The effect of tablet computers with a mobile patient portal application on hospitalized patients’ knowledge and activation

Abstract

Objective To assess the effect of tablet computers with a mobile patient portal application on hospitalized patients’ knowledge and activation.

Methods We developed a mobile patient portal application including pictures, names, and role descriptions of team members, scheduled tests and procedures, and a list of active medications. We evaluated the effect of the application using a controlled trial involving 2 similar units in a large teaching hospital. Patients on the intervention unit were offered use of tablet computers with the portal application during their hospitalization. We assessed patients’ ability to correctly name their nurse, primary service physicians, physician roles, planned tests and procedures, medications started, and medications stopped since admission. We also administered the Short Form of the Patient Activation Measure.

Results Overall, 100 intervention- and 102 control-unit patients participated. A higher percentage of intervention-unit patients correctly named ≥1 physician (56% vs 29.4%; P < .001) and ≥1 physician role (47% vs 15.7%; P < .001). Knowledge of nurses’ names, planned tests, planned procedures, and medication changes was generally low and not significantly different between the study units. The Short Form of the Patient Activation Measure mean (SD) score was also not significantly different at 64.1 (13.4) vs 62.7 (12.8); P = .46.

Conclusions Additional research is needed to identify optimal methods to engage and inform patients during their hospitalization, which will improve preparation for self- management after discharge.

INTRODUCTION

Hospital settings present unique challenges to patients’ capacity to learn about their care. Patients often do not have a prior relationship with hospital professionals, and hospital team memberships are highly dynamic. ¹ Patients frequently have multiple active conditions and undergo complex tests, procedures, and treatments. Physicians provide more information than a typical patient can retain and seldom complement the verbal information provided during daily rounds with any other mode of communication. ^2,3 Not surprisingly, research has shown that many hospitalized patients do not know the names and roles of their hospital physicians, planned tests and procedures for the day, and the names of medications they are receiving. ^4–9 These knowledge deficits have important implications with regard to consent for treatment and the ability of patients to assume responsibility for their own care after discharge.

The use of an electronic health record (EHR) creates the opportunity to assemble clinical information residing within the system into tools designed to inform and engage patients. A patient portal is one such tool, providing convenient 24-hour access to personal health information from any location with an Internet connection. ¹⁰ Although the degree of benefit is still unclear, the use of patient portals is increasing in ambulatory settings. ^11,12 Very little research has been dedicated to assessing the effect of patient portals on hospitalized patients, many of whom receive primary care elsewhere and do not have an Internet capable device with them in the hospital. ¹³ Vawdrey and colleagues ¹⁴ used tablet computers with a patient portal application to provide information about team members and medications to hospital patients. Although patients were enthusiastic about the application, the study was small (5 patients) and did not assess the application’s impact on patient comprehension. Greysen and colleagues ¹⁵ provided tablet computers with an educational module on patient safety and patient portal access to 30 hospitalized patients. Patients were highly satisfied with the device, but the intervention’s impact on their knowledge and engagement in care was not assessed.

We sought to assess the effect of tablet computers with a mobile patient portal application on hospitalized patients’ knowledge and activation. We hypothesized that use of the patient portal would result in greater knowledge of team members’ names and roles, planned tests and procedures, medications, and higher patient activation.

METHODS

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital, a large academic hospital in Chicago, Illinois. We conducted a controlled trial, selecting 1 of 2 similar general medical teaching service units as the intervention unit while the other served as the control. Patients were assigned to these 30-bed units by the hospital admissions office based on bed availability. Six physician teams provided care for patients on both units. Each physician team consisted of 1 attending, 1 senior resident, and 1 to 2 interns. Nonphysician staffing levels were identical across the study units. The Northwestern University Institutional Review Board approved the study.

Development of the Patient Portal

We identified patients’ preferred content for the portal from a prior study by our research group. ¹³ We then created a test version of the portal application, leveraging information from within our EHR (PowerChart Millennium; Cerner Corporation, Kansas City, Missouri).

