Patient portal use and hospital outcomes

Patient portal use and hospital outcomes Abstract Objectives To determine whether use of a patient portal during hospitalization is associated with improvement in hospital outcomes, 30-day readmissions, inpatient mortality, and 30-day mortality. Materials and Methods We performed a retrospective propensity score–matched study that included all adult patients admitted to Mayo Clinic Hospital in Jacksonville, Florida, from August 1, 2012, to July 31, 2014, who had signed up for a patient portal account prior to hospitalization (N = 7538). Results Out of the admitted patients with a portal account, 1566 (20.8%) accessed the portal while in the hospital. Compared to patients who did not access the portal, patients who accessed the portal were younger (58.8 years vs 62.3 years), had fewer elective admissions (54.2% vs 64.1%), were more frequently admitted to medical services (45.8% vs 35.2%), and were more likely to have liver disease (21.9% vs 12.9%) and higher disease severity scores (0.653 vs 0.456). After propensity score matching, there was no statistically significant difference between the 2 cohorts with respect to 30-day readmission (P = .13), inpatient mortality (P = .82), or 30-day mortality (P = .082). Conclusion Use of the patient portal in the inpatient setting may not improve hospital outcomes. Future research should examine the association of portal use with more immediate inpatient health outcomes such as patient experience, patient engagement, medication reconciliation, and prevention of adverse events. Patient portal, Health information technology, electronic health record, patient access, hospital outcomes BACKGROUND In 1996, the Health Insurance Portability and Accountability Act gave patients the legal right to view and own their medical records.1 Unrestricted patient access to medical records was somewhat difficult in the era of paper charts, but after the introduction of the electronic health record (EHR) and, subsequently, EHR-tethered personal health record electronic entry, or “patient portals,” patients’ access to their records became less challenging. In 2009, the Health Information Technology for Economic and Clinical Health Act published meaningful use rules for using the EHR,2 which included that the EHR should be used to provide effective health information exchange to improve health care quality and patient-centered care. In light of these rules, it is expected that patient portal use will be more associated with medical outcomes.3 Over the past several years, patient portal use has increased in both the outpatient and inpatient settings.4,5 Online, patients can review labs, provider notes, and medication and problem lists, request prescription renewals, manage appointments, and communicate with their physicians. While patient portal use is increasing,6 the inpatient environment presents additional challenges for patients in receiving information about their care, which may affect use. The clinical conditions leading to hospitalization are often acute, at times with devastating consequences. The amount of medical information generated during a hospitalization can be extensive and delivered to the patient rapidly over a short period of time. Consequently, patients may have difficulty comprehending and integrating this information, as well as understanding the plan of care.7 Furthermore, it is not uncommon for care teams to be new to patients and their families, further complicating communication. It is known that the majority of patients are unable to identify their primary inpatient providers.8 Hospitalized patients’ immediate access to the online portal is an additional challenge, since devices to access the Internet are not always available in the hospital. The advent of Internet-connected smartphones and tablets has facilitated online access.9 Despite the barriers, patients want to receive and review health information electronically during their hospitalization and after discharge.9–13 There is limited information regarding patient portal use in the inpatient setting and its association with clinical outcomes.14 Most of the information regarding inpatient portal use is obtained from survey-based quasi-experimental studies of limited numbers of patients.14 These studies report usability and satisfaction outcomes: patients’ intention to use the portal,9 inpatient portal utilization,15 and satisfaction with use of the portal tool.13 Offering hospitalized patients access to a web portal via tablet computers increased their ability to correctly name more than one physician and their roles.15 While these studies offer an indication that immediate health outcomes may be positive, to date no study has examined the association between hospital outcomes of 30-day readmissions, inpatient mortality, and 30-day mortality and patient portal use. Patients who use the portal to access their medical records while hospitalized may have an active interest in their outcome and information about their hospital stay. We pose the question of whether this engagement would reduce 30-day readmissions or decrease mortality. We asked two main questions in this study: How do patients who access their portal accounts in the inpatient setting differ from those with access who do not? Is there a relationship between inpatient portal use and hospital outcomes of 30-day readmission, 30-day mortality, and inpatient mortality? To our knowledge, this is the first study to examine hospital outcomes associated with inpatient portal use. MATERIALS AND METHODS Study design and subjects This retrospective study included all adult patients admitted to Mayo Clinic Hospital in Jacksonville, Florida, from August 1, 2012, to July 31, 2014, who had a patient portal account prior to hospitalization. If a patient had more than one inpatient admission during this time frame, only the demographic and clinical data from the first encounter was collected and analyzed. Patients were excluded if they were under 18 years of age at the time of admission. Hospital observation stays were not included. The study was approved by the Mayo Clinic Institutional Review Board. Data collection The EHR was used to obtain information on patient demographics (age at admission, sex, race, ethnicity, marital status, primary language, payor information, and employment status) and clinical information known at the time of admission (admission status, admission service, and comorbidities), and hospital outcomes (30-day readmission, 30-day mortality, and inpatient mortality). Principal and secondary diagnoses and procedures, comorbidities, and All-Patient Refined Diagnosis Related Group (APR-DRG) weight16 were collected based on International Classification of Diseases, Ninth Revision, Clinical Modification codes extracted from hospital discharge abstracts. Hospital length of stay (LOS) was calculated using the dates and times of hospital admission and hospital discharge. Patient Online Services (patient portal) The Florida campus of Mayo Clinic implemented the Patient Online Services (patient portal) in 2010, initially providing patients with secure Internet access to their laboratory data. On March 1, 2012, access to outpatient physician notes was added. In May 2011, Mayo Clinic launched the Mayo Clinic Mobile app, which allows patients the same access as the web portal. This app was compatible with Apple iOS initially, and then expanded to Android as well. The app gives patients the flexibility to view their information on the go, to sync appointments to electronic calendars, and to set appointment reminders. The patient portal is tethered to Mayo Clinic Florida’s EHR, PowerChart (Cerner Corporation, North Kansas City, MO, USA). Information is extracted from PowerChart and displayed by Mayo’s proprietary online and mobile-based applications. The date and time that patients access the portal can be determined; however, the content or modules accessed cannot, due to a system limitation with PowerChart. Patients routinely create portal accounts in the outpatient setting. Once an account is created, the portal gives a patient the opportunity to view lab results, clinic progress notes, and medication and allergy lists, and to send messages to care providers. When a patient who signed up for an account is hospitalized, the patient can then use the portal to access inpatient labs, inpatient admission notes, consultation reports, operative notes, and discharge summaries in real time. Radiology and pathology reports are available with a 72-h delay. Providers’ daily progress notes are not viewable. While patient-to-provider electronic messaging is encouraged in the outpatient setting, communication between hospitalized patients and hospital care teams is not available. Statistical analysis Preadmission characteristics were summarized according to portal use during hospitalization among the cohort of patients who had a portal account prior to admission. We aimed to estimate the effect of portal use during hospitalization on hospital outcomes (inpatient mortality, 30-day readmission, and 30-day mortality). In an attempt to control confounding, propensity score matching was used to identify a cohort of patients with similar baseline characteristics. Propensity score is defined here as the conditional probability of an inpatient using his or her existing portal account given a set of covariates known at the time of admission (baseline). A multivariable logistic regression model with inpatient portal use as the dependent variable and all the baseline characteristics displayed in Table 1 as covariates was used to estimate propensity score. Up to 3 non–portal users17 for each portal user were selected utilizing a greedy matching algorithm without replacement18 with a caliper width equal to 0.2 of the standard deviation (SD) of the logit of the propensity score.19,20 To assess potential imbalance in baseline characteristics between the 2 groups, standardized differences were estimated before and after matching. A standardized difference of <10% for a given baseline characteristic was considered a negligible imbalance between groups.20 In