Health Literacy, Education Levels, and Patient Portal Usage During Hospitalizations

Abstract

Patient portal adoption has rapidly increased, and portal usage has been associated with patients’ sociodemographics, health literacy, and education. Research on patient portals has primarily focused on the outpatient setting. We explored whether health literacy and education were associated with portal usage in an inpatient population. Among 60,159 admissions in 2012–2013, 23.3% of patients reported limited health literacy; 50.4% reported some post-secondary education; 34.4% were registered for the portal; and 23.4% of registered patients used the portal during hospitalization. Probability of registration and inpatient portal use increased with educational attainment. Health literacy was associated with registration but not inpatient use. Among admissions with inpatient use, educational attainment was associated with viewing health record data, and health literacy was associated use of appointment and health education tools. The inpatient setting may provide an opportunity to overcome barriers to patient portal adoption and reduce disparities in use of health information technologies.

Introduction

Patient portals are online tools allowing patients and their families to interact with health care systems, health care providers, and portions of the patient’s electronic health record^1–3. Over the last several decades, many health care organizations have adopted patient portals in response to consumer demands and regulatory pressures^4–17. Patient portals have been shown to increase patient satisfaction, promote adherence to preventative and treatment recommendations, and improve clinical outcomes in the management of chronic diseases such as diabetes, hypertension, and depression^{12,14,17–22}.

There have been well-documented associations among health literacy, education levels, and the use of technology, including patient portals. In a large outpatient study of patients with diabetes, individuals with limited health literacy were less likely to use electronic health portals, a difference that persisted after adjustment for internet access²³. Patients with a high school education or less were also less likely to use health portals²⁴. Indeed, when asked about their preferences, patients with low health literacy and those with lower educational attainment expressed a preference for self-management support to be delivered in person or by phone, rather than via internet²⁵. Complementary research shows related relationships, for example, between socioeconomic status or race and portal use^{26, 27}. However, not all have found a relationship between health literacy and the use of internet portals or mobile health tools^{28, 29}.

Nearly all patient portal research has been conducted in the outpatient setting in either primary care or medical specialty practices^{7, 9, 10}. The inpatient setting, however, may represent an opportunity to explore barriers to patient portal adoption and to promote portal use^{30, 31}. The Vanderbilt University Medical Center (VUMC) has a well-established patient portal, My Health at Vanderbilt (MHAV), which has been deployed throughout all adult and pediatric specialties¹⁶. MHAV was designed and promoted for outpatient use, and some of the patient portal policies, such as a delay in availability of test results or a two to three business day expectation for secure message responses, might discourage portal usage in the acute hospital setting¹⁶. However, in recent years, both clinicians and the portal development team have observed an increase in the number patients using MHAV not only while remote from the health care system, but also when admitted to the hospital. In addition to a widely adopted portal, VUMC routinely collects health literacy data on adult patients during hospital admission intake. Integrating these data, we examined whether health literacy and educational attainment were associated with patient portal registration and portal usage in the inpatient setting.

Methods

Admission data

This study was conducted at VUMC, a comprehensive, academic health care facility in middle Tennessee and a major referral center for the Southeastern United States. We collected data on all admissions to Vanderbilt University Hospital (VUH) between January 1, 2012 and December 31, 2013. The VUMC Institutional Review Board approved this study.

Data on patient sociodemographics and clinical characteristics for all admissions were extracted from VUMC’s clinical data warehouse. We extracted information on patient age (date of birth), gender, race, date of admission, date of discharge, length of stay, chief complaint, and admitting service for each admission. Patients with “unknown” or “declined” race responses were coded as “unknown.” Anesthesiology, ophthalmology, psychiatry, research, dermatology, radiology, and radiation oncology, each of which had <100 admissions during the study period, were grouped into an “other” category. An emergency medicine category included patients who were admitted to inpatient services, but never physically left the emergency department prior to discharge. Admissions to pediatric services reflected adult patients admitted to pediatric providers at the adult hospital, which sometimes occurs for selected cases such as adult patients with pediatric cancers.

Patient portal usage

MHAV was launched in 2005 and has been implemented throughout the clinical enterprise since 2008. MHAV provides patients with electronic access to components of their health record, supports secure messaging with health care providers, allows for appointment management, and delivers personalized health education information based on patients’ diagnoses. All VUMC patients are eligible for a MHAV account, and MHAV users may grant account access to their caregivers and/or family members as delegate users¹⁶. After implementation, MHAV was rolled out incrementally through the VUMC outpatient clinics. Physician champions encouraged provider use of MHAV, and technical support staff was available to patients, physicians, and staff as the portal was brought to individual clinics. Patients initially register for the portal online, with full access to electronic health record information being provided only after in-person user identity verification.