Within the portal, information was organized into pages reflecting general patient information, the care team, the medication list, and the agenda for the day ( supplementary appendix 1 ). General patient information included documented allergies, the name of the patient’s primary care physician, and the problem list. Care team information included names and pictures of the patient’s primary service physicians and nurse along with descriptions for each role (ie, intern, resident, attending physician, nurse). Care team names were populated from data within the care team tab in the EHR and linked to photograph files for each member. The medication list was populated from active medication orders (scheduled and as needed) with hyperlinks to UpToDate (Wolters Kluwer Health, Alphen aan den Rijn, Netherlands) patient drug information pages. In addition to mealtimes, the agenda included a list of scheduled tests and procedures populated from information in the scheduling component of the EHR. Throughout the portal pages, we made deliberate efforts to use language that was simple and easy to read. We presented the test version of the portal to the Northwestern Memorial Hospital Patient and Family Advisory Council for feedback. The Council was generally pleased with the portal and had no recommendations for revision.

The portal application was installed on 15 Apple iPads with Retina display Wi-Fi 16 GB, fourth generation, (Apple Inc, Cupertino, California) for use in the study. The iPads were configured with a Mobile Device Management Solution provided by Symantec Corporation, so they would only be functional on the secured hospital network. Unique passwords were installed on each iPad prior to distribution, and patients were allowed to update the password if desired. The iPads had a fully enclosed OtterBox (Otter Products, LLC, Fort Collins, Colorado) protective cover and screen protector and were outfitted with a stand for landscape viewing. We factory restored the iPads between uses and sanitized them in accordance with hospital recommendations.

Participant Enrollment and Data Collection

Each weekday morning from June through October 2014, a research coordinator (MEL or AK) approached newly admitted patients (ie, morning of second hospital day) on the intervention unit and offered them the use of an iPad, loaded with the patient portal application, for the duration of their hospital stay. We excluded patients whose preferred language was not English and those disoriented to person, place, or time. Patients were also excluded if they had physical or visual impairments that impeded use of a tablet computer. Enrolled patients were given a brief orientation to the iPad and the patient portal application, and given contact information for one-on-one support. We conducted structured interviews of the patients enrolled on the intervention unit as well as patients on the control unit during the afternoon of their second or third hospital day. For each patient interviewed, research coordinators also conducted a structured interview of his or her resident physician.

Third level

 

Measures

We assessed patients’ frequency of use and satisfaction with the portal through questions specifically created for this study. We used a 5-point Likert scale for satisfaction items. Patients’ knowledge of care team members and the plan of care was assessed using methods adapted from prior research. ^4,6–8 Specifically, the research coordinator asked patients to name their nurse for the day, their hospital physicians, and to identify each named physician’s role. Patients were then asked what tests and procedures were planned for the day, what medications had been started since admission, and what home medications had been stopped since admission. Patients’ responses were recorded verbatim by the research coordinator for comparison to data obtained from physician interviews and the medical record. Study patients’ physicians were asked to verify physician team member names and what tests and procedures were planned for the day. This information was combined with data from the EHR indicating ordered tests and procedures at the time of the patient interview, the home medication list, and the inpatient medication list at the time of the patient’s interview.

A physician researcher (GRS) performed reviews, reconciling patients’ and physicians’ verbatim responses and medical record data, in a blinded fashion to determine patients’ knowledge regarding team membership and aspects of the plan of care. Correctness for each aspect of the plan of care was categorized as none, partial, or complete, allowing for different language with similar meanings, as done in our prior research. ⁶ For example, if the patient responded that planned testing for the day included only an ultrasound of the legs and the physician stated the patient was planned to undergo a lower extremity venous duplex study and a chest computed tomography, partial correctness was given. If the patient responded that the planned procedure for the day was an angiogram and the physician responded that the patient was planned to undergo a cardiac catheterization, then that aspect of care (planned procedures) was rated as completely correct. If a patient indicated that he or she did not know the plan, the aspect was coded as incorrect.