the matched cohort of non–portal users, summary statistics were weighted such that each patient was assigned a weight of 1 divided by the number of non–portal users matched to the corresponding portal user for the given analysis. Table 1. Preadmission characteristics of the full cohort of patients who had a portal account established prior to hospital admission according to portal use during the inpatient stay Characteristic  No portal use (n = 5972)  Portal use (n = 1566)  Standardized difference before matching, %  Mean age, years  62.3 ± 15.1  58.8 ± 15.7  22.7  Female sex (%)  3121 (52.3)  739 (47.2)  10.2  Race (%)         Black  361 (6.0)  53 (3.4)  12.6   White  5374 (90.0)  1428 (91.2)  4.1   Other  144 (2.4)  71 (4.5)  11.6   Unknown  93 (1.6)  14 (0.9)  6.0  Ethnic group (%)         Hispanic or Latino  202 (3.4)  81 (5.2)  8.9   Not Hispanic or Latino  5625 (94.2)  1457 (93.0)  4.7   Unknown  145 (2.4)  28 (1.8)  4.5  Marital status (%)         Married or life partner  4371 (73.2)  1194 (76.2)  7.0   Divorced or separated  502 (8.4)  103 (6.6)  7.0   Widowed  491 (8.2)  95 (6.1)  8.4   Single  603 (10.1)  171 (10.9)  2.7   Unknown  5 (0.1)  3 (0.2)  2.9  Language (%)         English  5857 (98.1)  1504 (96.0)  12.1   Spanish  49 (0.8)  29 (1.9)  9.0   Other  31 (0.5)  19 (1.2)  7.5   Unknown  35 (0.6)  14 (0.9)  3.6  Government payor (%)  3315 (55.5)  723 (46.2)  18.8  Employment status (%)         Employed  1730 (29.0)  488 (31.2)  4.8   Not employed  663 (11.1)  249 (15.9)  14.1   Retired  2040 (34.2)  489 (31.2)  6.3   Disabled  351 (5.9)  107 (6.8)  3.9   Unknown  1188 (19.9)  233 (14.9)  13.3  Admission status (%)         Routine/elective  3831 (64.1)  848 (54.2)  20.4   Semi-urgent  161 (2.7)  80 (5.1)  12.5   Urgent  1718 (28.8)  564 (36.0)  15.5   Emergency  239 (4.0)  61 (3.9)  0.5   Other  23 (0.4)  13 (0.8)  5.7  Admission service (%)         Critical  244 (4.1)  85 (5.4)  6.3   Medical  2104 (35.2)  717 (45.8)  21.6   Surgical  3624 (60.7)  764 (48.8)  24.1  Myocardial infarct (%)  291 (4.9)  87 (5.6)  3.1  Congestive heart failure (%)  541 (9.1)  147 (9.4)  1.1  Peripheral vascular disease (%)  1045 (17.5)  262 (16.7)  2.0  Cerebrovascular disease (%)  718 (12.0)  162 (10.3)  5.3  Dementia (%)  253 (4.2)  67 (4.3)  0.2  Chronic pulmonary disease (%)  851 (14.2)  196 (12.5)  5.1  Ulcer (%)  150 (2.5)  47 (3.0)  3.0  Mild liver disease (%)  766 (12.8)  335 (21.4)  22.9  Diabetes (%)  1050 (17.6)  285 (18.2)  1.6  Diabetes with organ damage (%)  130 (2.2)  39 (2.5)  2.1  Hemiplegia (%)  78 (1.3)  22 (1.4)  0.9  Moderate/severe renal disease (%)  690 (11.6)  196 (12.5)  3.0  Moderate/severe liver disease (%)  273 (4.6)  168 (10.7)  23.3  Metastatic solid tumor (%)  473 (7.9)  178 (11.4)  11.7  AIDS (%)  8 (0.1)  4 (0.3)  2.8  Rheumatologic disease (%)  208 (3.5)  52 (3.3)  0.9  Other cancer (%)  1743 (29.2)  569 (36.3)  15.3  Mean APR-DRG weight, log scale  0.456 ± 0.725  0.653 ± 0.859  24.7  Characteristic  No portal use (n = 5972)  Portal use (n = 1566)  Standardized difference before matching, %  Mean age, years  62.3 ± 15.1  58.8 ± 15.7  22.7  Female sex (%)  3121 (52.3)  739 (47.2)  10.2  Race (%)         Black  361 (6.0)  53 (3.4)  12.6   White  5374 (90.0)  1428 (91.2)  4.1   Other  144 (2.4)  71 (4.5)  11.6   Unknown  93 (1.6)  14 (0.9)  6.0  Ethnic group (%)         Hispanic or Latino  202 (3.4)  81 (5.2)  8.9   Not Hispanic or Latino  5625 (94.2)  1457 (93.0)  4.7   Unknown  145 (2.4)  28 (1.8)  4.5  Marital status (%)         Married or life partner  4371 (73.2)  1194 (76.2)  7.0   Divorced or separated  502 (8.4)  103 (6.6)  7.0   Widowed  491 (8.2)  95 (6.1)  8.4   Single  603 (10.1)  171 (10.9)  2.7   Unknown  5 (0.1)  3 (0.2)  2.9  Language (%)         English  5857 (98.1)  1504 (96.0)  12.1   Spanish  49 (0.8)  29 (1.9)  9.0   Other  31 (0.5)  19 (1.2)  7.5   Unknown  35 (0.6)  14 (0.9)  3.6  Government payor (%)  3315 (55.5)  723 (46.2)  18.8  Employment status (%)         Employed  1730 (29.0)  488 (31.2)  4.8   Not employed  663 (11.1)  249 (15.9)  14.1   Retired  2040 (34.2)  489 (31.2)  6.3   Disabled  351 (5.9)  107 (6.8)  3.9   Unknown  1188 (19.9)  233 (14.9)  13.3  Admission status (%)         Routine/elective  3831 (64.1)  848 (54.2)  20.4   Semi-urgent  161 (2.7)  80 (5.1)  12.5   Urgent  1718 (28.8)  564 (36.0)  15.5   Emergency  239 (4.0)  61 (3.9)  0.5   Other  23 (0.4)  13 (0.8)  5.7  Admission service (%)         Critical  244 (4.1)  85 (5.4)  6.3   Medical  2104 (35.2)  717 (45.8)  21.6   Surgical  3624 (60.7)  764 (48.8)  24.1  Myocardial infarct (%)  291 (4.9)  87 (5.6)  3.1  Congestive heart failure (%)  541 (9.1)  147 (9.4)  1.1  Peripheral vascular disease (%)  1045 (17.5)  262 (16.7)  2.0  Cerebrovascular disease (%)  718 (12.0)  162 (10.3)  5.3  Dementia (%)  253 (4.2)  67 (4.3)  0.2  Chronic pulmonary disease (%)  851 (14.2)  196 (12.5)  5.1  Ulcer (%)  150 (2.5)  47 (3.0)  3.0  Mild liver disease (%)  766 (12.8)  335 (21.4)  22.9  Diabetes (%)  1050 (17.6)  285 (18.2)  1.6  Diabetes with organ damage (%)  130 (2.2)  39 (2.5)  2.1  Hemiplegia (%)  78 (1.3)  22 (1.4)  0.9  Moderate/severe renal disease (%)  690 (11.6)  196 (12.5)  3.0  Moderate/severe liver disease (%)  273 (4.6)  168 (10.7)  23.3  Metastatic solid tumor (%)  473 (7.9)  178 (11.4)  11.7  AIDS (%)  8 (0.1)  4 (0.3)  2.8  Rheumatologic disease (%)  208 (3.5)  52 (3.3)  0.9  Other cancer (%)  1743 (29.2)  569 (36.3)  15.3  Mean APR-DRG weight, log scale  0.456 ± 0.725  0.653 ± 0.859  24.7  Categorical characteristics are given as the number and percentage of patients, while numerical characteristics are given as mean ± SD. The sample mean of APR-DRG weight (0.497, log scale) was imputed for 14 patients (12 with no portal use and 2 with portal use) who were missing that information. Abbreviations: APR-DRG, All-Patient Refined Diagnosis Related Group. Conditional logistic regression was used to evaluate the impact of portal use on inpatient mortality, 30-day readmission, and 30-day mortality. All models included a variable representing the match identifier as a stratification factor in order to preserve the benefit of matching. Models evaluating 30-day outcomes were adjusted for LOS (log scale). Patients were excluded from analyses evaluating 30-day outcomes if they died during their hospital stay or if they did not have a corresponding match after patients who died during their hospital stay were excluded. Statistical analyses were performed using SAS statistical software (version 9.4, SAS Institute Inc., Cary, NC, USA), and all reported P-values are 2-sided without adjustment for multiple testing. RESULTS We identified 17 050 patients in the study period: 7538 (44.2%) had a portal account established at the time of admission and 9512 (55.8%) did not. Of the patients who already had portal accounts, 1566 (20.8%) accessed the portal during hospitalization. Patients’ preadmission characteristics before matching are summarized in Table 1 and show substantial differences (standardized difference >20%) among those who accessed the portal by age, elective admission status, admission to medical or surgical services, comorbidity of liver disease, and disease severity (APR-DRG weight). Patients who accessed their portal account were more likely to be younger, to have more elective vs urgent admissions, to be admitted to medical services versus surgical services, and to have liver disease, and had higher disease severity scores than patients who did not access the portal during their stay. We included 7538 patients with a portal account at the time of admission in the logistic regression model to estimate the propensity score for using the portal account during hospitalization. The c index from that model was 0.67. The mean propensity score among the patients who accessed their portal account during hospitalization was 0.260 (SD = 0.125), compared with 0.194 (SD = 0.094) for patients who did not access their portal account during hospitalization. As seen in Table 2, all differences in preadmission patient characteristics between those who did and did not use the portal during hospitalization were considered negligible (all standardized differences ≤2.0%) after propensity score matching. Table 2. Preadmission characteristics after propensity score matching Characteristic  No in-hospital portal use (n = 3922)  In-hospital portal use (n = 1559)  Standardized difference after matching, %  Age, years  58.7 ± 10.0  58.9 ± 15.7  1.5  Female sex  47.4  47.2  0.3  Race         Black  3.3  3.4  0.4   White  91.5  91.4  0.2   Other  4.2  4.3  0.5   Unknown  1.0  0.9  1.1  Ethnic group         Hispanic or Latino  4.9  5.1  0.6   Not Hispanic or Latino  93.2  93.1  0.4   Unknown  1.8  1.8  0.2  Marital status         Married or life partner  75.6  76.3  1.7   Divorced or separated  6.7  6.6  0.4   Widowed  6.1  6.1  0.0   Single  11.5  10.8  2.0   Unknown  0.2  0.2  0.2  Language         English  96.3  96.2  0.6   Spanish  1.8  1.8  0.0   Other  1.2  1.2  0.1   Unknown  0.8  0.9  1.4  Government payor  47.0  46.3  1.5  Employment status         Employed  31.0  31.2  0.6   Not employed  15.8  15.7  0.5   Retired  31.1  31.3  0.4   Disabled  7.3  6.9  1.5   Unknown  14.8  14.9  0.3  Admission status         Routine/elective  53.7  54.2  0.9   Semi-urgent  5.0  5.0  0.0   Urgent  36.7  36.0  1.3   Emergency  3.8  3.9  0.5   Other  0.7  0.8  1.0  Admission service         Critical  5.6  5.3  1.2   Medical  45.6  45.7  0.2   Surgical  48.8  49.0  0.3  Myocardial infarct  5.8  5.6  0.8  Congestive heart failure  9.7  9.4  0.9  Peripheral vascular disease  16.5  16.8  0.8  Cerebrovascular disease  10.2  10.4  0.7  Dementia  4.4  4.3  0.4  Chronic pulmonary disease  12.5  12.6  0.2  Ulcer  2.8  3.0  1.1  Mild liver disease  21.0  21.0  0.0  Diabetes  18.5  18.2  0.7  Diabetes with organ damage  2.6  2.5  0.9  Hemiplegia  1.5  1.4  0.9  Moderate/severe renal disease  12.6  12.6  0.1  Moderate/severe liver disease  10.2  10.5  1.1  Metastatic solid tumor  11.5  11.2  0.9  AIDS  0.2  0.3  1.4  Rheumatologic disease  3.2  3.3  0.5  Other cancer  36.3  36.2  0.2  APR-DRG weight, log scale  0.639 ± 0.555  0.645 ± 0.852  0.8  Characteristic  