We extracted data on patient MHAV registration and use for all patients admitted to the hospital during the study period. Dates of MHAV registration were compared with dates of inpatient stay to determine whether patients had portal accounts during hospitalization. Patients were considered registered with a portal account during an admission if they were registered prior to discharge. From MHAV usage logs, we determined the functions accessed within the portal and user status (i.e., the patient or his/her delegate) during each admission. Admissions were classified as having inpatient portal use if any activity by the patient or his/her delegate occurred between the time of admission and the time of discharge. Portal activity during hospitalization was grouped into major functions: account management (e.g., assigning a delegate), appointment management, secure messaging with health care providers, viewing portions of the electronic health record (i.e., laboratory test results, radiology reports, clinical visit summaries, medication lists, immunization history), accessing health educational materials, and other (e.g., viewing website help).

Health literacy and educational attainment

Since 2010, nursing intake forms for adult admissions to VUH have included questions on educational attainment and health literacy³². Years of educational attainment was collected as a continuous variable in the electronic health records by nurses, and was later categorized as less than high school, completed high school, some post-secondary education, and a 4-year college degree or more education. Health literacy was assessed using the Brief Health Literacy Screen (BHLS), a 3-item survey tool designed to identify patients with limited health literacy, which has been validated in research settings and as administered by nurses in our medical center^33,34. The three questions of the BHLS survey ask patients: 1) “How confident are you filling out medical forms by yourself?”; 2) “How often do you have someone help you read hospital materials?”; and 3) “How often do you have problems learning about your medical condition because of difficulty understanding written information?”. Patients responded to each question using a 5-point Likert scale, with the question regarding confidence filling out forms reverse coded. Responses were summed, and higher scores indicate higher levels of literacy. We operationalized health literacy as a continuous score and a binary indicator of adequate versus limited health literacy. Patients with scores ≤9 were classified as having limited health literacy³².

Statistical analysis

We restricted our analysis dataset to admissions of adult patients (age ≥18 years) with known gender, documented educational attainment, complete BHLS responses reported during admission, and discharge date no later than March 31, 2014, which was the end of our follow-up period.

Unadjusted and adjusted logistic regression models examined the relationships between patient characteristics, including sociodemographics and health literacy, and 1) MHAV registration status, 2) inpatient use during admissions of registered patients, and 3) portal functions accessed during admissions with inpatient use. For the fully adjusted models, we included age, gender, race, length of stay, month of admission, and admitting service. Educational attainment and health literacy were both included in multivariable models. Continuous variables – BHLS score, educational attainment in years, age in years, length of stay in days, and month of admission – were fit with 4-knot restricted cubic splines. In all models, we applied the Huber-White method to correct standard errors for correlation among multiple admissions to the same patient. Our analyses were conducted in R 3.1.2.

Results

Between January 1, 2012 and December 31, 2013, there were 84,185 admissions to VUH. We excluded 7 admissions with discharge dates after our follow-up period, 638 admissions of patients with age <18 years, and 5 admissions for which patient gender was not available. Of the 83,535 remaining admissions, the BHLS was not completed in 21,903 admissions (26.2%) and educational attainment was missing in another 1,473 admissions, restricting our analysis to 60,159 admissions and to 41,176 unique patients. Whether health literacy and education data were record was associated with sociodemographics and clinical characteristics (see Table 1). Compared to patients with health literacy and education data, those excluded due to missing health literacy or education data were more likely to be younger, female, and Asian/Pacific Islander or unknown race (p<0.05). Patients without health literacy or education data also had shorter length of stays than patients with these data (p<0.05). Admissions to Obstetrics/Gynecology were least likely to have health literacy or education data available.

Table 1.

Sociodemographics and clinical characteristics of the study population by health literacy status.