Patient activation was assessed using the Short Form of the Patient Activation Measure (PAM-SF). ¹⁶ The PAM-SF is a 13-item instrument (score ranging from 0 to 100) shown to be reliable and valid in clinical settings. ^16–19 A recent study found that hospitalized patients with lower PAM scores had higher rates of utilization at 30 days after being discharged . ²⁰

Data Analysis

Information from structured interviews was combined with data from the Northwestern Medicine Enterprise Data Warehouse, an integrated repository of all clinical and research data for patients receiving care in the system. Primary discharge diagnosis ICD-9 codes were grouped using the Healthcare Cost and Utilization Project Clinical Classification Software. ²¹ We compared patient characteristics using the chi-square test and t test. We report descriptive statistics for frequency of use and satisfaction with the portal. For ease of interpretation, we dichotomized variables related to patients’ knowledge of physicians and aspects of the plan of care. We compared the percentage of patients knowing 1 or more physician name, 1 or more physician role, and the name of the nurse, using chi-square tests. Similarly, we compared the percentage of patients with complete knowledge of planned tests, planned procedures, medications started since admission, and home medications discontinued since admission, using chi-square tests. We compared the PAM-SF score using t tests. In light of our finding that intervention-unit patients were younger than controls, we also performed multivariable regression analyses using age as a covariate. We estimated a need to enroll 200 total patients to have >80% power to detect a 60% increase in patients’ knowledge of 1 or more physician and a 40% improvement in patients’ knowledge of planned tests. These anticipated results were felt to be both clinically relevant and achievable based on prior research. ^6,22 All analyses were conducted using Stata Version 11.2 (StataCorp LP, College Station, Texas).

RESULTS

Participation and Patient Characteristics

Overall, 283 intervention-unit patients were approached for use of the patient portal during their hospitalization. Fifty-nine were excluded because of disorientation, 19 were excluded because their preferred language was not English, 21 were excluded because of physical or visual deficits, and 64 declined to participate in the study. One hundred of 120 (83%) patients given the patient portal completed the structured interview. Reasons for not completing the interview included early discharge, clinical deterioration, and scheduling conflicts. Overall, 184 control-unit patients were approached for a structured interview. Thirty-three were excluded because of disorientation, 15 were excluded because their preferred language was not English, 6 were excluded because of physical or visual deficits, and 28 declined to participate in the study.

Although patients were similar in sex, race, case mix, educational level, Elixhauser index, ²³ length of stay, and discharge disposition ( table 1 ), intervention-unit patients’ mean (SD) age was younger at 46.7 years (16.7) vs 51.4 years (17.3); P = .05.

Table 1.

Patient characteristics

Characteristics	Control Unit	Intervention Unit	P value
	( n = 102)	( n = 100)
Age, mean (SD), y	51.4 (17.3)	46.7 (16.7)	.05
Women, No. (%)	54 (52.9)	63 (63)	.15
Race, No. (%) Nonwhite	44 (43.1)	44 (44)	.90
Day of interview, No. (%)			.21
2	54 (52.9)	53 (53)
3	48 (47.1)	47 (47)
Case Mix, No. (%)			.31
Diseases of the digestive system	25 (24.5)	30 (30)
Diseases of the circulatory system	14 (13.7)	12 (12)
Symptoms, signs, and ill-defined conditions	9 (8.8)	10 (10)
Diseases of the skin and subcutaneous tissue	5 (4.9)	7 (7)
Infectious and parasitic diseases	6 (5.9)	6 (6)
Injury and poisoning	11 (10.8)	1 (1)
Diseases of the genitourinary system	4 (3.9)	6 (6)
Diseases of the musculoskeletal system	6 (5.9)	4 (4)
Diseases of the respiratory system	4 (3.9)	6 (6)
Other	18 (17.7)	18 (18)
Highest education completed, No. (%) a			.65
Did not complete high school	7 (7)	4 (4)
High school	44 (44)	39 (39.4)
4-year college	23 (23)	25 (25.3)
Advanced degree	26 (26)	31 (31.3)
Elixhauser score, mean (SD)	4.4 (6.6)	3.7 (6)	.44
Length of stay, mean (SD), d	4.4 (5.7)	4.3 (6.2)	.89
Discharge disposition, No. (%)
Home	98 (96.1)	96 (96)	.26
Skilled nursing facility/acute rehab	4 (3.9)	2 (2)
Other	0 (0)	2 (2)

^a For highest education completed, control unit n  = 100 and intervention unit n  = 99.

Use and Satisfaction with the Patient Portal Application

Use of the patient portal application varied among intervention patients, with 57% (57 of 100) using it more than once per day and 20% (20 of 100) never using it ( table 2 ). Many patients (18%, 18 of 100) allowed visitors to use the portal. Overall satisfaction with the portal was high, with 76% (76 of 100) indicating the portal was easy to use and 71% (71 of 100) indicating that the portal provided useful information.

Table 2.