No in-hospital portal use (n = 3922)  In-hospital portal use (n = 1559)  Standardized difference after matching, %  Age, years  58.7 ± 10.0  58.9 ± 15.7  1.5  Female sex  47.4  47.2  0.3  Race         Black  3.3  3.4  0.4   White  91.5  91.4  0.2   Other  4.2  4.3  0.5   Unknown  1.0  0.9  1.1  Ethnic group         Hispanic or Latino  4.9  5.1  0.6   Not Hispanic or Latino  93.2  93.1  0.4   Unknown  1.8  1.8  0.2  Marital status         Married or life partner  75.6  76.3  1.7   Divorced or separated  6.7  6.6  0.4   Widowed  6.1  6.1  0.0   Single  11.5  10.8  2.0   Unknown  0.2  0.2  0.2  Language         English  96.3  96.2  0.6   Spanish  1.8  1.8  0.0   Other  1.2  1.2  0.1   Unknown  0.8  0.9  1.4  Government payor  47.0  46.3  1.5  Employment status         Employed  31.0  31.2  0.6   Not employed  15.8  15.7  0.5   Retired  31.1  31.3  0.4   Disabled  7.3  6.9  1.5   Unknown  14.8  14.9  0.3  Admission status         Routine/elective  53.7  54.2  0.9   Semi-urgent  5.0  5.0  0.0   Urgent  36.7  36.0  1.3   Emergency  3.8  3.9  0.5   Other  0.7  0.8  1.0  Admission service         Critical  5.6  5.3  1.2   Medical  45.6  45.7  0.2   Surgical  48.8  49.0  0.3  Myocardial infarct  5.8  5.6  0.8  Congestive heart failure  9.7  9.4  0.9  Peripheral vascular disease  16.5  16.8  0.8  Cerebrovascular disease  10.2  10.4  0.7  Dementia  4.4  4.3  0.4  Chronic pulmonary disease  12.5  12.6  0.2  Ulcer  2.8  3.0  1.1  Mild liver disease  21.0  21.0  0.0  Diabetes  18.5  18.2  0.7  Diabetes with organ damage  2.6  2.5  0.9  Hemiplegia  1.5  1.4  0.9  Moderate/severe renal disease  12.6  12.6  0.1  Moderate/severe liver disease  10.2  10.5  1.1  Metastatic solid tumor  11.5  11.2  0.9  AIDS  0.2  0.3  1.4  Rheumatologic disease  3.2  3.3  0.5  Other cancer  36.3  36.2  0.2  APR-DRG weight, log scale  0.639 ± 0.555  0.645 ± 0.852  0.8  Categorical characteristics are given as the percentage of patients, while numerical characteristics are given as mean ± SD. Summary statistics in the group with no in-hospital portal use are weighted such that each patient was assigned a weight of 1 divided by the number of patients with no in-hospital portal use matched to the corresponding portal user. Abbreviations: APR-DRG, All-Patient Refined Diagnosis Related Group. All but 7 patients who used the portal were matched to one or more patients who did not use the portal during their hospital stay. Matching retained 72.7% of the original cohort. The mean propensity score in the matched groups was 0.258 in the group that used the portal (SD = 0.120, n = 1559) and 0.227 in the group that did not use the portal (SD = 0.098, n = 3922). Associations of portal use with hospital outcomes in the propensity-matched cohort are shown in Table 3. There was no significant association of portal use with inpatient mortality (P = .82), and, after adjusting for LOS, no statistically significant association with 30-day readmission (P = .13) or 30-day mortality (P = .087). Table 3. Association of patient portal use with hospital outcomes in the propensity-matched cohort Outcome  No.  OR (95% CI)  P-value  Inpatient mortality  5481  0.94 (0.55, 1.61)  .82  30-day readmission  5374  1.16 (0.96, 1.39)  .13  30-day mortality  5374  1.37 (0.96, 1.97)  .087  Outcome  No.  OR (95% CI)  P-value  Inpatient mortality  5481  0.94 (0.55, 1.61)  .82  30-day readmission  5374  1.16 (0.96, 1.39)  .13  30-day mortality  5374  1.37 (0.96, 1.97)  .087  Odds ratios (ORs) and 95% confidence intervals (CIs) result from conditional logistic regression models with in-hospital patient portal use as a predictor variable. Propensity match identifier was included in each model as a stratification factor to preserve the benefit of matching. Models of 30-day outcomes include length of hospital stay on the logarithm scale as a covariate. Patients were excluded from the 30-day outcomes analysis (n = 107) if they died in the hospital (n = 65: 44 patients with no in-hospital portal use and 21 patients with in-hospital portal use) or if they did not have a corresponding match after patients who died in the hospital were excluded (n = 42: 34 patients with no in-hospital portal use and 8 patients with in-hospital portal). Among the 5374 patients included in the 30-day outcomes analyses, 609 patients were readmitted within 30 days after hospital discharge (403 patients with no in-hospital portal use and 206 patients with in-hospital portal use) and 142 patients died within 30 days after hospital discharge (83 patients with no in-hospital portal use and 59 patients with in-hospital portal use). DISCUSSION The primary objective of this study was to compare patients with portal accounts who accessed and did not access the portal during hospitalization and to determine if there was an association between inpatient portal use and 30-day readmissions and mortality. In this study, patients who accessed their portal account were younger and had greater disease severity and more urgent admissions. It is possible that patients (or their families) access the portal when they are uncertain of the hospital course. This would account for lower use among surgical patients (large portion of elective surgery) and those with shorter stays. After propensity matching, there were no differences in 30-day readmissions, inpatient mortality, or 30-day mortality after adjusting for LOS. Of the 44.2% of patients who had a portal account at the time of admission, only 20.8% accessed the portal while hospitalized. There are several possible reasons for low inpatient portal use. Patient education regarding portal access and the potential benefits of portal use was not provided. The inpatient portal lacks specific features that allow communication between patients and care teams. Only admission notes, operative notes, consultations, and laboratory studies are available in real time. Daily progress notes cannot be viewed, and there is a 72-h delay in the viewing of radiology and pathology reports. Finally, the inpatient portal does not provide education regarding patient-specific disease processes. Other studies similarly report low inpatient use. Davis noted that 34.4% of admitted patients were registered for the portal, but only 23.4% used it during hospitalization.21 Among surgical patients admitted to a tertiary-care hospital, 25.3% had a portal account but only 16% of registered users accessed their account.22 The lack of features designed specifically for inpatient use was previously emphasized in a systematic review.14 Consequently, several medical centers designed hospital-specific applications aimed at improving the use and usability of inpatient portals.23–25 In a realistic review, Roberts indicated that patient participation with inpatient health information technology (including patient portals) can be augmented by interactive learning focused on information sharing, self-assessment and feedback, tailored education, user-centered design, and user support.26 Studies show that in the outpatient setting, patients with severe disease use portals more frequently,27,28 and patients who access portals have better outcomes in certain chronic medical conditions such as diabetes (lower HbA1c at 6 months),29 hypertension (improved blood pressure control at 12 months),30 depression management (increased medication adherence),31 and preventative care (up-to-date immunizations and mammograms).32,33 Our study indicates that unprompted and unguided patient portal use does not have any benefit with respect to 30-day readmissions and mortality. There are several possible explanations for this. First, use of a patient portal may play a more important role in the longitudinal management of chronic medical issues than in the outcomes associated with acute management of a decompensating clinical situation. While patients and their families can review clinical information via patient portals, this passive, unguided review of clinical data may have little bearing on the overall hospital and immediate posthospital course. In a systematic review, Goldzweig et al.34 noted that positive outcomes for patients with chronic diseases such as diabetes, hypertension, and depression were obtained only when portal use was enhanced with case management intervention, and the cohorts that showed clinical benefit received tailored education about portal use. Our patients were not specifically trained to use the portal, nor did they receive targeted intervention such as case management via the portal. Secondly, in the outpatient setting, outcome improvement was noted mostly for chronic diseases where the disease course is influenced by self-monitoring and behavioral modifications (diabetes, hypertension, asthma, HIV, fertility management, glaucoma, and hyperlipidemia).12 Due to the acute nature of inpatient conditions, behavioral modification is less feasible in the short time associated with hospitalization. Use of the patient portal is strongly encouraged by the Centers for Medicare and Medicaid Services, presumably to better inform patients and improve outcomes.35 Our study shows that clinical outcomes are not improved. Our study has a number of limitations. The population cohorts were selected from patients admitted to a single academic tertiary-care referral center, which precludes generalizability of our findings. Education level, which could affect portal use, is not routinely recorded in the EHR, so we could not use this information. Another limitation is that only certain information can be accessed by patients in real time, while there is a 72-h delay with pathology reports and radiology testing. The effect of patient portal use may be different in hospitals that embrace the “open notes” concept, allowing patients immediate access to provider notes and test results. Due to a system limitation, we could not verify the type of information accessed by patients in the portal. Also, we did not control the device type (PC, tablet, phone) used to access the patient portal. Free Internet connection is ubiquitous in the study institution, and the patient portal has iterations that can be accessed on different platforms under multiple operating