Characteristic	Study Population by Health Literacy a						Excluded Admissions b
	All		Adequate		Limited
	n	%	n	%	n	%	n	%
	Total	60,159		46,139		14,020		23,376
Age at admission (years) c	41/55/66		39/53/65		46/60/73		27/36/59
Gender
Female	29,476	49.0	23,016	49.9	6,460	46.1	16,305	69.8
Male	30,683	51.0	23,123	50.1	7,560	53.9	7,071	30.2
Race
Asian/Pacific Islander	545	0.9	350	0.8	195	1.4	660	2.8
African American	9,314	15.5	7,049	15.3	2,265	16.2	3,547	15.2
Native American/Alaskan	183	0.3	109	0.2	74	0.5	103	0.4
Caucasian	48,769	81.1	37,679	81.7	11,090	79.1	17,778	76.1
Unknown/Declined	1,348	2.2	952	2.1	396	2.8	1,288	5.5
Educational attainment category
No high school	2,853	4.7	1,074	2.3	1,779	12.7	–
Some high school	5,336	8.9	3,290	7.1	2,046	14.6	–
High school/GED	21,673	36.0	16,051	34.8	5,622	40.1	–
Some post-high school	16,221	27.0	13,513	29.3	2,708	19.3	–
College degree or more	14,076	23.4	12,211	26.5	1,865	13.3	–
Year of admission
2012	29,611	49.2	23,032	49.9	6,579	46.9	11,330	48.5
2013	30,548	50.8	23,107	50.1	7,441	53.1	12,046	51.5
Length of stay (days)c	2/3/5		1/3/5		2/3/6		1/2/3
Admitting service category
Emergency Medicine	491	0.8	377	0.8	114	0.8	157	0.7
Medicine	30,934	51.4	23,286	50.5	7,648	54.6	5,278	22.6
Neurology	3,326	5.5	2,273	4.9	1,053	7.5	774	3.3
Obstetrics/Gynecology	1,119	1.9	984	2.1	135	1.0	10,101	43.2
Pediatrics	102	0.2	84	0.2	18	0.1	21	0.1
Surgery	24,087	40.0	19,059	41.3	5,028	35.9	7,011	30.0
Other	100	0.2	76	0.2	24	0.2	34
MHAV registered user	20,671	34.4	17,340	37.6	3,331	23.8	7,551	32.3
MHAV use during admission	4,829	8.0	3,956	8.6	873	6.2	930	4.0

^aLimited health literacy defined as score ≤9.

^bAdmissions excluded due to missing education or health literacy data.

^cmedian and interquartile range (25^th percentile / median / 75^th percentile) included for continuous variables.

A description of the study population is presented in Table 1. The study population was predominantly Caucasian (81.1%), with a median age at admission of 55 years (interquartile range 41 – 66 years). Half of all hospitalizations were admitted by Medicine (51.4%) and the median length of stay was three days (interquartile range 2 – 5 days). Patients were registered for MHAV during 34.4% of admissions and used MHAV during inpatient stay in 8.0% of all admissions. Among admissions involving patient registered for MHAV, inpatient portal use occurred in 23.4%.

Across all admissions, 23.3% of patients reported limited health literacy (Table 2), with reading hospital materials being the most challenging question. Half of admissions involved patients with some post-secondary education (50.4%), while 13.6% involved patients with less than a high school education. Literacy levels and educational attainment differed by demographics (p<0.05). Correlation between continuous BHLS scores and educational attainment in completed years was 0.32.

Table 2.

Health literacy and educational attainment by MHAV registration and inpatient use.

	All	Portal registration		Inpatient portal use
		Registered	Unregistered	User	Nonuser
N	60,159	20,671	39,488	4,829	15,842
Health literacy
BHLS score (mean, sd)	12.1 (3.3)	12.8 (3.0)	11.7 (3.4)	12.6 (3.1)	12.9 (3.0)
Limited literacy a (%)	23.3	16.1	27.1	18.1	15.5
Educational attainment
Completed years (mean, sd)	13.3 (3.0)	14.4 (2.8)	12.7 (3.0)	14.5 (3.0)	14.3 (2.8)
No high school	4.7	1.6	6.4	1.8	1.6
Some high school	8.9	3.6	11.6	3.6	3.6
High school/GED	36.0	27.2	40.6	27.3	27.2
Some post-high school	27.0	31.2	24.7	28.8	32.0
College degree or more	23.4	36.4	16.6	38.4	35.7

^aLimited health literacy defined as score ≤9.

Portal registration

Health literacy and educational attainment were higher for admissions of patients registered for MHAV compared to admissions of patients not registered for MHAV (Table 2). Two-thirds (67.6%) of registered patients reported some post-secondary education, while only 41.3% of unregistered patients reported similar educational attainment. In unadjusted and adjusted models, the probability of portal registration increased across the range of educational attainment (p<0.05; Figure 1). The largest increase in the probability of registration occurred between 10 and 15 years of education (roughly between high school and college degrees).

