Health Information Technology and Physician-Patient Interactions: Impact of Computers on Communication during Outpatient Primary Care Visits
- Affiliations of the authors: Kaiser Permanente Medical Care Program, Division of Research, Oakland, CA (JHs, JHu, VF); The Robertson Group, Lake Oswego, OR (NR); Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University, Portland, OR (HJ); Regenstrief Institute, Indiana University School of Medicine, Indianapolis, IN (RF)
- Correspondence and reprints: John Hsu, MD, MBA, MSCE, Physician Scientist, 2000 Broadway, 3rd Floor, Oakland, CA 94612; e-mail: < >
- Received 12 November 2004
- Accepted 23 March 2005
Objective The aim of this study was to evaluate the impact of introducing health information technology (HIT) on physician-patient interactions during outpatient visits.
Design This was a longitudinal pre-post study: two months before and one and seven months after introduction of examination room computers. Patient questionnaires (n = 313) after primary care visits with physicians (n = 8) within an integrated delivery system. There were three patient satisfaction domains: (1) satisfaction with visit components, (2) comprehension of the visit, and (3) perceptions of the physician's use of the computer.
Results Patients reported that physicians used computers in 82.3% of visits. Compared with baseline, overall patient satisfaction with visits increased seven months after the introduction of computers (odds ratio [OR] = 1.50; 95% confidence interval [CI]: 1.01–2.22), as did satisfaction with physicians' familiarity with patients (OR = 1.60, 95% CI: 1.01–2.52), communication about medical issues (OR = 1.61; 95% CI: 1.05–2.47), and comprehension of decisions made during the visit (OR = 1.63; 95% CI: 1.06–2.50). In contrast, there were no significant changes in patient satisfaction with comprehension of self-care responsibilities, communication about psychosocial issues, or available visit time. Seven months post-introduction, patients were more likely to report that the computer helped the visit run in a more timely manner (OR = 1.76; 95% CI: 1.28–2.42) compared with the first month after introduction. There were no other significant changes in patient perceptions of the computer use over time.
Conclusion The examination room computers appeared to have positive effects on physician-patient interactions related to medical communication without significant negative effects on other areas such as time available for patient concerns. Further study is needed to better understand HIT use during outpatient visits.
Innovations in health information technology (HIT) have great potential for improving the practice of medicine; their use is encouraged by groups including national governments, the Institute of Medicine (IOM), and purchaser coalitions.1 2 3 4 5 6 In particular, computers at the point of care, e.g., in the examination room, provide physicians with real-time access to resources such as an electronic health record, clinical decision support tools, and order entry systems during the medical visit. As examination room computing becomes more popular, it is important to understand the effects of HIT on communication and the patient-physician relationship.
Some studies suggest that HIT can improve the quality and efficiency of care delivery through better decision support.7 Documented benefits include greater adherence to preventive care guidelines, reductions in inpatient medication errors, and reductions in the cost of care.8 9 10 Other studies have found that new technology can have unintended consequences, such as increased medication order errors or increased physician time investments.11 12 13 14 15 16 There is limited information on how computer use affects interactions between physicians and patients.17 Previous studies have had limited ability to differentiate between changes in physician-patient communication related to physicians initially learning how to use the computer system and changes related to physicians integrating computer use into their clinical workflow.
In theory, greater and faster information availability could allow physicians more time to thoroughly explain diagnoses and treatments or address patient concerns. Greater access to information about previous care, medication prescriptions, laboratory test results, or clinical guidelines also could support more productive discussions about medical issues. At the same time, there might be unintended consequences of examination room computing, such as shifting the physician's attention away from face-to-face engagement with the patient and toward the computer screen and lack of focus on patients' psychosocial concerns.18 19 20 Time spent navigating the computer system, searching for information, and documenting visit activities also could leave less time for patient needs, especially given the limited time available for ambulatory visits.21 22 23 As with most change processes, potential adverse effects might be particularly prominent in the period shortly after implementation.
We conducted a longitudinal quantitative study to investigate how the use of computers in ambulatory primary care visits affected physician-patient interactions. Using the longitudinal experience of eight primary care physicians (PCPs), we evaluated patient satisfaction with regularly scheduled visits at three points in time: two months before and one and seven months after the introduction of examination room computers. We addressed three questions: (1) How did examination room computing affect patient satisfaction with various components of the visit, such as time spent on patient concerns? (2) How did examination room computing affect patient comprehension of the visit, such as understanding diagnoses or postvisit needs? (3) How did patient perceptions of the computer use change over time? We hypothesized that patient satisfaction with communication about medical information would increase after the introduction of the computer and that the increased emphasis on medical issues would decrease time available for patient concerns. We also hypothesized that patients would have greater comprehension of the visit and their postvisit needs after the introduction of the computer. Finally, we hypothesized that patient perceptions of computer use would improve significantly during the initial seven months after the computer introduction.