systems and via web browsing. Mobile device availability and patient familiarity with the mobile version of the patient portal were not evaluated. Although the information in the patient portal is private, some hospitalized patients may have shared their log-in information with their family, friends, or primary care physician, and we did not account for the patient’s level of alertness at the time of portal access, nor did we evaluate who actually accessed the information. Prior inpatient or ambulatory use may be associated with inpatient portal use, and this was not assessed. Finally, there may be immediate outcomes that we did not directly examine where use of an inpatient portal may be of benefit, such as medication reconciliation, prevention of adverse events, patient satisfaction, and patient engagement. CONCLUSION This is the first study to examine the association between 30-day readmissions, 30-day mortality, and inpatient mortality and inpatient portal use. We found that there was no difference in 30-day readmissions, inpatient mortality, or 30-day mortality between portal and non–portal users and that inpatient portal use was low. The first step in determining whether patient portal use can improve hospital outcomes is to increase adoption and use by designing inpatient-specific portal tools that can engage patients and make them active participants in their health care. Future research should examine the role of portal use on more immediate outcomes such as patient experience and engagement to help clarify how patient portals can be used to improve outcomes. FUNDING This work was supported by the Robert D and Patricia E Kern Center for the Science of Health Care Delivery, Mayo Clinic, Jacksonville, Florida. COMPETING INTERESTS The authors have no competing interests to declare. CONTRIBUTORS AGD, MCB, and NLD conceived and designed the study. AGD, LMN, and CST collected the data. CST and JMN analyzed the data, designed the statistical method, and drafted the tables. AGD, MCB, NLD, and CST drafted the article. HEG, DIA, and JMN critically revised the article for important intellectual content. All co-authors contributed to writing, proofreading, and editing the manuscript. ACKNOWLEDGMENTS Launia J. White for her help with data extraction. REFERENCES 1 Health Insurance Portability and Accountability Act of 1996: Public Law 104–191—Aug. 21, 1996. US Government Publishing Office. www.gpo.gov/fdsys/pkg/PLAW-104publ191/pdf/PLAW-104publ191.pdf. Accessed March 18, 2014. 2 Title XIII: Health Information Technology (HITECH Act). www.healthit.gov/sites/default/files/hitech_act_excerpt_from_arra_with_index.pdf. Accessed March 18, 2014. 3 Kruse CS, Bolton K, Freriks G. The effect of patient portals on quality outcomes and its implications to meaningful use: a systematic review. J Med Internet Res.  2015; 17 2: e44. Google Scholar CrossRef Search ADS PubMed  4 Patel V, Barker W, Siminerio E. ONC Data Brief, No. 30: Trends in Consumer Access and Use of Electronic Health Information . Washington, DC: Office of the National Coordinator for Health Information Technology; 2015. www.healthit.gov/sites/default/files/briefs/oncdatabrief30_accesstrends_.pdf. Accessed November 1, 2016. 5 Charles D, Gabriel M, Henry J. ONC Data Brief 29: Electronic Capabilities for Patient Engagement among U.S. Non-Federal Acute Care Hospitals: 2012–2014 . Washington, DC: Office of the National Coordinator for Health Informaiton Technology; 2015. dashboard.healthit.gov/evaluations/data-briefs/hospitals-patient-engagement-electronic-capabilities.php. Accessed November 1, 2016. 6 Ammenwerth E, Schnell-Inderst P, Hoerbst A. The impact of electronic patient portals on patient care: a systematic review of controlled trials. J Med Internet Res.  2012; 14 6: e162. Google Scholar CrossRef Search ADS PubMed  7 O’Leary KJ, Kulkarni N, Landler MPet al.  , Hospitalized patients’ understanding of their plan of care. Mayo Clin Proc.  2010; 85 1: 47– 52. Google Scholar CrossRef Search ADS PubMed  8 Olson DP, Windish DM. Communication discrepancies between physicians and hospitalized patients. Arch Intern Med.  2010; 170 15: 1302– 07. Google Scholar CrossRef Search ADS PubMed  9 O’Leary KJ, Balabanova A, Patyk Met al.  , Medical inpatients’ use of information technology: characterizing the potential to share information electronically. J Healthc Qual.  2015; 37 4: 207– 20. Google Scholar CrossRef Search ADS PubMed  10 Burke RP, Rossi AF, Wilner BRet al.  , Transforming patient and family access to medical information: utilisation patterns of a patient-accessible electronic health record. Cardiol Young.  2010; 20 5: 477– 84. Google Scholar CrossRef Search ADS PubMed  11 Sprague J, Pell J, Lin CT. Divergent care team opinions about online release of test results to an ICU patient. J Participat Med.  2013; 5: e24. 12 Price M, Bellwood P, Kitson Net al.  , Conditions potentially sensitive to a personal health record (PHR) intervention, a systematic review. BMC Med Inform Decis Mak.  2015; 15: 32. Google Scholar CrossRef Search ADS PubMed  13 Greysen SR, Khanna RR, Jacolbia Ret al.  , Tablet computers for hospitalized patients: a pilot study to improve inpatient engagement. J Hosp Med.  2014; 9 6: 396– 69. Google Scholar CrossRef Search ADS PubMed  14 Prey JE, Woollen J, Wilcox Let al.  , Patient engagement in the inpatient setting: a systematic review. J Am Med Inform Assoc.  2014; 21 4: 742– 50. Google Scholar CrossRef Search ADS PubMed  15 O’Leary KJ, Lohman ME, Culver Eet al.  , The effect of tablet computers with a mobile patient portal application on hospitalized patients’ knowledge and activation. J Am Med Inform Assoc.  2016; 23 1: 159– 65. Google Scholar CrossRef Search ADS PubMed  16 Averill RF, Goldfield N, Bonazelli Jet al.  , All Patient Refined Diagnosis Related Groups (APR-DRGs), Version 20.0: Methodology Overview . Wallingford, CT: 3M Health Information Systems; 2003. 17 Ming K, Rosenbaum PR. Substantial gains in bias reduction from matching with a variable number of controls. Biometrics.  2000; 56 1: 118– 24. Google Scholar CrossRef Search ADS PubMed  18 Austin PC. Statistical criteria for selecting the optimal number of untreated subjects matched to each treated subject when using many-to-one matching on the propensity score. Am J Epidemiol.  2010; 172 9: 1092– 97. Google Scholar CrossRef Search ADS PubMed  19 Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat.  2011; 10 2: 150– 61. Google Scholar CrossRef Search ADS PubMed  20 Faries D, Leon AC, Haro JMet al.  , Analysis of Observational Health Care Data Using SAS® . Cary, NC: SAS Institute; 2010. 21 Davis SE, Osborn CY, Kripalani Set al.  , Health literacy, education levels, and patient portal usage during hospitalizations. AMIA Annu Symp Proc.  2015; 2015: 1871– 80. Google Scholar PubMed  22 Robinson JR, Davis SE, Cronin RM, Jackson GP. Use of a patient portal during hospital admissions to surgical services. AMIA Annu Symp Proc.  2016; 2016: 1967– 76. Google Scholar PubMed  23 Dykes PC, Stade D, Chang Fet al.  , Participatory design and development of a patient-centered toolkit to engage hospitalized patients and care partners in their plan of care. AMIA Annu Symp Proc.  2014; 2014: 486– 95. Google Scholar PubMed  24 Masterson Creber R, Prey J, Ryan Bet al.  , Engaging hospitalized patients in clinical care: study protocol for a pragmatic randomized controlled trial. Contemp Clin Trials.  2016; 47: 165– 171. Google Scholar CrossRef Search ADS PubMed  25 McAlearney AS, Sieck CJ, Hefner JLet al.  , High touch and high tech (HT2) proposal: transforming patient engagement throughout the continuum of care by engaging patients with portal technology at the bedside. JMIR Res Protoc.  2016; 5 4: e221. Google Scholar CrossRef Search ADS PubMed  26 Roberts S, Chaboyer W, Gonzalez R, Marshall A. Using technology to engage hospitalised patients in their care: a realist review. BMC Health Services Res.  2017; 17: 388. Google Scholar CrossRef Search ADS   27 Yamin CK, Emani S, Williams DHet al.  , The digital divide in adoption and use of a personal health record. Arch Intern Med.  2011; 171 6: 568– 74. Google Scholar CrossRef Search ADS PubMed  28 Hsu J, Huang J, Kinsman Jet al.  , Use of e-Health services between 1999 and 2002: a growing digital divide. J Am Med Inform Assoc.  2005; 12 2: 164– 71. Google Scholar CrossRef Search ADS PubMed  29 Tang PC, Overhage JM, Chan ASet al.  , Online disease management of diabetes: engaging and motivating patients online with enhanced resources-diabetes (EMPOWER-D), a randomized controlled trial. J Am Med Inform Assoc.  2013; 20 3: 526– 34. Google Scholar CrossRef Search ADS PubMed  30 Green BB, Cook AJ, Ralston JDet al.  , Effectiveness of home blood pressure monitoring, Web communication, and pharmacist care on hypertension control: a randomized controlled trial. JAMA.  2008; 299 24: 2857– 67. Google Scholar CrossRef Search ADS PubMed  31 Simon GE, Ralston JD, Savarino Jet al.  , Randomized trial of depression follow-up care by online messaging. J Gen Intern Med.  2011; 26 7: 698– 704. Google Scholar CrossRef Search ADS PubMed  32 Krist AH, Woolf SH, Rothemich SFet al.  , Interactive preventive health record to enhance delivery of recommended care: a randomized trial. Ann Fam Med.  2012; 10 4: 312– 19. Google Scholar CrossRef Search ADS PubMed  33 Wright A, Poon EG, Wald Jet al.  , Randomized controlled trial of health maintenance reminders provided directly to patients through an electronic PHR. J Gen Intern Med.  2012; 27 1: 85– 92. Google Scholar CrossRef Search ADS PubMed  34 Goldzweig CL, Orshansky G, Paige NMet al.  , Electronic patient portals: evidence on health outcomes, satisfaction, efficiency, and attitudes: a systematic review. Ann Intern Med.  2013; 159 10: 677– 87. Google Scholar CrossRef Search ADS PubMed  35 Office of the National Coordinator for Health Information. What is a patient portal? Updated November 2, 2015. www.healthit.gov/providers-professionals/faqs/what-patient-portal. Accessed June 25, 2017. © The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of the American Medical Informatics Association Oxford University Press

Loading next page...