Figure 1.

Predicted probability of MHAV portal registration by educational attainment (left) and health literacy (right) with adjustment for sociodemographics and clinical characteristics.

Health literacy was limited in 16.1% and 27.1% of admissions of registered and unregistered patients, respectively. In both unadjusted and adjusted models for portal registration, patients with limited health literacy were less likely to be registered than patient with adequate health literacy (p<0.05). For BHLS scores above 10, the predicted probability of registration increased with BHLS score; a clear and significant pattern of association within lower ranges of health literacy was not observed (Figure 1).

Inpatient portal use

Among admissions of registered patients, admissions with and without inpatient use had similar levels of health literacy and educational attainment. For both registered patients using the portal during their inpatient stay and those registered patients not using the portal during their inpatient stay, two-thirds reported education beyond the high school level and approximately 5% reported never attending high school. After adjustment for sociodemographics, clinical characteristics, and health literacy, educational attainment remained significantly associated with inpatient portal use (p<0.05). For MHAV registered patients with approximately 15 or more years of education, the predicted probability of inpatient portal use increased with additional years of education, although confidence intervals became wide due to small sample sizes for the highest educational levels reported (Figure 2).

Figure 2.

Predicted probability of MHAV inpatient use by educational attainment (left) and health literacy (right) with adjustment for sociodemographics and clinical characteristics.

Limited health literacy was reported for 18.1% of admissions with inpatient portal use and 15.5% of admissions of registered without inpatient portal use; mean BHLS scores were not different between the two groups (12.6 for users and 12.9 for non-users). With adjustment for sociodemographics and clinical characteristics, we did not observe an association between health literacy and the inpatient portal use (Figure 2).

Portal functions accessed during inpatient stay

During the 4,829 admissions with inpatient portal use, patients viewed their health record information in 71.8% of admissions, used secure messaging in 57.2% of admissions, managed medical appointments in 21.0% of admissions, viewed health education materials in 20.1% of admissions, conducted portal account management activities in 4.5% of admissions, and used other functions in 8.0% of admissions. In adjusted models, educational attainment was associated with whether data from the electronic health record was viewed through the portal during admission (p<0.05). The predicted probability of viewing health record information through the portal steadily increased with years of education (Figure 3). We also observed, in adjusted models, associations between health literacy and whether appointment management tools and health education materials were accessed through the portal during admission (p<0.05). The predicted probability of using either function increased over the lower ranges of BHLS, peaking around scores of 10–11 and then declining somewhat (Figure 4).

Figure 3.

Among admissions with inpatient portal use, predicted probability of viewing health record data through the portal by educational attainment, after adjustment for sociodemographics and clinical characteristics.

Figure 4.

Among admissions with inpatient portal use, predicted probability of accessing appointment management tools and viewing health education materials through the portal by health literacy, after adjustment for sociodemographic and clinical characteristics.

Discussion

This study is one of the first to examine patient portal usage during hospitalizations and the relationships between patient characteristics, including sociodemographics, health literacy, and education. In a large population of adult patients with diverse health care needs and diagnoses, we found that both health literacy and educational attainment appeared to have a significant effect on the likelihood of being registered for a patient portal. Probability of registration and inpatient portal use increased with educational attainment. As educational attainment increased, portal users were more likely to access health record information within the patient portal. While health literacy was associated with registration, it was not associated with whether patients accessed the portal during admission. Among inpatient portal users, however, health literacy was associated with use of appointment management and targeted health education tools. Similar to our results, Mayberry et al did not observe an association between health literacy and portal use²⁸. In contrast, Sarkar et al observed lower levels of portal use among patients with limited health literacy²³. Both studies were conducted in outpatient populations limited to patients with diabetes.

