participants in the study, who then reported their level of trust in the oncologist. They therefore actually measured what the observer of the video expected the patient in the video to report on trust. This present observational study, in contrast, uses the reported trust of the patients themselves, based on their experience in a real life setting.
Not finding a strong correlation between gaze and trust, appears to negate the notion that there be a strong necessity to attain as much gaze as possible during a consultation. On the contrary, it gives basis to the assumption that the attention needed for notes taking or EHR attention, does not diminish the value of the consultation, as perceived by the patient. It may be, however, difficult to measure the effects of gaze on trust in a patient-physician setting as people are known to report high trust in their doctor due to their strong dependence on the doctor. Another study found that patient characteristics, such as gender and educational level, have an effect on trust, be it negative or positive (Bonds et al., 2004)
We also found that the average gaze time during the first minute predicted the total average gaze time during a consultation, indicating that physicians’ amount of gazing at the patient during the first minute is highly predictive of their gaze behavior during the entire consultation. Physicians appear to be reasonably unchanging in their gaze pattern towards their patients. This might lead to the
conclusion that only a short measurement is needed to get a representative indication of the amount of gaze by the physician. This is in line with the thin slice theory (Carney et al., 2007). In view of the fact that accurate and representative data may be retrieved from only the first minute of the consultation, without a significant loss of the usability of such data, future research might, on this basis, benefit from easier data gathering and simpler data analysis, thus saving time and money. As the results of this study with regards to a correlation between doctor’s gaze and patient’s trust is at odds with the results of previous studies, which could be due to the advanced methods used in this present study, follow up research may be warranted. In view of our finding with regards to the first minute, the use of thin slicing in follow up research could be considered. Since, however, there is a possibility that other patterns of gaze might be found in analyzing the whole consultation, the use of thin slicing in follow up research should still be considered carefully.
4.2 Limitations
The study had several limitations. First, the facial recognition software that was used, recognized all faces in the video, both patients’ and their caregivers’. This might influence the gaze time as the physician needs to switch attention between patient and caregiver(s). However, in other studies patients were also accompanied by one or more caregivers and that did not seem to influence the results (Street Jr et al., 2014).
The population of analysis of the present study is a mixed group of patients of the department of internal medicine of one hospital. Patients who suffered from major psychological or psychiatric disorder were excluded, but the severity of illness patients were treated for during this study, was
14 not part of the data used in this study, which may have had an effect on the results of this study, as varying illnesses may play a role in communication between physicians and patients.
This study used self-reports, and although measures with demonstrated reliability and validity were used, bias inherent in self-reports (e.g. social desirability and recall) may have influenced the results. Since this, however, is a widely used and generally accepted method, this can be deemed a minor disadvantage. Trust score was high and generally older people tend to report higher trust. This might be due to social differences with attitude towards the physician but might also be due to having more contact with doctors during later stages in life.
The physicians were all residents, therefore approximately around the same age (mean = 34, min 29, max 38). As it is plausible that patients’ attitude towards older or younger physicians may have an impact on patient trust, our results cannot be extrapolated beyond this population of physicians.
4.3 Future research
For future research the facial recognition software can be adjusted so that only the patient is
recognized. The same software might also be used to recognize other features such as the physician’s computer screen, enabling researchers to measure time spent by physicians looking at the EHR. During the consultations, sound was also recorded. It is, therefore, possible to transcribe what is being said and natural language processing can be used to analyze gaze in response to what is being said. Future studies might look at a more diverse or more extensive group of patients and physicians. Future research could also include physicians from different age groups.
More research can be done on Gaze shift, i.e. the measurement of gaze at other subjects of gaze than the patient, and patterns in shifting gaze from one subject to another. It could very well be relevant what physicians look at before and after looking at the patient. Gaze should be looked at not just in terms of amount of time but in relation to the courses of actions and systems of activity (Robinson, 1998). This study suggests that the duration of behaviors such as eye contact should be measured relative to actions taken by the physician.
Information on the severity of the illness patients were treated for during this study, was not part of the data used in this study. This may have had an effect on the results of this study, as varying illnesses may play a role in communication between physicians and patients. A recommendation for future research could be to limit the study to a group of patients with one specific illness, thus excluding a possible influence of the diversity of illnesses on the test results.
4.4 Strengths
This is the first study that uses mobile eye tracking and facial recognition to assess eye gaze in a consultation between a physician and patient. It therefore gives a more accurate measurement of the physician’s gaze towards the patient during the consultation, than previous studies. All
consultations in this study between physician and patient concerned patients who had had no prior contact with this particular physician. No bias can therefore come from familiarity of the patient with their physician involved.
15
4.5 Implications
As no significant correlation was found on gaze and reported trust, this might suggest that patients understand the need for their physician to look at the EHR, or that other verbal or non-verbal aspects, such as continuity in the conversation, are more important for patients’ trust in their physician. This notion, as it is, might be of interest to any and all caregivers.
In this view, the fact that our secondary research goal, i.e. whether the amount of the physicians’ gaze towards their patients during first minute of a consultation is representative of the amount of gaze during the whole consultation, did present a significant correlation, is of importance. This outcome may enable future researchers to more efficiently and effectively study the correlation between the doctor’s gaze and the trust reported by patients.
5 Conclusion
This study originated from the notion that there might be a correlation between patients’ trust in their physician and the amount of gaze by the physician toward their patients during consultations in which an EHR was used. This was based on the hypothesis that the use of EHRs could diminish the amount of the physician’s gaze towards the patient, which subsequently could result in a diminished trust of the patient in the physician. The study results show that non-verbal communication through eye contact by the physician does not influence the trust reported by the patient. Possibly, other factors are more influential for patients’ trust than the overall amount of eye contact made by the physician. Finally, the strong prediction of the first minute of gaze time suggests that physicians employ distinct gaze patterns. It may very well be possible to base eye gaze measurements on shorter segments of clinical consultations. This would simplify eye gaze analysis in future research.
16
6 References
ANDERSON, L. A. & DEDRICK, R. F. 1990. Development of the Trust in Physician scale: a measure to assess interpersonal trust in patient-physician relationships. Psychological reports, 67, 1091-1100.
ASAN, O., D. SMITH, P. & MONTAGUE, E. 2014. More screen time, less face time–implications for EHR design. Journal of evaluation in clinical practice, 20, 896-901.
BACHINGER, S. M., KOLK, A. M. & SMETS, E. M. 2009. Patients’ trust in their physician—Psychometric properties of the Dutch version of the “Wake Forest Physician Trust Scale”. Patient education and counseling, 76, 126-131.
BEKKER, H. L., LEGARE, F., STACEY, D., O’CONNOR, A. & LEMYRE, L. 2003. Is anxiety a suitable measure of decision aid effectiveness: a systematic review? Patient education and counseling, 50, 255-262.
BENSING, J. M., KERSSENS, J. J. & VAN DER PASCH, M. 1995. Patient-directed gaze as a tool for discovering and handling psychosocial problems in general practice. Journal of Nonverbal Behavior, 19, 223-242.
BONDS, D. E., FOLEY, K. L., DUGAN, E., HALL, M. A. & EXTROM, P. 2004. An exploration of patients' trust in physicians in training. Journal of Health Care for the Poor and Underserved, 15, 294-306.
CALLEMEIN, T., VAN BEECK, K., BRÔNE, G. & GOEDEMÉ, T. Automated analysis of eye-tracker-based human-human interaction studies. International Conference on Information Science and Applications, 2018. Springer, 499-509.
CARNEY, D. R., COLVIN, C. R. & HALL, J. A. 2007. A thin slice perspective on the accuracy of first impressions. Journal of Research in Personality, 41, 1054-1072.
CHAUDHRY, B., WANG, J., WU, S., MAGLIONE, M., MOJICA, W., ROTH, E., MORTON, S. C. & SHEKELLE, P. G. 2006. Systematic Review: Impact of Health Information Technology on Quality,
Efficiency, and Costs of Medical Care. Annals of Internal Medicine, 144, 742-752. DUCHOWSKI, A. T. 2002. A breadth-first survey of eye-tracking applications. Behavior Research
Methods, Instruments, & Computers, 34, 455-470.
DUKE, P., FRANKEL, R. M. & REIS, S. 2013. How to integrate the electronic health record and patient-centered communication into the medical visit: a skills-based approach. Teaching and learning in medicine, 25, 358-365.
EPIC. 2019. Epic [Online]. Available: www.epic.com [Accessed].
HALL, M. A., DUGAN, E., ZHENG, B. & MISHRA, A. K. 2001. Trust in Physicians and Medical
Institutions: What Is It, Can It Be Measured, and Does It Matter? The Milbank Quarterly, 79, 613-639.
HALL, M. A., ZHENG, B., DUGAN, E., CAMACHO, F., KIDD, K. E., MISHRA, A. & BALKRISHNAN, R. 2002. Measuring patients’ trust in their primary care providers. Medical care research and review, 59, 293-318.
HALPERN, J. 2007. Empathy and patient–physician conflicts. Journal of general internal medicine, 22, 696-700.
HENRY, S. G., FUHREL-FORBIS, A., ROGERS, M. A. & EGGLY, S. 2012. Association between nonverbal communication during clinical interactions and outcomes: a systematic review and meta-analysis. Patient education and counseling, 86, 297-315.
HILLEN, M. A., DE HAES, H. C., VAN TIENHOVEN, G., BIJKER, N., VAN LAARHOVEN, H. W., VERMEULEN, D. M. & SMETS, E. M. 2015. All eyes on the patient: the influence of oncologists’ nonverbal communication on breast cancer patients’ trust. Breast cancer research and treatment, 153, 161-171.
HOLMQVIST, K., NYSTRÖM, M., ANDERSSON, R., DEWHURST, R., JARODZKA, H. & VAN DE WEIJER, J. 2011. Eye tracking: A comprehensive guide to methods and measures, OUP Oxford.
17 HUPCEY, J. E., PENROD, J., MORSE, J. M. & MITCHAM, C. 2001. An exploration and advancement of
the concept of trust. Journal of advanced nursing, 36, 282-293.
KNAPP, M. L., HALL, J. A. & HORGAN, T. G. 2013. Nonverbal communication in human interaction, Cengage Learning.
LEVINSON, W., ROTER, D. L., MULLOOLY, J. P., DULL, V. T. & FRANKEL, R. M. 1997. Physician-Patient Communication: The Relationship With Malpractice Claims Among Primary Care Physicians and Surgeons. JAMA, 277, 553-559.
MAST, M. S. 2007. On the importance of nonverbal communication in the physician–patient interaction. Patient education and counseling, 67, 315-318.
MONTAGUE, E. & ASAN, O. 2014. Dynamic modeling of patient and physician eye gaze to understand the effects of electronic health records on doctor–patient communication and attention. International journal of medical informatics, 83, 225-234.
MOORE, P., VARGAS, A., NÚÑEZ, S. & MACCHIAVELLO, S. 2011. [A study of hospital complaints and the role of the doctor-patient communication]. Revista medica de Chile, 139, 880-885. OROM, H., UNDERWOOD III, W., CHENG, Z., HOMISH, D. L. & SCOTT, I. Y. 2018. Relationships as
medicine: quality of the physician–patient relationship determines physician influence on treatment recommendation adherence. Health services research, 53, 580-596.
PEARSON, S. D. & RAEKE, L. H. 2000. Patients' trust in physicians: many theories, few measures, and little data. Journal of general internal medicine, 15, 509-513.
QUALTRICS. 2019. Qualtrics [Online]. Available: www.qualtrics.com [Accessed].
ROBINSON, J. D. 1998. Getting down to business: Talk, gaze, and body orientation during openings of doctor-patient consultations. Human communication research, 25, 97-123.
RSTUDIO. 2019. RStudio [Online]. Available: www.rstudio.com [Accessed].
RUUSUVUORI, J. 2001. Looking means listening: coordinating displays of engagement in doctor– patient interaction. Social science & medicine, 52, 1093-1108.
SHACHAK, A. & REIS, S. 2009. The impact of electronic medical records on patient–doctor
communication during consultation: a narrative literature review. Journal of evaluation in clinical practice, 15, 641-649.
STEWART, M., BROWN, J. B., BOON, H., GALAJDA, J., MEREDITH, L. & SANGSTER, M. 1999. Evidence on patient-doctor communication. Cancer, 25, 30.
STEWART, M., BROWN, J. B., DONNER, A., MCWHINNEY, I. R., OATES, J., WESTON, W. W. & JORDAN, J. 2000. The impact of patient-centered care on outcomes. J Fam Pract, 49, 796-804.
STEWART, M. A. 1995. Effective physician-patient communication and health outcomes: a review. CMAJ: Canadian Medical Association Journal, 152, 1423.
STREET JR, R. L., LIU, L., FARBER, N. J., CHEN, Y., CALVITTI, A., ZUEST, D., GABUZDA, M. T., BELL, K., GRAY, B. & RICK, S. 2014. Provider interaction with the electronic health record: the effects on patient-centered communication in medical encounters. Patient education and
counseling, 96, 315-319.
THOM, D. H., RIBISL, K. M., STEWART, A. L., LUKE, D. A. & PHYSICIANS, T. S. T. S. 1999. Further validation and reliability testing of the Trust in Physician Scale. Medical care, 510-517. TOBII. 2019a. Tobii Pro [Online]. Available: https://www.tobiipro.com/ [Accessed].
TOBII. 2019b. Tobii Pro Lab [Online]. Available: www.tobiipro.com/product-listing/tobii-pro-lab [Accessed].
VAN DER BIJ, A. K., DE WEERD, S., CIKOT, R. J., STEEGERS, E. A. & BRASPENNING, J. C. 2003. Validation of the dutch short form of the state scale of the Spielberger State-Trait Anxiety Inventory: considerations for usage in screening outcomes. Public Health Genomics, 6, 84-87.
18
A. Appendix A – Wake Forest Physician Trust Scale
Wake Forest Physician Trust Scale
Response categories: Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree. (1) Your doctor will do whatever it takes to get you all the care you need.
(2) Sometimes your doctor cares more about what is convenient for him or her than about your medical needs.
(3) Your doctor’s medical skills are not as good as they should be. (4) Your doctor is extremely thorough and careful.
(5) You completely trust your doctor’s decisions about which medical treatments are best for you.
(6) Your doctor is totally honest in telling you about all of the different treatment options available for your condition.
(7) Your doctor only thinks about what is best for you.
(8) Sometimes your doctor does not pay full attention to what you are trying to tell him or her. (9) You have no worries about putting your life in your doctor’s hands.
19