 
/lp/ou_press/patient-portal-use-and-hospital-outcomes-0WeYvfDlWK
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
1067-5027
eISSN
1527-974X
D.O.I.
10.1093/jamia/ocx149
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives To determine whether use of a patient portal during hospitalization is associated with improvement in hospital outcomes, 30-day readmissions, inpatient mortality, and 30-day mortality. Materials and Methods We performed a retrospective propensity score–matched study that included all adult patients admitted to Mayo Clinic Hospital in Jacksonville, Florida, from August 1, 2012, to July 31, 2014, who had signed up for a patient portal account prior to hospitalization (N = 7538). Results Out of the admitted patients with a portal account, 1566 (20.8%) accessed the portal while in the hospital. Compared to patients who did not access the portal, patients who accessed the portal were younger (58.8 years vs 62.3 years), had fewer elective admissions (54.2% vs 64.1%), were more frequently admitted to medical services (45.8% vs 35.2%), and were more likely to have liver disease (21.9% vs 12.9%) and higher disease severity scores (0.653 vs 0.456). After propensity score matching, there was no statistically significant difference between the 2 cohorts with respect to 30-day readmission (P = .13), inpatient mortality (P = .82), or 30-day mortality (P = .082). Conclusion Use of the patient portal in the inpatient setting may not improve hospital outcomes. Future research should examine the association of portal use with more immediate inpatient health outcomes such as patient experience, patient engagement, medication reconciliation, and prevention of adverse events. Patient portal, Health information technology, electronic health record, patient access, hospital outcomes BACKGROUND In 1996, the Health Insurance Portability and Accountability Act gave patients the legal right to view and own their medical records.1 Unrestricted patient access to medical records was somewhat difficult in the era of paper charts, but after the introduction of the electronic health record (EHR) and, subsequently, EHR-tethered personal health record electronic entry, or “patient portals,” patients’ access to their records became less challenging. In 2009, the Health Information Technology for Economic and Clinical Health Act published meaningful use rules for using the EHR,2 which included that the EHR should be used to provide effective health information exchange to improve health care quality and patient-centered care. In light of these rules, it is expected that patient portal use will be more associated with medical outcomes.3 Over the past several years, patient portal use has increased in both the outpatient and inpatient settings.4,5 Online, patients can review labs, provider notes, and medication and problem lists, request prescription renewals, manage appointments, and communicate with their physicians. While patient portal use is increasing,6 the inpatient environment presents additional challenges for patients in receiving information about their care, which may affect use. The clinical conditions leading to hospitalization are often acute, at times with devastating consequences. The amount of medical information generated during a hospitalization can be extensive and delivered to the patient rapidly over a short period of time. Consequently, patients may have difficulty comprehending and integrating this information, as well as understanding the plan of care.7 Furthermore, it is not uncommon for care teams to be new to patients and their families, further complicating communication. It is known that the majority of patients are unable to identify their primary inpatient providers.8 Hospitalized patients’ immediate access to the online portal is an additional challenge, since devices to access the Internet are not always available in the hospital. The advent of Internet-connected smartphones and tablets has facilitated online access.9 Despite the barriers, patients want to receive and review health information electronically during their hospitalization and after discharge.9–13 There is limited information regarding patient portal use in the inpatient setting and its association with clinical outcomes.14 Most of the information regarding inpatient portal use is obtained from survey-based quasi-experimental studies of limited numbers of patients.14 These studies report usability and satisfaction outcomes: patients’ intention to use the portal,9 inpatient portal utilization,15 and satisfaction with use of the portal tool.13 Offering hospitalized patients access to a web portal via tablet computers increased their ability to correctly name more than one physician and their roles.15 While these studies offer an indication that immediate health outcomes may be positive, to date no study has examined the association between hospital outcomes of 30-day readmissions, inpatient mortality, and 30-day mortality and patient portal use. Patients who use the portal to access their medical records while hospitalized may have an active interest in their outcome and information about their hospital stay. We pose the question of whether this engagement would reduce 30-day readmissions or decrease mortality. We asked two main questions in this study: How do patients who access their portal accounts in the inpatient setting differ from those with access who do not? Is there a relationship between inpatient portal use and hospital outcomes of 30-day readmission, 30-day mortality, and inpatient mortality? To our knowledge, this is the first study to examine hospital outcomes associated with inpatient portal use. MATERIALS AND METHODS Study design and subjects This retrospective study included all adult patients admitted to Mayo Clinic Hospital in Jacksonville, Florida, from August 1, 2012, to July 31, 2014, who had a patient portal account prior to hospitalization. If a patient had more than one inpatient admission during this time frame, only the demographic and clinical data from the first encounter was collected and analyzed. Patients were excluded if they were under 18 years of age at the time of admission. Hospital observation stays were not included. The study was approved by the Mayo Clinic Institutional Review Board. Data collection The EHR was used to obtain information on patient demographics (age at admission, sex, race, ethnicity, marital status, primary language, payor information, and employment status) and clinical information known at the time of admission (admission status, admission service, and comorbidities), and hospital outcomes (30-day readmission, 30-day mortality, and inpatient mortality). Principal and secondary diagnoses and procedures, comorbidities, and All-Patient Refined Diagnosis Related Group (APR-DRG) weight16 were collected based on International Classification of Diseases, Ninth Revision, Clinical Modification codes extracted from hospital discharge abstracts. Hospital length of stay (LOS) was calculated using the dates and times of hospital admission and hospital discharge. Patient Online Services (patient portal) The Florida campus of Mayo Clinic implemented the Patient Online Services (patient portal) in 2010, initially providing patients with secure Internet access to their laboratory data. On March 1, 2012, access to outpatient physician notes was added. In May 2011, Mayo Clinic launched the Mayo Clinic Mobile app, which allows patients the same access as the web portal. This app was compatible with Apple iOS initially, and then expanded to Android as well. The app gives patients the flexibility to view their information on the go, to sync appointments to electronic calendars, and to set appointment reminders. The patient portal is tethered to Mayo Clinic Florida’s EHR, PowerChart (Cerner Corporation, North Kansas City, MO, USA). Information is extracted from PowerChart and displayed by Mayo’s proprietary online and mobile-based applications. The date and time that patients access the portal can be determined; however, the content or modules accessed cannot, due to a system limitation with PowerChart. Patients routinely create portal accounts in the outpatient setting. Once an account is created, the portal gives a patient the opportunity to view lab results, clinic progress notes, and medication and allergy lists, and to send messages to care providers. When a patient who signed up for an account is hospitalized, the patient can then use the portal to access inpatient labs, inpatient admission notes, consultation reports, operative notes, and discharge summaries in real time. Radiology and pathology reports are available with a 72-h delay. Providers’ daily progress notes are not viewable. While patient-to-provider electronic messaging is encouraged in the outpatient setting, communication between hospitalized patients and hospital care teams is not available. Statistical analysis Preadmission characteristics were summarized according to portal use during hospitalization among the cohort of patients who had a portal account prior to admission. We aimed to estimate the effect of portal use during hospitalization on hospital outcomes (inpatient mortality, 30-day readmission, and 30-day mortality). In an attempt to control confounding, propensity score matching was used to identify a cohort of patients with similar baseline characteristics. Propensity score is defined here as the conditional probability of an inpatient using his or her existing portal account given a set of covariates known at the time of admission (baseline). A multivariable logistic regression model with inpatient portal use as the dependent variable and all the baseline characteristics displayed in Table 1 as covariates was used to estimate propensity score. Up to 3 non–portal users17 for each portal user were selected utilizing a greedy matching algorithm without replacement18 with a caliper width equal to 0.2 of the standard deviation (SD) of the logit of the propensity score.19,20 To assess potential imbalance in baseline characteristics between the 2 groups, standardized differences were estimated before and after matching. A standardized difference of <10% for a