There are several possible explanations for our findings. First, registration for the portal may be a formidable barrier that prohibits patients with limited levels of health literacy and educational attainment from becoming portal users. Prior research has also observed lower portal registration rates among patients with lower educational attainment and limited health literacy^{23, 24}. Once patients have overcome the registration barrier, education and health literacy may no longer affect portal usage. If this explanation were true, we expect to see similar trends in both the inpatient and outpatient settings. Studies of outpatient populations, however, report inconsistent findings regarding associations between portal use and literacy^{23, 28}. Alternatively, while admitted to the hospital, patients may receive encouragement and support in using the patient portal from hospital staff or family members, as well as contextual support such as continuous computer and internet access. This assistance may help patients overcome barriers resulting from limited health literacy or education. For example, during inpatient stay, all VUH patients may have access to the internet through computers available in patient rooms. Nurse may assist patients with use of in-room technologies, allowing patients to access the portal. If limited health literacy and education are associated with more limited computer or internet access in patients’ daily lives, then equity in computer access during an inpatient stay may have a clear connection to our results. For patients with limited access to technology in the outpatient setting, portal registration may be more difficult, or even limited to times during which patients interact with the health system, leading to associations such as those we observed between education, health literacy, and registration. During hospitalizations at VUMC, however, this technology access barrier is eliminated, which may explain the lack of an association we observed between health literacy and inpatient portal use. Whether registration serves as the prohibitive barrier to portal adoption or the presence of additional portal support during an inpatient stay is responsible for the observed associations, the inpatient setting may provide an opportunity to promote patient portals and reduce disparities in adoption of patient engagement technologies.

This study is strengthened by leveraging routinely-collected health literacy data obtained on a large cohort of adult patients; exploring adoption of a well-established patient portal with a large user community; and including patients treated by all clinical specialties rather than limiting to patients with particular diseases. However, the study is not without limitations. MHAV is a locally-developed patient portal deployed at a single institution and designed to support outpatient use cases. The findings for our portal may not necessarily translate to other portals with different registration processes and user interfaces, variations that may result in different barriers and associations between education, health literacy, and portal adoption. A number of other factors not examined in this study influence portal use, including socioeconomic status and whether or not the patient was an existing VUMC patient at admission. VUMC is a large referral center in the region, and thus patients who are transferred here might not have a portal account since they are unfamiliar with the medical center and most likely do not plan to return. Additionally, we were not able to ascertain how patients accessed the portal. During the study period, VUMC did not provide dedicated computers for patient use in patient rooms, and thus patients wanting to access the portal during hospitalization would have needed access to a personal web-enabled device. More recently, VUMC has implemented an interactive television system that provides patients with in-room web access to the MHAV portal. Follow-up studies will examine whether this increase in technology access is associated with increased inpatient portal use. Additionally, we did not account for severity of admitting diagnoses or affected systems, which may limit patient physical and mental ability to use the patient portal during a hospital stay.

While we can distinguish portal activity conducted through patient and delegate user accounts, we do not have health literacy or educational information for delegates. For the inpatient portal use models, we associated delegate activity during each admission with the admitted patient’s education or health literacy information, which may or may not be a reflection of the delegate’s traits. We observed delegate use in 4.2% of admissions with inpatient use, and 2.9% of admissions with inpatient use solely involved delegate activity and no portal access through the patient’s account. While the delegate account policy is designed to avoid multiple users accessing the patient account, we cannot be certain whether patients used the portal themselves or logged in to allow family members access to their health information, in which case the recorded education and health literacy information may not be representative of the individual interacting with the patient portal.

Additionally, almost one fourth of patients were missing health literacy or education data. Although this limited our sample size, the remaining cohort of patients was still large. As we continue to administer health literacy measures to patients, we will be able to add health literacy and education data for patients from previous admissions. Finally, these measures are beginning to be administered in the clinics at VUH, making clinic intake forms another valuable resource of health literacy and education data for patients.

Conclusion

Patient portals engage patients in their health care, increase patient self-efficacy, and improve the quality of care. While studies of portal adoption have focused on outpatient primary care settings, patient portals may provide valuable benefits to hospitalized patients. Leveraging data from a patient portal with a large population of registered users treated across inpatient clinical specialties and data from a health literacy assessment widely implemented throughout the adult inpatient setting, we sought to understand associations between patients’ characteristics, including sociodemographics and health literacy, and portal registration and inpatient portal use. In our diverse adult inpatient population, lower education and limited health literacy were independently associated with a reduced probability of being registered for a portal account. Among patients registered for the portal, however, health literacy was not associated with whether patients accessed the portal during a hospital stay. The inpatient setting may serve as an opportunity to promote patient portal adoption and increase patient confidence with portal tools, increasing patient engagement during admission and potentially maintaining this engagement as patients transition to outpatient care. A deeper understanding of how education and health literacy act as barriers to portal registration more so than portal use is necessary for designing and implementing successful patient portal promotion activities.