We conducted the study in one freestanding medical office building of Kaiser Permanente-Northwest, a prepaid, integrated delivery system (IDS) in the greater Portland, OR, metropolitan area.
Examination Room Computing
Although examination room computing was new to the study site, PCPs in the four clinics had had access to a commercially available electronic health record and order entry system since 1994. The HIT, developed by Epic Systems, was located in their personal offices but not in the examination room. Before the study, PCPs received basic training in how to use the system, and regularly had to use the computers to enter their progress notes as well as order medications or laboratory tests. All the physicians regularly used the computer system to document visits and enter orders but did not have the computers available in the examination room before the study; this study site was deliberately selected to permit evaluation of the introduction of examination room computers on physician-patient interactions separate from effects associated with learning how to use the system.
In August 2001, the IDS introduced the computer-based system into all the examination rooms of the four clinics. The system hardware consisted of a flat panel computer screen on an adjustable, multidirectional arm, a keyboard and mouse, and a wall-mounted central processing unit (CPU). The spatial relationship between the computer CPU and monitor, the examination table, and the physician's chair varied in each examination room, depending in large part on the room's preexisting architecture. All physicians in the practice used the same examination rooms when seeing their patients; there were no systematic changes in examination room assignments after examination room computers were introduced.
Between the second and third observation periods, PCPs received training in how to integrate computers into the visit. The two-hour on-site workshop involved a didactic lecture on using the computer in an outpatient visit, group assessment of a videotape of an artificial visit, and a role-playing session. The workshop covered communication topics such as making a connection with the patient, making decisions collaboratively, establishing closure for the visit, and expressing empathy for the patient. All physicians in the study completed the training. In addition to the workshop, on-site technical support was available during all clinic hours from two part-time HIT staff persons (equal to one 100% full-time employee).
Population and Study Design
Working with leadership of the clinic and the HIT implementation team, we designed a three-period longitudinal study beginning with a baseline period (two months) before the introduction of examination room computers (P1) and two subsequent points: one month after (P2) and seven months after (P3) their introduction. We recruited PCP volunteers from the four clinics in the IDS. Eligible PCPs included physicians trained in internal medicine and family practice who provided primary care to a regular panel of adult IDS members. Eligible patient subjects included all regularly scheduled IDS members for the PCP. We obtained written consent from all patients, accompanying family members, staff, and physicians involved in the study. The Kaiser Foundation Research Institute Institutional Review Board approved the study.
During the study period between June 2001 and April 2002, there were 17 PCPs trained in internal medicine and family practice who provided care in four clinics at the study site. Of these 17 PCPs, eight agreed to participate in the study. Among patients, the overall participation rate among eligible subjects was 80%.
During one to two days per physician per observation period, research assistants approached all patients in each physician's waiting room. We excluded patients receiving a gynecological examination during the visit. For each consenting patient, we administered pre- and postvisit patient questionnaires, videotaped physician-patient interactions, and videotaped the computer screen. For the videotapes, we mounted one digital video camcorder from the examination room ceiling corner; we used a second camcorder to capture the video feed between the computer and the computer monitor. In this article, we report only on findings from the questionnaires.
In our effort to minimize any intrusions on the medical visit, research assistants performed all the equipment setup, tape changes, and clean up. Physicians, staff, and patients were not responsible for any part of the data collection. The cameras were minimally intrusive; a small red light indicated that the camera was recording. Noise from the cameras was negligible. Physicians, staff, and patients could either cover the camera lens using a special lens net or turn off the camera using a remote control at any point during the encounter.
We pretested the written self-administered questionnaire to assess its general clarity and comprehensibility. After consenting to the study and before seeing the physician, subjects completed a one-page previsit questionnaire. Immediately after their visit, subjects completed a postvisit questionnaire, which assessed satisfaction with the visit, comprehension of diagnosis and treatment plan, satisfaction with examination room computer use, and sociodemographic characteristics. Subjects who were unable to fill out the questionnaire on site had the option of returning it by mail or completing it via telephone interview. Questionnaires were deliberately anonymous to encourage patient participation and candor.