given baseline characteristic was considered a negligible imbalance between groups.20 In the matched cohort of non–portal users, summary statistics were weighted such that each patient was assigned a weight of 1 divided by the number of non–portal users matched to the corresponding portal user for the given analysis. Table 1. Preadmission characteristics of the full cohort of patients who had a portal account established prior to hospital admission according to portal use during the inpatient stay Characteristic  No portal use (n = 5972)  Portal use (n = 1566)  Standardized difference before matching, %  Mean age, years  62.3 ± 15.1  58.8 ± 15.7  22.7  Female sex (%)  3121 (52.3)  739 (47.2)  10.2  Race (%)         Black  361 (6.0)  53 (3.4)  12.6   White  5374 (90.0)  1428 (91.2)  4.1   Other  144 (2.4)  71 (4.5)  11.6   Unknown  93 (1.6)  14 (0.9)  6.0  Ethnic group (%)         Hispanic or Latino  202 (3.4)  81 (5.2)  8.9   Not Hispanic or Latino  5625 (94.2)  1457 (93.0)  4.7   Unknown  145 (2.4)  28 (1.8)  4.5  Marital status (%)         Married or life partner  4371 (73.2)  1194 (76.2)  7.0   Divorced or separated  502 (8.4)  103 (6.6)  7.0   Widowed  491 (8.2)  95 (6.1)  8.4   Single  603 (10.1)  171 (10.9)  2.7   Unknown  5 (0.1)  3 (0.2)  2.9  Language (%)         English  5857 (98.1)  1504 (96.0)  12.1   Spanish  49 (0.8)  29 (1.9)  9.0   Other  31 (0.5)  19 (1.2)  7.5   Unknown  35 (0.6)  14 (0.9)  3.6  Government payor (%)  3315 (55.5)  723 (46.2)  18.8  Employment status (%)         Employed  1730 (29.0)  488 (31.2)  4.8   Not employed  663 (11.1)  249 (15.9)  14.1   Retired  2040 (34.2)  489 (31.2)  6.3   Disabled  351 (5.9)  107 (6.8)  3.9   Unknown  1188 (19.9)  233 (14.9)  13.3  Admission status (%)         Routine/elective  3831 (64.1)  848 (54.2)  20.4   Semi-urgent  161 (2.7)  80 (5.1)  12.5   Urgent  1718 (28.8)  564 (36.0)  15.5   Emergency  239 (4.0)  61 (3.9)  0.5   Other  23 (0.4)  13 (0.8)  5.7  Admission service (%)         Critical  244 (4.1)  85 (5.4)  6.3   Medical  2104 (35.2)  717 (45.8)  21.6   Surgical  3624 (60.7)  764 (48.8)  24.1  Myocardial infarct (%)  291 (4.9)  87 (5.6)  3.1  Congestive heart failure (%)  541 (9.1)  147 (9.4)  1.1  Peripheral vascular disease (%)  1045 (17.5)  262 (16.7)  2.0  Cerebrovascular disease (%)  718 (12.0)  162 (10.3)  5.3  Dementia (%)  253 (4.2)  67 (4.3)  0.2  Chronic pulmonary disease (%)  851 (14.2)  196 (12.5)  5.1  Ulcer (%)  150 (2.5)  47 (3.0)  3.0  Mild liver disease (%)  766 (12.8)  335 (21.4)  22.9  Diabetes (%)  1050 (17.6)  285 (18.2)  1.6  Diabetes with organ damage (%)  130 (2.2)  39 (2.5)  2.1  Hemiplegia (%)  78 (1.3)  22 (1.4)  0.9  Moderate/severe renal disease (%)  690 (11.6)  196 (12.5)  3.0  Moderate/severe liver disease (%)  273 (4.6)  168 (10.7)  23.3  Metastatic solid tumor (%)  473 (7.9)  178 (11.4)  11.7  AIDS (%)  8 (0.1)  4 (0.3)  2.8  Rheumatologic disease (%)  208 (3.5)  52 (3.3)  0.9  Other cancer (%)  1743 (29.2)  569 (36.3)  15.3  Mean APR-DRG weight, log scale  0.456 ± 0.725  0.653 ± 0.859  24.7  Characteristic  No portal use (n = 5972)  Portal use (n = 1566)  Standardized difference before matching, %  Mean age, years  62.3 ± 15.1  58.8 ± 15.7  22.7  Female sex (%)  3121 (52.3)  739 (47.2)  10.2  Race (%)         Black  361 (6.0)  53 (3.4)  12.6   White  5374 (90.0)  1428 (91.2)  4.1   Other  144 (2.4)  71 (4.5)  11.6   Unknown  93 (1.6)  14 (0.9)  6.0  Ethnic group (%)         Hispanic or Latino  202 (3.4)  81 (5.2)  8.9   Not Hispanic or Latino  5625 (94.2)  1457 (93.0)  4.7   Unknown  145 (2.4)  28 (1.8)  4.5  Marital status (%)         Married or life partner  4371 (73.2)  1194 (76.2)  7.0   Divorced or separated  502 (8.4)  103 (6.6)  7.0   Widowed  491 (8.2)  95 (6.1)  8.4   Single  603 (10.1)  171 (10.9)  2.7   Unknown  5 (0.1)  3 (0.2)  2.9  Language (%)         English  5857 (98.1)  1504 (96.0)  12.1   Spanish  49 (0.8)  29 (1.9)  9.0   Other  31 (0.5)  19 (1.2)  7.5   Unknown  35 (0.6)  14 (0.9)  3.6  Government payor (%)  3315 (55.5)  723 (46.2)  18.8  Employment status (%)         Employed  1730 (29.0)  488 (31.2)  4.8   Not employed  663 (11.1)  249 (15.9)  14.1   Retired  2040 (34.2)  489 (31.2)  6.3   Disabled  351 (5.9)  107 (6.8)  3.9   Unknown  1188 (19.9)  233 (14.9)  13.3  Admission status (%)         Routine/elective  3831 (64.1)  848 (54.2)  20.4   Semi-urgent  161 (2.7)  80 (5.1)  12.5   Urgent  1718 (28.8)  564 (36.0)  15.5   Emergency  239 (4.0)  61 (3.9)  0.5   Other  23 (0.4)  13 (0.8)  5.7  Admission service (%)         Critical  244 (4.1)  85 (5.4)  6.3   Medical  2104 (35.2)  717 (45.8)  21.6   Surgical  3624 (60.7)  764 (48.8)  24.1  Myocardial infarct (%)  291 (4.9)  87 (5.6)  3.1  Congestive heart failure (%)  541 (9.1)  147 (9.4)  1.1  Peripheral vascular disease (%)  1045 (17.5)  262 (16.7)  2.0  Cerebrovascular disease (%)  718 (12.0)  162 (10.3)  5.3  Dementia (%)  253 (4.2)  67 (4.3)  0.2  Chronic pulmonary disease (%)  851 (14.2)  196 (12.5)  5.1  Ulcer (%)  150 (2.5)  47 (3.0)  3.0  Mild liver disease (%)  766 (12.8)  335 (21.4)  22.9  Diabetes (%)  1050 (17.6)  285 (18.2)  1.6  Diabetes with organ damage (%)  130 (2.2)  39 (2.5)  2.1  Hemiplegia (%)  78 (1.3)  22 (1.4)  0.9  Moderate/severe renal disease (%)  690 (11.6)  196 (12.5)  3.0  Moderate/severe liver disease (%)  273 (4.6)  168 (10.7)  23.3  Metastatic solid tumor (%)  473 (7.9)  178 (11.4)  11.7  AIDS (%)  8 (0.1)  4 (0.3)  2.8  Rheumatologic disease (%)  208 (3.5)  52 (3.3)  0.9  Other cancer (%)  1743 (29.2)  569 (36.3)  15.3  Mean APR-DRG weight, log scale  0.456 ± 0.725  0.653 ± 0.859  24.7  Categorical characteristics are given as the number and percentage of patients, while numerical characteristics are given as mean ± SD. The sample mean of APR-DRG weight (0.497, log scale) was imputed for 14 patients (12 with no portal use and 2 with portal use) who were missing that information. Abbreviations: APR-DRG, All-Patient Refined Diagnosis Related Group. Conditional logistic regression was used to evaluate the impact of portal use on inpatient mortality, 30-day readmission, and 30-day mortality. All models included a variable representing the match identifier as a stratification factor in order to preserve the benefit of matching. Models evaluating 30-day outcomes were adjusted for LOS (log scale). Patients were excluded from analyses evaluating 30-day outcomes if they died during their hospital stay or if they did not have a corresponding match after patients who died during their hospital stay were excluded. Statistical analyses were performed using SAS statistical software (version 9.4, SAS Institute Inc., Cary, NC, USA), and all reported P-values are 2-sided without adjustment for multiple testing. RESULTS We identified 17 050 patients in the study period: 7538 (44.2%) had a portal account established at the time of admission and 9512 (55.8%) did not. Of the patients who already had portal accounts, 1566 (20.8%) accessed the portal during hospitalization. Patients’ preadmission characteristics before matching are summarized in Table 1 and show substantial differences (standardized difference >20%) among those who accessed the portal by age, elective admission status, admission to medical or surgical services, comorbidity of liver disease, and disease severity (APR-DRG weight). Patients who accessed their portal account were more likely to be younger, to have more elective vs urgent admissions, to be admitted to medical services versus surgical services, and to have liver disease, and had higher disease severity scores than patients who did not access the portal during their stay. We included 7538 patients with a portal account at the time of admission in the logistic regression model to estimate the propensity score for using the portal account during hospitalization. The c index from that model was 0.67. The mean propensity score among the patients who accessed their portal account during hospitalization was 0.260 (SD = 0.125), compared with 0.194 (SD = 0.094) for patients who did not access their portal account during hospitalization. As seen in Table 2, all differences in preadmission patient characteristics between those who did and did not use the portal during hospitalization were considered negligible (all standardized differences ≤2.0%) after propensity score matching. Table 2. Preadmission characteristics after propensity score matching Characteristic  No in-hospital portal use (n = 3922)  In-hospital portal use (n = 1559)  Standardized difference after matching, %  Age, years  58.7 ± 10.0  58.9 ± 15.7  1.5  Female sex  47.4  47.2  0.3  Race         Black  3.3  3.4  0.4   White  91.5  91.4  0.2   Other  4.2  4.3  0.5   Unknown  1.0  0.9  1.1  Ethnic group         Hispanic or Latino  4.9  5.1  0.6   Not Hispanic or Latino  93.2  93.1  0.4   Unknown  1.8  1.8  0.2  Marital status         Married or life partner  75.6  76.3  1.7   Divorced or separated  6.7  6.6  0.4   Widowed  6.1  6.1  0.0   Single  11.5  10.8  2.0   Unknown  0.2  0.2  0.2  Language         English  96.3  96.2  0.6   Spanish  1.8  1.8  0.0   Other  1.2  1.2  0.1   Unknown  0.8  0.9  1.4  Government payor  47.0  46.3  1.5  Employment status         Employed  31.0  31.2  0.6   Not employed  15.8  15.7  0.5   Retired  31.1  31.3  0.4   Disabled  7.3  6.9  1.5   Unknown  14.8  14.9  0.3  Admission status         Routine/elective  53.7  54.2  0.9   Semi-urgent  5.0  5.0  0.0   Urgent  36.7  36.0  1.3   Emergency  3.8  3.9  0.5   Other  0.7  0.8  1.0  Admission service         Critical  5.6  5.3  1.2   Medical  45.6  45.7  0.2   Surgical  48.8  49.0  0.3  Myocardial infarct  5.8  5.6  0.8  Congestive heart failure  9.7  9.4  0.9  Peripheral vascular disease  16.5  16.8  0.8  Cerebrovascular disease  10.2  10.4  0.7  Dementia  4.4  4.3  0.4  Chronic pulmonary disease  12.5  12.6  0.2  Ulcer  2.8  3.0  1.1  Mild liver disease  21.0  21.0  0.0  Diabetes  18.5  18.2  0.7  Diabetes with organ damage  2.6  2.5  0.9  Hemiplegia  1.5  1.4  0.9  Moderate/severe renal disease  12.6  12.6  0.1  Moderate/severe liver disease  10.2  10.5  1.1  Metastatic solid tumor  11.5  11.2  0.9  AIDS  0.2  0.3  1.4  Rheumatologic disease  3.2  3.3  0.5  Other cancer  36.3  36.2  0.2  APR-DRG weight, log scale  0.639 ± 0.555  0.645 ± 0.852  0.8  Characteristic  No in-hospital portal use (n = 3922)  In-hospital portal use (n = 1559)  Standardized difference after matching, %  Age, years  58.7 ± 10.0  58.9 ± 15.7  1.5  Female sex  47.4  47.2  0.3  Race         Black  3.3  3.4  0.4   White  91.5  91.4  0.2   Other  4.2  4.3  0.5   Unknown  1.0  0.9  1.1  Ethnic group         Hispanic or Latino  4.9  5.1  0.6   Not Hispanic or Latino  93.2  93.1  0.4   Unknown  1.8  1.8  0.2  Marital status         Married or life partner  75.6  76.3  1.7   Divorced or separated  6.7  6.6  0.4   Widowed  6.1  6.1  0.0   Single  11.5  10.8  2.0   Unknown  0.2  0.2  0.2  Language         English  96.3  96.2  0.6   Spanish  1.8  1.8  0.0   Other  1.2  1.2  0.1   Unknown  0.8  0.9  1.4  Government payor  47.0  46.3  1.5  Employment status         Employed  31.0  31.2  0.6   Not employed  15.8  15.7  0.5   Retired  31.1  31.3  0.4   Disabled  7.3  6.9  1.5   Unknown  14.8  14.9  0.3  Admission status         Routine/elective  53.7  54.2  0.9   Semi-urgent  5.0  5.0  0.0   Urgent  36.7  36.0  1.3   Emergency  3.8  3.9  0.5   Other  0.7  0.8  1.0  Admission service         Critical  5.6  5.3  1.2   Medical  45.6  45.7  0.2   Surgical  48.8  49.0  0.3  Myocardial infarct  5.8  5.6  0.8  Congestive heart failure  9.7  9.4  0.9  Peripheral vascular disease  16.5  16.8  0.8  Cerebrovascular disease  10.2  10.4  0.7  Dementia  4.4  4.3  0.4  Chronic pulmonary disease  12.5  12.6  0.2  Ulcer  2.8  3.0  1.1  Mild liver disease  21.0  21.0  0.0  Diabetes  18.5  18.2  0.7  Diabetes with organ damage  2.6  2.5  0.9  Hemiplegia  1.5  1.4  0.9  Moderate/severe renal disease  12.6  12.6  0.1  Moderate/severe liver disease  10.2  10.5  1.1  Metastatic solid tumor  11.5  11.2  0.9  AIDS  0.2  0.3  1.4  Rheumatologic disease  3.2  3.3  0.5  Other cancer  36.3  36.2  0.2  APR-DRG weight, log scale  0.639 ± 0.555  0.645 ± 0.852  0.8  Categorical characteristics are given as the percentage of patients, while numerical characteristics are given as mean ± SD. Summary statistics in the group with no in-hospital portal use are weighted such that each patient was assigned a weight of 1 divided by the number of patients with no in-hospital portal use matched to the corresponding portal user. Abbreviations: APR-DRG, All-Patient Refined Diagnosis Related Group. All but 7 patients who used the portal were matched to one or more patients who did not use the portal during their hospital stay. Matching retained 72.7% of the original cohort. The mean propensity score in the matched groups was 0.258 in the group that used the portal (SD = 0.120, n = 1559) and 0.227 in the group that did not use the portal (SD = 0.098, n = 3922). Associations of portal use with hospital outcomes in the propensity-matched cohort are shown in Table 3. There was no significant association of portal use with inpatient mortality (P = .82), and, after adjusting for LOS, no statistically significant association with 30-day readmission (P = .13) or 30-day mortality (P = .087). Table 3. Association of patient portal use with hospital outcomes in the propensity-matched cohort Outcome  No.  OR (95% CI)  P-value  Inpatient mortality  5481  0.94 (0.55, 1.61)  .82  30-day readmission  5374  1.16 (0.96, 1.39)  .13  30-day mortality  5374  1.37 (0.96, 1.97)  .087  Outcome  No.  OR (95% CI)  P-value  Inpatient mortality  5481  0.94 (0.55, 1.61)  .82  30-day readmission  5374  1.16 (0.96, 1.39)  .13  30-day mortality  5374  1.37 (0.96, 1.97)  .087  Odds ratios (ORs) and 95% confidence intervals (CIs) result from conditional logistic regression models with in-hospital patient portal use as a predictor variable. Propensity match identifier was included in each model as a stratification factor to preserve the benefit of matching. Models of 30-day outcomes include length of hospital stay on the logarithm scale as a covariate. Patients were excluded from the 30-day outcomes analysis (n = 107) if they died in the hospital (n = 65: 44 patients with no in-hospital portal use and 21 patients with in-hospital portal use) or if they did not have a corresponding match after patients who died in the hospital were excluded (n = 42: 34 patients with no in-hospital portal use and 8 patients with in-hospital portal). Among the 5374 patients included in the 30-day outcomes analyses, 609 patients were readmitted within 30 days after hospital discharge (403 patients with no in-hospital portal use and 206 patients with in-hospital portal use) and 142 patients died within 30 days after hospital discharge (83 patients with no in-hospital portal use and 59 patients with in-hospital portal use). DISCUSSION The primary objective of this study was to compare patients with portal accounts who accessed and did not access the portal during hospitalization and to determine if there was an association between inpatient portal use and 30-day readmissions and mortality. In this study, patients who accessed their portal account were younger and had greater disease severity and more urgent admissions. It is possible that patients (or their families) access the portal when they are uncertain of the hospital course. This would account for lower use among surgical patients (large portion of elective surgery) and those with shorter stays. After propensity matching, there were no differences in 30-day readmissions, inpatient mortality, or 30-day mortality after adjusting for LOS. Of the 44.2% of patients who had a portal account at the time of admission, only 20.8% accessed the portal while hospitalized. There are several possible reasons for low inpatient portal use. Patient education regarding portal access and the potential benefits of portal use was not provided. The inpatient portal lacks specific features that allow communication between patients and care teams. Only admission notes, operative notes, consultations, and laboratory studies are available in real time. Daily progress notes cannot be viewed, and there is a 72-h delay in the viewing of radiology and pathology reports. Finally, the inpatient portal does not provide education regarding patient-specific disease processes. Other studies similarly report low inpatient use. Davis noted that 34.4% of admitted patients were registered for the portal, but only 23.4% used it during hospitalization.21 Among surgical patients admitted to a tertiary-care hospital, 25.3% had a portal account but only 16% of registered users accessed their account.22 The lack of features designed specifically for inpatient use was previously emphasized in a systematic review.14 Consequently, several medical centers designed hospital-specific applications aimed at improving the use and usability of inpatient portals.23–25 In a realistic review, Roberts indicated that patient participation with inpatient health information technology (including patient portals) can be augmented by interactive learning focused on information sharing, self-assessment and feedback, tailored education, user-centered design, and user support.26 Studies show that in the outpatient setting, patients with severe disease use portals more frequently,27,28 and patients who access portals have better outcomes in certain chronic medical conditions such as diabetes (lower HbA1c at 6 months),29 hypertension (improved blood pressure control at 12 months),30 depression management (increased medication adherence),31 and preventative care (up-to-date immunizations and mammograms).32,33 Our study indicates that unprompted and unguided patient portal use does not have any benefit with respect to 30-day readmissions and mortality. There are several possible explanations for this. First, use of a patient portal may play a more important role in the longitudinal management of chronic medical issues than in the outcomes associated with acute management of a decompensating clinical situation. While patients and their families can review clinical information via patient portals, this passive, unguided review of clinical data may have little bearing on the overall hospital and immediate posthospital course. In a systematic review, Goldzweig et al.34 noted that positive outcomes for patients with chronic diseases such as diabetes, hypertension, and depression were obtained only when portal use was enhanced with case management intervention, and the cohorts that showed clinical benefit received tailored education about portal use. Our patients were not specifically trained to use the portal, nor did they receive targeted intervention such as case management via the portal. Secondly, in the outpatient setting, outcome improvement was noted mostly for chronic diseases where the disease course is influenced by self-monitoring and behavioral modifications (diabetes, hypertension, asthma, HIV, fertility management, glaucoma, and hyperlipidemia).12 Due to the acute nature of inpatient conditions, behavioral modification is less feasible in the short time associated with hospitalization. Use of the patient portal is strongly encouraged by the Centers for Medicare and Medicaid Services, presumably to better inform patients and improve outcomes.35 Our study shows that clinical outcomes are not improved. Our study has a number of limitations. The population cohorts were selected from patients admitted to a single academic tertiary-care referral center, which precludes generalizability of our findings. Education level, which could affect portal use, is not routinely recorded in the EHR, so we could not use this information. Another limitation is that only certain information can be accessed by patients in real time, while there is a 72-h delay with pathology reports and radiology testing. The effect of patient portal use may be different in hospitals that embrace the “open notes” concept, allowing patients immediate access to provider notes and test results. Due to a system limitation, we could not verify the type of information accessed by patients in the portal. Also, we did not control the device type (PC, tablet, phone) used to access the patient portal. Free Internet connection is ubiquitous in the study institution, and the patient portal has iterations that can be accessed on different platforms under multiple operating systems