Use and perceptions of the portal

	No. (%) of Patients Using the Portal ( n = 100)
Frequency of use
Never	20 (20)
Less than once a day	4 (4)
Once a day	19 (19)
More than once a day	57 (57)
Allowed a visitor to use portal	18 (18)
Satisfaction with portal a
The patient portal was easy to use	76 (76)
The patient portal provided useful information	71 (71)

^a Number and percentage of respondents who agreed or strongly agreed.

Knowledge of Team Members, Roles, and Aspects of Care

A larger percentage of intervention-unit patients correctly named 1 or more of their hospital physicians (56% vs 29.4%; P < .001) and the role of 1 or more physicians (47% vs 15.7%; P < .001) ( table 3 ). A similar percentage of intervention- and control-unit patients were able to correctly name their nurse (76% vs 70.2%; P = .39). Results were similar in multivariable analyses controlling for patient age.

Table 3:

Patient portal effect on patients’ knowledge of the care team

	Unadjusted Analysis			Adjusted Analysis a
Effect, No. (%)	Control Unit	Intervention Unit	P value	Odds Ratio (95% CI)	P value
	( n = 102)	( n = 100)
Patient correctly named ≥1 hospital physician	30 (29.4)	56 (56)	< .001	3.06 (1.70-5.51)	< .001
Patient correctly named role of hospital physician	16 (15.7)	47 (47)	< .001	4.87 (2.49-9.53)	< .001
Patient correctly named nurse	72 (70.6)	76 (76)	.39	1.27 (0.68-2.40)	.45

^a Adjusted for patient age.

We found no difference in the percentage of intervention-unit patients with complete knowledge of planned tests, planned procedures, new medications, or discontinued medications compared with control-unit patients ( table 4 ). Results were similar in multivariable analyses controlling for patient age.

Table 4:

Patients’ knowledge of the care plan

	Unadjusted Analysis			Adjusted Analysis a
Aspect of Care, No. (%) b	Control Unit ( n = 102)	Intervention Unit ( n = 100)	P value	Odds Ratio (95% CI)	P value
Planned tests	70 (68.6)	63 (63)	.40	0.75 (0.41-1.35)	.33
Planned procedures	97 (95.1)	90 (90)	.17	0.40 (0.13-1.24)	.11
New medications since admission	11 (10.9)	17 (17)	.21	1.74 (0.76-3.97)	.19
Discontinued medications since admission	45 (44.6)	51 (51)	.36	1.18 (0.66-2.08)	.58

^a Adjusted for patient age.

^b Percent rated as having complete knowledge.

Patient Activation

The mean (SD) PAM-SF score was higher for intervention-unit patients, but the result was not statistically significant at 64.1 (13.4) vs 62.7 (12.8); P = .46. Similarly, the PAM-SF was not significantly higher in multivariable analyses controlling for patient age (β = 1.58; P = .41).

DISCUSSION

We found that hospitalized patients given a tablet computer with a mobile patient portal application had greater knowledge of the names and roles of their hospital physicians. Knowledge of nurses’ names, planned tests, planned procedures, new medications, and discontinued medications was not different between intervention- and control-unit patients. Similarly, patient activation was not different between intervention- and control-unit patients.

There are several potential explanations for our findings. First, it may have been relatively easy for patients to learn physicians’ names and roles and more difficult for them to learn about planned tests, procedures, and medication changes, all of which the terminology may be unfamiliar. Prior studies have shown that simple interventions such as use of a physician facecard can improve patients’ knowledge of the names and roles of physicians. ^22,24 Second, patients may not have had enough time to review the portal and acquire new knowledge about planned tests, procedures, and medication changes. We conducted interviews in the afternoon of the same day or the day after patients received the tablet computer in an attempt to optimize efficiency of data collection. Allowing more time to review information in the portal may have resulted in greater acquisition of knowledge about the plan of care. Perhaps most importantly, a fifth of intervention-unit patients did not use the portal. Studies in ambulatory settings have similarly found that many patients offered access to a patient portal do not enroll, and that many who enroll do not use the portal. ^25,26 A lack of motivation has been identified as a contributing factor to nonuse of ambulatory portals. Engaging and motivating hospitalized patients may be especially challenging, as prior research suggests that hospitalized patients accept and often prefer a passive role during their acute illness. ^27,28 A recent systematic review concluded that there are considerable gaps in our knowledge regarding methods to engage patients during hospitalization, and that further research is needed to identify optimal methods. ²⁹

Enhancements to functionality may increase the likelihood of portal use and the potential to improve knowledge and activation. The portal application used in our study did not display results of tests or procedures. In the ambulatory setting, physicians often approve the release of results to their patients. Requiring physician approval to release each result may not be feasible in the hospital setting due to the large volume of results. One approach may be to release preselected results (eg, chemistries but not pathology) to patients at a specified time each day (eg, noon for results prior to 7 AM). Secure messaging between the patient and his or her physician(s) is another patient portal feature used to engage patients in the outpatient setting. Messaging via the portal may be considered for use in hospital settings but would require workflow design to ensure that messages are received by the most appropriate clinician (eg, nurse, physician, social worker). Future research should be conducted to identify features most useful to patients and determine how to implement enhancements without negatively impacting clinician workflow. In our study, research coordinators distributed tablet computers. Large-scale implementation of tablet computers would require identification of the best team member to administer the tablets, teach patients how to use them, and follow up on questions they have.

Importantly, we found that many patients allowed visitors to use the portal application. Torke and colleagues ³⁰ recently reported that surrogates are involved in decision making for nearly half of hospitalized older adults. Offering use of a patient portal to designated surrogates may be another important enhancement to hospital-based patient portals. Because many surrogates are unable to be present during physicians’ rounds, portal access could be offered remotely. Sarkar and Bates recently recommended that health care organizations develop approaches to allow patients to give permission to caregivers and care partners to access their portal and emphasized the need to ensure strong authentication procedures using a unique login and password for each designee. ³¹

Studies like ours—evaluating methods to improve hospitalized patients’ knowledge of their plan of care—are important because the principles of informed consent apply to any treatment that carries risk, including the prescription of medications and initiation of diagnostic evaluations. ^32–34 Moreover, better understanding of the inpatient plan of care lays the foundation for patients to assume their own care after discharge. Patients experience difficulty adhering to medication regimens that have undergone multiple changes during a hospitalization. ^35,36 Similarly, patients frequently do not complete recommended outpatient workups after hospital discharge. ³⁷ Such failures often reflect inadequate preparation of patients and their caregivers as they transition from the hospital to becoming fully responsible for their care. ^38,39 Focusing educational effort at the time of discharge misses the opportunity to provide ongoing education throughout the hospital stay. Research shows that hospitalized patients have ample time ² and motivation to engage in health education. ⁴⁰

Providing hospitalized patients with concurrent access to selected chart content and to electronic educational materials is consistent with Meaningful Use objectives for hospitals and providers. ^41,42 Existing Stage 2 requirements for hospitals allow for delays in online chart content access for up to 36 hours following discharge. As currently proposed for Stage 3, the threshold percentage of patients required to possess electronic access to their chart content rises. ⁴³ In addition, the potential wait time for information to be made available following discharge is reduced. Establishing electronic chart access during a hospital stay offers the prospect of more easily meeting existing and future requirements.

Our study has several limitations. First, we assessed the impact of a custom designed, mobile patient portal application at a single site. Many EHR vendors have developed patient portals and some have developed mobile applications for these portals. Features and functionality differ among patient portals and their corresponding mobile applications. Importantly, most portals have been developed for ambulatory care. Our portal application included features designed specifically for hospitalized patients. Second, intervention-unit patients were younger than control-unit patients in our study. Intervention-unit patients consented to participate in both use of the tablet computer and completion of the structured interview, while control patients consented only to completion of the structured interview. Older patients may have been less comfortable with the use of a tablet computer and therefore less likely to participate among eligible intervention-unit patients. We performed multivariable analyses to control for differences in patient age and found similar results to unadjusted analyses. Third, our method to assess patients’ knowledge of the care plan, though based on several prior research studies, has not been validated. ^4,6–8 Prior studies have shown that patients and their physicians tend to overestimate patients’ comprehension of their hospital care plan; therefore, patients’ self-report of overall knowledge is unlikely to be valid. ^6,9 By asking patients specific questions regarding their plan of care, we hoped to gain a more accurate measure of patients’ knowledge. Finally, we did not assess health literacy, which may affect knowledge, and did not offer the portal application in any language other than English.