Outcome Variables: Visit Satisfaction
Using items based in part on the Medical Outcomes Study,24 the survey questions addressed patient satisfaction with three visit-related domains: (1) visit components, e.g., overall visit satisfaction, PCP's familiarity with the patient, communication about medical issues, communication about psychosocial issues, and time spent on patient concerns; (2) comprehension of the visit, e.g., understanding visit activities, such as diagnosis or treatment plans and determinations, and postvisit self-care needs, such as potential side effects or complications; and (3) examination room computing, e.g., impact of computer use on comprehension and personalization of care, visit efficiency and flow, and overall satisfaction with computer use (Table 1 for additional details on the wording of each item).24 25 Rather than create summary scores for each group of satisfaction measures (e.g., satisfaction with visit components or satisfaction with psychosocial communication), we present all scores individually to allow readers to interpret each item (Table 1). Responses were based on a six-point Likert scale, which ranged from 1 (excellent) to 6 (very poor), with an additional option of N/A (not applicable).
The unit of analysis was the patient visit. We first compared characteristics of subjects in P2 and P3 with P1; then we evaluated all the satisfaction item responses in P2 and P3 with P1 using multivariate regression models. We examined the effects of examination room computing on satisfaction levels using three different coding schemes for the outcome variables: (1) dichotomous, wherein responses of “excellent” were compared with all other responses; (2) a six-level categorical variable, which ranged from 1 (excellent) to 6 (very poor), and (3) a three-level categorical variable, which included 1 (excellent), 2 (very good), and 3 (good, fair, poor, and very poor). We collapsed the last category due to the few responses in the lowest four levels. We excluded N/A responses and missing values; however, we repeated analyses coding both or either value as either high or low satisfaction. Overall, the findings were robust across all approaches. In this article (Tables 1 and 2), we present the unadjusted percentage of “excellent” responses, i.e., the number of subjects who had excellent satisfaction on each item, and the model results that treated the outcomes as a three-level ordered categorical variable.
In the multivariate logistic and ordinal logistic regression models, we included all the patient demographic characteristics, i.e., age, gender, self-reported health status, race/ethnicity, annual household income, education, and whether the patient had previously seen the PCP. We also examined the contribution of physician characteristics to patient satisfaction in bivariate analyses but excluded them from the multivariate models. Instead, we adjusted for potential clustering of patient responses by PCP through a generalized estimating equation (GEE) approach (PROC GENMOD procedure with REPEATED option in SAS 8.2). We present the data without any specific adjustments for multiple comparisons to allow readers to make their own inferences about the appropriate confidence intervals (CIs).26 In the text, we also focus on the comparison between P3 (seven months after introduction) and P1 (baseline), although the tables show both P2 vs. P1 and P3 vs. P1 comparisons.
Eight PCPs and 313 patients participated in the study: 107 patients in the precomputer baseline period (P1), 81 in the first month after the computer introduction (P2), and 125 in the seventh month after the computer introduction (P3). Table 3 displays the characteristics of the patient subjects. The mean age was 55.2 years old (standard deviation [SD] = 16.5); 63.9% were female; 28.5% reported being in excellent or very good health; 75.4% reported being of white race/ethnicity; 31.6% reported having at least a college degree; 27.2% reported an annual household income of less than $35,000; and 79.9% reported having a previous visit with the PCP before the study visit. There were no statistically significant differences in patient characteristics across the three time periods. Table 4 displays the characteristics of the PCPs participating in all three study periods. The PCPs were evenly divided between the Departments of Family Practice and Internal Medicine; 62.5% were male; 62.5% reported being of white race/ethnicity, and 50% had 3+ years' experience within the health system.
Patients reported that their physician used the computer in the examination room in 82.3% of visits: 84.1% and 81.3% of visits in P2 and P3, respectively. As expected, patient satisfaction with the physician's use of the latest medical technology increased after the computer introduction with 35.4%, 55.7%, and 59.1% reporting “excellent” satisfaction in P1, P2, and P3, respectively (odds ratio [OR] = 1.71, 95% CI: 1.05–2.79 for P2 vs. P1; OR = 2.03; 95% CI: 1.47–2.80 for P3 vs. P1).
Postvisit Satisfaction with Visit Components
In general, patients reported high levels of satisfaction with the visit. Table 1 displays the percentage reporting “excellent” satisfaction with various visit-related items, i.e., a score of one out of six possible choices. There was a significant increase in the level of overall patient satisfaction with the PCP during the visit after the introduction of the computer into the examination room in the seventh month after introduction as compared with baseline, i.e., P3 vs. P1 (OR = 1.50, 95% CI: 1.01–2.22), adjusting for patient age, gender, self-reported health status, whether the visit was an initial visit, household income, and educational attainment, while allowing for clustering by physician. In addition, there was no significant drop in overall satisfaction immediately after the computer introduction (P2 vs. P1).
Familiarity and Medical Communication
Compared with the baseline period, patients in P3 also were more likely to report that physicians were familiar with them as persons (OR = 1.60, 95% CI: 1.01–2.52) and familiar with their medical history (OR = 1.42, 95% CI: 1.03–1.96). Similarly, patients were more likely to be satisfied with the level of communication about their medical care, including the explanation of diagnoses and treatments (OR = 1.61, 95% CI: 1.05–2.47), their participation in the decision-making process (OR = 1.94, 95% CI: 1.12–3.38), and the focus on preventing illness and promoting good health (OR = 1.61, 95% CI: 1.07–2.43).
Psychosocial Communication and Available Time
Patients' satisfaction with communication about psychosocial concerns was not significantly different after the computer introduction compared with the baseline: satisfaction with the personal manner of their PCP (P3 vs. P1, OR = 1.21, 95% CI: 0.70–2.09), with the PCP's concern for their emotional and physical well-being (P3 vs. P1, OR = 0.99, 95% CI: 0.55–1.79) or with how carefully the PCP listened to them (P3 vs. P1, OR = 1.02, 95% CI: 0.61–1.70).
There also were no significant differences in satisfaction with the amount of time available during the visit across the three study periods. For example, there were no statistically significant differences in satisfaction with time spent discussing the main reason for the visit (P3 vs. P1, OR = 1.18, 95% CI: 0.70–1.99), emotional concerns (P3 vs. P1, OR = 1.23, 95% CI: 0.74–2.05), or the total time available to address all concerns (P3 vs. P1, OR = 1.17, 95% CI: 0.70–1.95).
Patient Comprehension of the Visit
Table 1 displays patients' satisfaction levels with their comprehension of the visit. Consistent with satisfaction regarding medical communication, patients at seven months reported having greater comprehension about their medical care during the visit, including understanding of their diagnosis or treatment plan (OR = 1.63, 95% CI: 1.06–2.50), and understanding how their diagnosis or treatment was determined during the visit (OR = 1.65, 95% CI: 1.09–2.50).
There were no significant differences in patient comprehension of medical advice at seven months after computer introduction compared with the baseline. For example, there were no statistically significant changes in understanding self-care activities needed to improve health (P3 vs. P1, OR = 1.29, 95% CI: 0.73–2.27) or knowledge of the potential side effects or complications associated with their treatments or diagnoses (P3 vs. P1, OR = 0.89, 95% CI: 0.53–1.50).
Patient Perceptions of Computer Use during the Visit
Patients reported positive overall impressions of examination room computer use during the visit. The majority of patients (85.4%) reported that they totally agreed (51.4%) or agreed (34.0%) that they liked the way that their PCP used the computer during the visit. In contrast, only 6.2% of patients reported that the computer use created a distraction during the visit; 3.8% and 7.7% in P2 and P3, respectively (p = 0.37 in both bivariate and multivariate analyses). Table 2 displays the changes in perceptions of computer use between P2 and P3. The only statistically significant change in patient perceptions from P2 to P3 was an increase in satisfaction with the computer's effect on timeliness of visit activities (OR = 1.76, 95% CI: 1.28–2.42).
In this longitudinal study of the impact of examination room computing on physician-patient interactions, overall visit satisfaction, satisfaction with the physician's level of familiarity, communication about medical issues, and the degree of comprehension with decisions made during the visit all improved significantly by seven months after implementation. Surprisingly, we did not find that the enhanced medical communication “crowded out” discussions about psychosocial issues or time for patient concerns from the patient perspective, even during the period immediately after implementation. We also did not detect any significant changes in comprehension about post-visit needs or satisfaction with the physician's personal manner, level of concern for the patient, or level of listening. Finally, we detected few changes in patient perceptions of computer use between one month and seven months after implementation.
We originally hypothesized that examination room computing might make the medical decision-making process more transparent and collaborative. In fact, patients reported that their physicians were more familiar with them, communication about medical care was better, and they understood and participated more in the medical decision-making process on average. Increases in satisfaction with the physician's use of the latest technology and familiarity with patients were expected after implementation and serve as a validity check on patient perceptions.