and via web browsing. Mobile device availability and patient familiarity with the mobile version of the patient portal were not evaluated. Although the information in the patient portal is private, some hospitalized patients may have shared their log-in information with their family, friends, or primary care physician, and we did not account for the patient’s level of alertness at the time of portal access, nor did we evaluate who actually accessed the information. Prior inpatient or ambulatory use may be associated with inpatient portal use, and this was not assessed. Finally, there may be immediate outcomes that we did not directly examine where use of an inpatient portal may be of benefit, such as medication reconciliation, prevention of adverse events, patient satisfaction, and patient engagement. CONCLUSION This is the first study to examine the association between 30-day readmissions, 30-day mortality, and inpatient mortality and inpatient portal use. We found that there was no difference in 30-day readmissions, inpatient mortality, or 30-day mortality between portal and non–portal users and that inpatient portal use was low. The first step in determining whether patient portal use can improve hospital outcomes is to increase adoption and use by designing inpatient-specific portal tools that can engage patients and make them active participants in their health care. Future research should examine the role of portal use on more immediate outcomes such as patient experience and engagement to help clarify how patient portals can be used to improve outcomes. FUNDING This work was supported by the Robert D and Patricia E Kern Center for the Science of Health Care Delivery, Mayo Clinic, Jacksonville, Florida. COMPETING INTERESTS The authors have no competing interests to declare. CONTRIBUTORS AGD, MCB, and NLD conceived and designed the study. AGD, LMN, and CST collected the data. CST and JMN analyzed the data, designed the statistical method, and drafted the tables. AGD, MCB, NLD, and CST drafted the article. HEG, DIA, and JMN critically revised the article for important intellectual content. All co-authors contributed to writing, proofreading, and editing the manuscript. ACKNOWLEDGMENTS Launia J. White for her help with data extraction. REFERENCES 1 Health Insurance Portability and Accountability Act of 1996: Public Law 104–191—Aug. 21, 1996. US Government Publishing Office. www.gpo.gov/fdsys/pkg/PLAW-104publ191/pdf/PLAW-104publ191.pdf. Accessed March 18, 2014. 2 Title XIII: Health Information Technology (HITECH Act). www.healthit.gov/sites/default/files/hitech_act_excerpt_from_arra_with_index.pdf. Accessed March 18, 2014. 3 Kruse CS, Bolton K, Freriks G. The effect of patient portals on quality outcomes and its implications to meaningful use: a systematic review. J Med Internet Res.  2015; 17 2: e44. Google Scholar CrossRef Search ADS PubMed  4 Patel V, Barker W, Siminerio E. ONC Data Brief, No. 30: Trends in Consumer Access and Use of Electronic Health Information . Washington, DC: Office of the National Coordinator for Health Information Technology; 2015. www.healthit.gov/sites/default/files/briefs/oncdatabrief30_accesstrends_.pdf. Accessed November 1, 2016. 5 Charles D, Gabriel M, Henry J. ONC Data Brief 29: Electronic Capabilities for Patient Engagement among U.S. Non-Federal Acute Care Hospitals: 2012–2014 . Washington, DC: Office of the National Coordinator for Health Informaiton Technology; 2015. dashboard.healthit.gov/evaluations/data-briefs/hospitals-patient-engagement-electronic-capabilities.php. Accessed November 1, 2016. 6 Ammenwerth E, Schnell-Inderst P, Hoerbst A. The impact of electronic patient portals on patient care: a systematic review of controlled trials. J Med Internet Res.  2012; 14 6: e162. Google Scholar CrossRef Search ADS PubMed  7 O’Leary KJ, Kulkarni N, Landler MPet al.  , Hospitalized patients’ understanding of their plan of care. Mayo Clin Proc.  2010; 85 1: 47– 52. Google Scholar CrossRef Search ADS PubMed  8 Olson DP, Windish DM. Communication discrepancies between physicians and hospitalized patients. Arch Intern Med.  2010; 170 15: 1302– 07. Google Scholar CrossRef Search ADS PubMed  9 O’Leary KJ, Balabanova A, Patyk Met al.  , Medical inpatients’ use of information technology: characterizing the potential to share information electronically. J Healthc Qual.  2015; 37 4: 207– 20. Google Scholar CrossRef Search ADS PubMed  10 Burke RP, Rossi AF, Wilner BRet al.  , Transforming patient and family access to medical information: utilisation patterns of a patient-accessible electronic health record. Cardiol Young.  2010; 20 5: 477– 84. Google Scholar CrossRef Search ADS PubMed  11 Sprague J, Pell J, Lin CT. Divergent care team opinions about online release of test results to an ICU patient. J Participat Med.  2013; 5: e24. 12 Price M, Bellwood P, Kitson Net al.  , Conditions potentially sensitive to a personal health record (PHR) intervention, a systematic review. BMC Med Inform Decis Mak.  2015; 15: 32. Google Scholar CrossRef Search ADS PubMed  13 Greysen SR, Khanna RR, Jacolbia Ret al.  , Tablet computers for hospitalized patients: a pilot study to improve inpatient engagement. J Hosp Med.  2014; 9 6: 396– 69. Google Scholar CrossRef Search ADS PubMed  14 Prey JE, Woollen J, Wilcox Let al.  , Patient engagement in the inpatient setting: a systematic review. J Am Med Inform Assoc.  2014; 21 4: 742– 50. Google Scholar CrossRef Search ADS PubMed  15 O’Leary KJ, Lohman ME, Culver Eet al.  , The effect of tablet computers with a mobile patient portal application on hospitalized patients’ knowledge and activation. J Am Med Inform Assoc.  2016; 23 1: 159– 65. Google Scholar CrossRef Search ADS PubMed  16 Averill RF, Goldfield N, Bonazelli Jet al.  , All Patient Refined Diagnosis Related Groups (APR-DRGs), Version 20.0: Methodology Overview . Wallingford, CT: 3M Health Information Systems; 2003. 17 Ming K, Rosenbaum PR. Substantial gains in bias reduction from matching with a variable number of controls. Biometrics.  2000; 56 1: 118– 24. Google Scholar CrossRef Search ADS PubMed  18 Austin PC. Statistical criteria for selecting the optimal number of untreated subjects matched to each treated subject when using many-to-one matching on the propensity score. Am J Epidemiol.  2010; 172 9: 1092– 97. Google Scholar CrossRef Search ADS PubMed  19 Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat.  2011; 10 2: 150– 61. Google Scholar CrossRef Search ADS PubMed  20 Faries D, Leon AC, Haro JMet al.  , Analysis of Observational Health Care Data Using SAS® . Cary, NC: SAS Institute; 2010. 21 Davis SE, Osborn CY, Kripalani Set al.  , Health literacy, education levels, and patient portal usage during hospitalizations. AMIA Annu Symp Proc.  2015; 2015: 1871– 80. Google Scholar PubMed  22 Robinson JR, Davis SE, Cronin RM, Jackson GP. Use of a patient portal during hospital admissions to surgical services. AMIA Annu Symp Proc.  2016; 2016: 1967– 76. Google Scholar PubMed  23 Dykes PC, Stade D, Chang Fet al.  , Participatory design and development of a patient-centered toolkit to engage hospitalized patients and care partners in their plan of care. AMIA Annu Symp Proc.  2014; 2014: 486– 95. Google Scholar PubMed  24 Masterson Creber R, Prey J, Ryan Bet al.  , Engaging hospitalized patients in clinical care: study protocol for a pragmatic randomized controlled trial. Contemp Clin Trials.  2016; 47: 165– 171. Google Scholar CrossRef Search ADS PubMed  25 McAlearney AS, Sieck CJ, Hefner JLet al.  , High touch and high tech (HT2) proposal: transforming patient engagement throughout the continuum of care by engaging patients with portal technology at the bedside. JMIR Res Protoc.  2016; 5 4: e221. Google Scholar CrossRef Search ADS PubMed  26 Roberts S, Chaboyer W, Gonzalez R, Marshall A. Using technology to engage hospitalised patients in their care: a realist review. BMC Health Services Res.  2017; 17: 388. Google Scholar CrossRef Search ADS   27 Yamin CK, Emani S, Williams DHet al.  , The digital divide in adoption and use of a personal health record. Arch Intern Med.  2011; 171 6: 568– 74. Google Scholar CrossRef Search ADS PubMed  28 Hsu J, Huang J, Kinsman Jet al.  , Use of e-Health services between 1999 and 2002: a growing digital divide. J Am Med Inform Assoc.  2005; 12 2: 164– 71. Google Scholar CrossRef Search ADS PubMed  29 Tang PC, Overhage JM, Chan ASet al.  , Online disease management of diabetes: engaging and motivating patients online with enhanced resources-diabetes (EMPOWER-D), a randomized controlled trial. J Am Med Inform Assoc.  2013; 20 3: 526– 34. Google Scholar CrossRef Search ADS PubMed  30 Green BB, Cook AJ, Ralston JDet al.  , Effectiveness of home blood pressure monitoring, Web communication, and pharmacist care on hypertension control: a randomized controlled trial. JAMA.  2008; 299 24: 2857– 67. Google Scholar CrossRef Search ADS PubMed  31 Simon GE, Ralston JD, Savarino Jet al.  , Randomized trial of depression follow-up care by online messaging. J Gen Intern Med.  2011; 26 7: 698– 704. Google Scholar CrossRef Search ADS PubMed  32 Krist AH, Woolf SH, Rothemich SFet al.  , Interactive preventive health record to enhance delivery of recommended care: a randomized trial. Ann Fam Med.  2012; 10 4: 312– 19. Google Scholar CrossRef Search ADS PubMed  33 Wright A, Poon EG, Wald Jet al.  , Randomized controlled trial of health maintenance reminders provided directly to patients through an electronic PHR. J Gen Intern Med.  2012; 27 1: 85– 92. Google Scholar CrossRef Search ADS PubMed  34 Goldzweig CL, Orshansky G, Paige NMet al.  , Electronic patient portals: evidence on health outcomes, satisfaction, efficiency, and attitudes: a systematic review. Ann Intern Med.  2013; 159 10: 677– 87. Google Scholar CrossRef Search ADS PubMed  35 Office of the National Coordinator for Health Information. What is a patient portal? Updated November 2, 2015. www.healthit.gov/providers-professionals/faqs/what-patient-portal. Accessed June 25, 2017. © The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of the American Medical Informatics AssociationOxford University Press

Published: Apr 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial