INTRODUCTION

As physicians increasingly integrate the Electronic Medical Record (EMR) into medical practice, it is important to understand its impact on the patient–doctor communication dynamic. Unfortunately, concerns have been raised over physicians who pay more attention to the “iPatient” on the computer screen than to the real patient during a clinical interaction.1

Leading primary care physician organizations issued the Joint Principles of the Patient-Centered Medical Home (PCMH) in February 2007, a model that affirms that patient satisfaction with their doctor is an important marker in health care,2 and that patient compliance,3 health outcomes,4 6 perceptions of physician competence,7 9 and incidence of malpractice suits10 are all closely related to the doctor’s interpersonal skills and quality of the patient–doctor relationship.

While benefits of computerization in health care are well described,11 important drawbacks exist. For instance, some studies show EMR use can prevent doctors from focusing on patients, impede communication, and be detrimental to the patient–doctor relationship.12 15 In order to provide patient-centered care in the digital age, it is critical to understand how EMR use impacts the quality of communication and the patient–doctor relationship.

Two prior systematic reviews have examined the impact of EMR use on patient–doctor communication; however, both had limitations impeding application to current clinical practice.16 , 17 First, due to limited search terminology, publication sources and minimal inclusion of international or inpatient studies, the scope of literature reviewed limits its inclusivity. Second, the studies lack results past 2012, which is prior to increased meaningful use participation, and dates many of the findings. To provide a more comprehensive representation of the current literature, the aims of this systematic literature review were to examine the impact of EMR use on the patient–doctor relationship and communication with a focus on patient perspectives and to identify future directions for study.

METHODS

Data Sources and Searches

We conducted an electronic systematic search of the English literature in Ovid MEDLINE from 1995 to 2015 by exploring Medical Subject Heading (MeSH) terms and keywords related to technology, communication, and relationship terminology in consultation with a biomedical librarian (DW). Given the heterogeneity and lack of standardized terms or MeSH headings used to describe the various types of technologies used in clinical care, additional terms were included (Appendix available online). Only studies or systematic reviews were included; editorials and commentaries were excluded.

We conducted parallel searches in PubMed, Scopus, PsycINFO and the Cochrane Library. In addition, we examined references of prior review articles16 26 and had two independent expert reviewers evaluate the results to ensure key articles were included. To explore publication bias, we reviewed meeting abstracts from two previous years of Society of General Internal Medicine, American Academy of Family Physicians and International Conference on Communication in Healthcare and European Association of Communication in Healthcare conferences for studies that may not have been published.

Inclusion criteria included studies related to EMR use, the patient–doctor relationship, and face-to-face communication. We included all study designs, all patient populations, and international studies. We excluded studies that reported only physician attitudes and perceptions, as well as articles that did not pertain to face-to-face patient–doctor communication (i.e., patient portals and remote EMR access).

Study Selection

Following the initial search, duplicates were eliminated. For the title and abstract review, each article was independently reviewed for inclusion by three co-authors (ML, SI, ANA). Articles were secondarily reviewed by two senior authors (LAA, WWL). For any titles or abstracts that were unclear, authors erred on the side of including for full review.

Data Extraction and Quality Assessment

To ensure consistent article extraction, all reviewers participated in a training process. Ten articles were randomly selected and reviewed by three title abstraction reviewers (LAA, WWL, VGP) to ensure that training was successful and definitions were applied appropriately. All discrepancies were resolved by consensus. Following training, all articles were extracted onto a standard extraction form focused on identifying the following for each study: physician type and characteristics (position such as faculty or residents, age, sex, specialty), patient type and demographics (age, sex, race/ethnicity), study design (observational, RCT, single or pre-post survey), setting (inpatient, outpatient, academic, nonacademic, practice type), recruitment methods, study aims, primary and secondary outcomes, identified barriers & facilitators to patient–doctor communication in the setting of technology use, study strengths and limitations. The validated Downs and Black (DB) checklist27 was going to be used to assess study quality and bias; however, since very few studies were interventional by design, this was not feasible.

Funding for this review was made possible from a grant from the Arnold P. Gold Foundation Research Institute Call for Reviews of Research on Humanistic Healthcare. Funding did not influence our study design, conduct or reporting.

Data Synthesis and Analysis

Authors systematically examined studies qualitatively by comparing the study population, design and outcomes. Studies were sub-divided according to method of data collection. A structured data extraction table was created to facilitate collection of these key elements. Articles not meeting inclusion criteria were excluded. Added to this were studies meeting inclusion criteria identified from reference mining systematic review articles, expert opinion, and review of conference abstracts from unpublished studies.

Our review conforms to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standards.28 Our systematic review did not meet guidelines for submission to a systematic review protocol registry nor did it facilitate a meta analysis due to the varied interventions, methodology and outcomes, reported in our included studies.

RESULTS

Among 7445 total articles identified, 53 were eligible for review (Fig. 1, Tables 1, 2, and 3). Just over half (n = 28) objectively measured communication behaviors using videotaped or direct observation of clinical encounters and 25 examined patient perceptions by survey or interview. Seven studies examined both patient perceptions and observed behaviors.30 , 31 , 43 , 44 , 48 , 49 , 55 Only two studies were interventional in nature.61 , 70 Forty-seven of the 53 studies were published after the year 2000, and 19 (35.8 %) were published since 2011. Only nine studies reported on both pre- and post-EMR implementation findings,2 , 12 , 49 , 56 61 of which only one included both patient perceptions and objective behavioral analysis.49

Figure 1
figure 1

PRISMA 2009 flow diagram.

Table 1 Behavioral Analysis Studies (n = 28)
Table 2 Patient Perceptions Studies: Pre- and Post-EMR Surveys (n = 8)
Table 3 Patient Perceptions Studies: Cross-Sectional Surveys (n = 17)

Study Setting

Thirty-one (58 %) studies were conducted in the US. Most were conducted in an outpatient setting (n = 51, 96 %), with two (4 %) in the inpatient setting. Of outpatient studies, 39 (76 %) were in an adult primary care (i.e., family practice, internal medicine) clinic; only seven (14 %) included pediatric patients or their families. Eight studies (16 %) took place in a specialty clinic. Approximately half (n = 28) were at a single clinic or institution, one-third (n = 19) included multiple sites (range: 2–78), and one examined 2988 unique hospitals.

Twenty of the studies (42 %) were conducted at academic or academically-affiliated training sites; however, only six (26 %) examined outcomes related to residents or their patients.

Behavioral Analysis Outcomes

Of 28 studies15 , 29 55 utilizing behavioral analysis to objectively observe communication behaviors, 23 (82 %) analyzed videotaped interactions and five (18 %) used direct observation.42 , 44 , 45 , 49 , 54 An average of 162 patients (range 10–1170) and 15 providers (range 3–59) participated per study, and three studies included resident observations.37 , 40 , 44

Characterizing EMR Communication Behaviors

Six studies quantified EMR use during a clinical encounter, with an average of 32 % of visit time spent using the computer (range 12–55 %).37 , 42 , 45 , 51 53 Although many studies reported long periods of silence during the encounter, only one study actually defined it as a percentage of the interaction (12 %, mean duration 15.7 s).47 Studies reported changes in speech style of both providers (i.e., abrupt topic shifts)29 , 30 , 42 , 48 , 51 and patients (i.e., synchronizing speech with typing pauses).29 , 30 , 50 Eight studies described variation in the amount and manner in which the EMR was used.15 , 36 , 39 , 41 , 46 , 48 , 51 , 54 Four studies examined typing behaviors33 , 37 , 42 , 48 and six reported on screen positioning, with only 8–10 % active screen sharing during the visit.29 , 30 , 33 , 37 , 38 , 52 Interestingly, one study noted patients had a more positive attitude towards the EMR when they were shown the screen.30

Provider multitasking was another theme that emerged, highlighting providers were unsuccessful at concentrating on complex computer interactions while attending to the patient simultaneously.29 , 33

There were also instances of communication behaviors that researchers believed promoted communication. Four studies noted that EMR use appeared to facilitate clarification, questions and discussion, as well as more open-ended questions and partnership strategies.32 , 34 , 35 , 40 Specific behaviors that seemed to facilitate a more patient-centered interaction included actively inviting patients to look at the screen and using it as an educational tool (i.e., showing test results), signposting computer use, maintaining eye contact, cessation of computer use when patients spoke about sensitive or important topics, continued verbal and nonverbal cues of listening, and reading aloud while typing.30 , 34 , 38 , 42 , 54 Additionally, being able to make computer use less obvious (i.e., typing softly, continuing to speak while typing) resulted in fewer patient speech pattern modifications.29

Five studies29 , 32 , 36 , 40 , 49 included both a pre- and a post-EMR implementation observation group; however, only two paired findings were to the same physician at both points.29 , 36 Paired observations showed greater doctor preoccupation with computer use and alterations in doctor and patient speech patterns, such as delaying speech until finished with the computer.29 Two studies demonstrated doctors tended to adopt a more active role in clarifying information and encouraging questions when the EMR was used.32 , 40

Correlating EMR Communication Behaviors with Patient Perceptions

While 11 studies29 31 , 34 , 37 , 38 , 43 , 44 , 48 , 49 , 55 attempted to correlate objective observations of communication behaviors with patient perceptions of care, only seven of these studies elicited patient perspectives directly.30 , 31 , 43 , 44 , 48 , 49 , 55 The remaining studies used researcher perceptions of the patient perspective as a proxy. Studies noted mixed patient perceptions. An increase in provider screen gaze, keyboarding, silence and closed body posturing negatively impacted communication.43 , 55 However, certain behaviors enabled more successful integration of the EMR into the visit, such as screen sharing that did not obstruct the visual field between doctor and patient.30 Three studies directly examined patient perceptions of change in overall patient–doctor relationship and quality of care or satisfaction overall, and found no significant change as a result of EMR use.31 , 43 , 49 While two studies noted high rates of satisfaction or trust of their doctor, these studies did not report baseline data, thus making it unclear if there was any change in satisfaction related to the introduction of the EMR.48 , 55 Furthermore, two qualitative studies showed a mix of positive, negative and neutral patient responses without quantifying of the effect.30 , 44

Patient Perceptions: Pre- and Post-EMR Surveys

Eight studies used pre- and post-EMR patient surveys as their only method of data collection, with a range of 100 to 18,897 patients responses.2 , 12 , 56 61 Five2 , 56 58 , 60 studies (63 %), two of which had sample sizes over 10,000, found that most patients reported no change in measures of overall patient satisfaction, communication and the patient–doctor relationship as a result of the introduction of technology into the face-to-face clinical interaction.

Three12 , 59 , 61 studies (38 %) reported largely positive satisfaction with communication and patient–doctor relationship as a result of EMR use. One of these was unique because it was one of only two inpatient studies and it directly enabled patients to interact with the EMR.61 In this study, Furness et al. examined the effect of allowing inpatient trauma patients to view their radiographic images on a tablet with their consultant. After the introduction of tablets, patients perceived significantly more involvement in their care decisions and being given the “right amount of information” about their treatment as compared to before the introduction of tablets.61 Lastly, two studies reported that patients perceived their quality of care as higher with EMRs.56 , 59

Patient Perceptions: Cross-Sectional Surveys

Seventeen studies used single cross-sectional patient surveys as their only method of data collection, with a range of 65–518 patient participants per study. Nine of these examined patient perceptions of physician distraction by the computer, with a range of 3–40 % (mean 18 %) of patients expressing some level of concern.63 , 64 , 66 69 , 72 , 76 , 78

Eleven studies examined global perceptions, with eight62 , 67 71 , 73 , 77 studies (73 %) reporting no change in overall patient satisfaction, communication or the patient–doctor relationship as a result of the introduction of EMR. One study (9 %) demonstrated equally mixed positive, negative and neutral patient satisfaction,76 and two studies (18 %) demonstrated a majority of positive outcomes.72 , 75 Only one study reported patient-perceived quality-of-care (QOC), with the majority of patients reporting technology contributed to a better QOC.67

The remaining six cross-sectional studies63 66 , 74 , 78 (35 %) in this group also examined patient perceptions, but lacked global measures such as overall satisfaction with communication or the patient–doctor relationship. It appears, however, that they contained more positive (i.e., use of the computer was a “good thing”)63 than negative (i.e., the computer interfered with my doctor’s ability to hear my complaints)66 patient comments.

One study used a “patient-centered” spatial arrangement of the room and computer, and found no difference in patient satisfaction or perceptions of communication quality with the ergonomic change.70

Characterizing Positive Deviants

An important but limited number of studies (n = 4) examined increases in patient understanding of their condition as a result of their provider using the EMR in the clinical interaction, demonstrating increased perceptions of empowerment and informed decision-making.12 , 59 , 61 , 66 Also, of the 22 total articles examining impacts of the EMR on overall patient perceptions of satisfaction, communication or the patient–doctor relationship as a result of EMR use, five (23 %) found positive changes and these are important to highlight.12 , 59 , 61 , 72 , 75 Three of these (60 %) were conducted outside of the US in countries in which a Universal Health Care system exists (UK, Canada, Germany).61 , 72 , 75 Two focused on the use of a somewhat novel technology aide; one using a tablet to view radiologic images61 and another using an EMR decision aid,75 both of which resulted in increased satisfaction with the encounter, counseling, and involvement in their care. Of the two US studies, Hsu et al.’s Kaiser study was remarkable in that it was the only study that provided physician training on how to integrate computers into the visit.12 , 59 Although there was a decrease in patient satisfaction after physician training, from 67 % 1 month post-EMR (pre-training) to 63 % 7 months post-EMR (post-training), there was increased overall patient satisfaction 7 months post-EMR introduction compared to baseline.12 Due the observational design of the study, it is unknown whether the changes in satisfaction were related to the training; however, it is an important finding.

Interestingly, a greater percentage of positive studies emanated from the international community, with 43 % (n = 3 of 7 total studies)61 , 72 , 75 noting overall positive changes in satisfaction in communication or the patient–doctor relationship as a result of technology use versus 13 % (n = 2 of 15 total)12 , 59 of US studies.

CONCLUSION

This systematic review of the impact of the EMR on the doctor–patient relationship and communication found while physicians exhibited potentially negative communication behaviors with EMR use (i.e., interrupted patient and doctor speech patterns, increased gaze shifts and episodes of multitasking, and low rates of sharing the computer screen with patients), the majority of studies examining patient perceptions reported no change in overall patient satisfaction, communication, or the patient–doctor relationship. Furthermore, some studies identified instances in which patients felt the EMR facilitated the process of communication, clarification, and discussion as well as some potentially patient-centered communication behaviors. These “best practices” may be taught to providers in order to guide them towards more successful and collaborative EMR use. Given that the majority of studies were conducted in adult primary care clinics, these findings are highly pertinent to adult providers since communication is key to the patient–doctor primary care relationship and patient outcomes.3 10 Lack of change in overall patient perceptions may be surprising to clinicians, given accounts of negative provider attitudes to EMR implementation.2 , 65 However, knowing that patient perceptions did not suffer, providers and administrators should not be deterred by fears of its adoption and instead learn to actively use it in a more patient-centered manner.

It is important to reflect on the five of 22 studies that reported positive changes in overall patient satisfaction, communication or the patient–doctor relationship as a result of EMR use. These positive patient perceptions are perhaps reflective of a different culture of EMR use in these settings, and an increased acceptance in other countries or highly integrated healthcare systems. It is possible that improved patient and provider familiarity with the EMR in these environments created a different culture of practice that enabled EMR use in a more patient-centered manner. Comparatively, US patients and physicians are perhaps not as cognizant or experienced in achieving this, as evidenced by the greater positive patient perceptions abroad. Given the tremendous potential of EMR integration for patient education, it is important to highlight best practices in order to maximize EMR use as an educational tool.

It is also worth considering why patient and physician perception discrepancies regarding EMR use exist. Patient satisfaction or their perceived quality of care may be driven by factors other than provider communication behaviors. For example, the EMR may improve clinical efficiency by making it easier for physicians to communicate with other providers, and in turn patients may perceive physician technology use as positive overall. Because the majority of studies were conducted in adult primary care settings, strong patient–doctor relationships may have contributed to patients being more accepting of their doctors being unfamiliar with the EMR at first and slowly becoming facile with the EMR. Additionally, patients may not consciously notice behavior differences as much as trained observers.

This review identifies the need for further study in a variety of areas related to EMR use. Future work should correlate observed physician behavior with direct patient perceptions rather than a trained observer as proxy, in order to identify how to best use the EMR during clinical interactions to engage patients in their care. Objective studies should further explore how to integrate EMR use to enhance patient engagement and communication.

Also, few studies took place in primary care academic settings, which is particularly interesting due to issues around the hidden curriculum and potential negative role-modeling by attendings, given the lack of training on patient-centered EMR use. This highlights the need to study academic settings further, and to develop and implement effective curricula for all providers on how to use the EMR to enhance patient–doctor communication.

In the future, greater attention should be given to studies outside of adult outpatient primary care. Since high levels of continuity may influence patient perceptions of EMR use, studies should specifically focus on inpatient or specialty settings to understand the impact of EMR use in low continuity settings. Also, given the increasing rates of technology adoption by younger “millennial” trainees (i.e., fellows, residents, medical students), further studies should look at how this group may differ from older providers in their EMR use. Lastly, future research should utilize randomized study designs where possible; for example, randomizing providers to EMR training and directly eliciting patient experience regarding technology use and the impact on patient–doctor communication.

Although this systematic review found several significant findings, there are important limitations to note. For instance, while nearly one-third of studies examined patient perceptions, the heterogeneity in the type of questions asked and lack of global measures such as overall satisfaction limited the ability to compare findings within this cohort. Also, analysis of the included studies reveals potential areas where study bias could exist. The majority of studies used direct observation methods, which are a proxy for the patient’s experience and are subject to inter-observer variability when multiple individuals are observing and reporting on behaviors observed. Interviewer bias could also have occurred when those observations were followed by questioning from the study personnel. There was also the potential for publication bias, and while we sought to address this by reviewing abstracts from related meetings, we were not able to review abstracts for all possible related meetings and could only review studies published in English. Another limitation is the paucity of studies documenting both specific observed communication behaviors pre- and post-EMR in addition to eliciting direct patient perceptions. With increasing rates of EMR adoption, it will be harder to conduct such a pre-post EMR study. Also, for those studies where the is no pre-EMR observation, it is quite possible that these providers were at baseline poor communicators, and thus the introduction of the EMR is not to account for the negative behaviors observed, but rather they are reflective of the providers’ poor baseline communication ability.

Reliance on convenience samples of both physician and patient subjects may have contributed to selection bias in both groups, as only one study70 was randomized. Most studies identified had small numbers of physician and patient participants as well as study sites. As such, external validity of the findings and the ability to generalize them to other groups or populations is not known. In addition, because many of the studies were observational in nature, causal inferences could not be made and unmeasured confounders may exist. Lastly, multiple variables contribute to the overall experience of the patient–doctor relationship and communication, and it is quite plausible that some other factor is contributing to the observations and effects seen.

In conclusion, it appears EMR use can improve patient understanding of conditions and treatment plans, and increase sharing and confirmation of medical information. Several studies identify behaviors that appear to facilitate patient-centered communication (i.e., screen sharing, signposting, cessation of typing during sensitive discussions) and future work should adapt these best practices into a curriculum to teach providers how to integrate patient-centered EMR use into their clinical workflow. Medical education targeting the continuum of learners can address this gap in training and help foster humanistic patient–doctor-EMR interactions in the digital age.