The lack of change from baseline to P2 in satisfaction with available time during visits was surprising. We had anticipated that physicians might have difficulty integrating computer use into their workflow during the initial months, leaving less time for patient needs, i.e., the computer would distract the PCP from the patient. We also hypothesized that availability of computer-based information could place greater emphasis on the medical aspects of the visit, thereby limiting the amount of time available for psychosocial aspects of care; however, patient satisfaction levels do not indicate that either the distraction or crowd-out phenomenon occurred. It is possible that previous experience with the computer-based electronic health record system used in the clinic could account for the absence of patient dissatisfaction after implementation.
Although the findings are generally reassuring, the data suggest opportunities for improving physician-patient interactions. For example, the level of patient comprehension of postvisit needs did not change significantly despite improvement in comprehension about what happened during the visit. Patient perceptions of the quality of computer use also did not appear to change over time, suggesting that time alone might not improve the quality of use. It may be important to continue to monitor computer use well after the initial implementation. Further research is needed to better understand the learning curve associated with successfully integrating examination room computing into ambulatory visits.
Previous studies on the impact of examination room computers are mixed. A few studies have found that introducing computers into examination rooms had an adverse effect on physician-patient communication.18 19 23 For example, using videos of ambulatory care visits, Greatbatch et al.20 found that physicians tended to be preoccupied with computer tasks, which hindered the flow of communication with their patients. These studies may have had limited ability to differentiate between the effects of physicians learning to use computers and electronic health records in the examination room and the office and experienced computer users attempting to integrate computers into the examination room during outpatient visits. A number of studies have found that examination room computers do not diminish patient satisfaction.21 27 28 29 30 31 In some cases, computer use may actually improve certain aspects of physician-patient communication, such as physicians taking a more active role in clarifying information or encouraging patient questions, a finding similar to ours in this study.18
Our findings might differ from other studies because we focused on sampling at three time points rather than a single cross-sectional sample. By measuring multiple time points for each physician, we were better able to control for individual physician behaviors. In addition, by including a second postimplementation period, we were able to account for changes that may have occurred due to greater physician experience in integrating the computer into the visit. Our study also gauged the quality of physician-patient interactions by querying patients directly about their satisfaction levels and separated the responses by measures expected to improve with greater information availability and measures expected to worsen because of greater visit complexity or increased emphasis on medical information. Last, many previous studies were conducted in the late 1980s or early to mid-1990s, when computer systems might have been less user-friendly or physicians and patients less computer savvy.
This study has several notable limitations. First, this was an observational study that relied on a convenience sample of physicians and patients. Because participation in the study was voluntary, there is the potential for selection bias, e.g., early adopters or individuals more predisposed to favor computers in the examination room may be more likely to participate. The observation process and especially the videotaping also could have influenced behavior or perceptions. The study, however, focused on relative changes over time; there is no reason to expect that there would be differential effects across the three time periods.
In addition, we studied a small number of PCPs who practiced in a single clinic, within a single, integrated system. We had limited power to detect small change in our outcomes; nevertheless, we found several significant findings consistent with our hypotheses. We also relied on patient perceptions and did not attempt to directly assess areas such as patient comprehension of self-care practices. Moreover, the setting, types of physicians, and previous experience of all the physicians with the electronic health record may limit the generalizability of these findings to other contexts. We also could not adjust for any secular trends in satisfaction or in ambulatory visits, given the absence of a concurrent control group. To our knowledge, however, there were no such changes during the study period at this clinic. Finally, we did not adjust the statistical analyses for multiple comparisons.26
In conclusion, this early study suggests that soon after the introduction of HIT into examination rooms, physicians used the computers in the majority of ambulatory care visits and that these activities appeared to have positive effects on several aspects of physician-patient interactions including overall visit satisfaction, satisfaction with the physician's level of familiarity, communication about medical decisions, and patient understanding of the medical decisions. There did not appear to be significant negative effects on other aspects of the relationship such as communication about psychosocial needs or available time for patients' concerns. Although these findings are generally positive, much additional research is needed to confirm and elaborate on these findings, and much opportunity remains for improving the quality of physician-patient communication and for improving the integration of computers into the clinical interaction.
Supported by the Garfield Memorial Fund. Neither the funding agency nor the health system had any role in the analysis, interpretation, writing of this report, or decision to submit this manuscript for publication.
The authors thank all the participants of this study, especially the clinicians, staff, and patients at the medical office building where the study was conducted. They also thank Kathy Poteraj, James Kinsman, Mary Reed, Alison Truman, and all the research assistants, without whose assistance this study would not have been possible.