This paper is in the following e-collection/theme issue:

Participatory Medicine & E-Patients (773) Adoption and Change Management of eHealth Systems (712)

Published on in Vol 21, No 7 (2019): July

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/12373, first published .
Factors Affecting Patients’ Use of Electronic Personal Health Records in England: Cross-Sectional Study

Factors Affecting Patients’ Use of Electronic Personal Health Records in England: Cross-Sectional Study

Factors Affecting Patients’ Use of Electronic Personal Health Records in England: Cross-Sectional Study

Authors of this article:

Alaa Abd-Alrazaq1, 2 Author Orcid Image ;   Bridgette M Bewick1 Author Orcid Image ;   Tracey Farragher1 Author Orcid Image ;   Peter Gardner3 Author Orcid Image
Article Authors Cited by (23) Tweetations (19) Metrics

    Original Paper

    1Leeds Institute of Health Sciences, School of Medicine, University of Leeds, London, United Kingdom

    2Division of Information and Computing Technology, College of Science and Engineering, Hamad Bin Khalifa University, Doha, Qatar

    3School of Psychology, University of Leeds, Leeds, United Kingdom

    Corresponding Author:

    Alaa Abd-Alrazaq, PhD

    Division of Information and Computing Technology

    College of Science and Engineering

    Hamad Bin Khalifa University

    Apartment 3

    21 Riding House Street

    Doha, 00000

    Qatar

    Phone: 974 55708549

    Fax:974 55708549

    Email: aabdalrazaq@hbku.edu.qa


    Background: Electronic personal health records (ePHRs) are secure Web-based tools that enable individuals to access, manage, and share their medical records. England recently introduced a nationwide ePHR called Patient Online. As with ePHRs in other countries, adoption rates of Patient Online remain low. Understanding factors affecting patients’ ePHR use is important to increase adoption rates and improve the implementation success of ePHRs.

    Objective: This study aimed to examine factors associated with patients’ use of ePHRs in England.

    Methods: The unified theory of acceptance and use of technology was adapted to the use of ePHRs. To empirically examine the adapted model, a cross-sectional survey of a convenience sample was carried out in 4 general practices in West Yorkshire, England. Factors associated with the use of ePHRs were explored using structural equation modeling.

    Results: Of 800 eligible patients invited to take part in the survey, 624 (78.0%) returned a valid questionnaire. Behavioral intention (BI) was significantly influenced by performance expectancy (PE; beta=.57, P<.001), effort expectancy (EE; beta=.16, P<.001), and perceived privacy and security (PPS; beta=.24, P<.001). The path from social influence to BI was not significant (beta=.03, P=.18). Facilitating conditions (FC) and BI significantly influenced use behavior (UB; beta=.25, P<.001 and beta=.53, P<.001, respectively). PE significantly mediated the effect of EE and PPS on BI (beta=.19, P<.001 and beta=.28, P=.001, respectively). Age significantly moderated 3 paths: PE→BI, EE→BI, and FC→UB. Sex significantly moderated only the relationship between PE and BI. A total of 2 paths were significantly moderated by education and internet access: EE→BI and FC→UB. Income moderated the relationship between FC and UB. The adapted model accounted for 51% of the variance in PE, 76% of the variance in BI, and 48% of the variance in UB.

    Conclusions: This study identified the main factors that affect patients’ use of ePHRs in England, which should be taken into account for the successful implementation of these systems. For example, developers of ePHRs should involve patients in the process of designing the system to consider functions and features that fit patients’ preferences and skills to ensure systems are useful and easy to use. The proposed model accounted for 48% of the variance in UB, indicating the existence of other, as yet unidentified, factors that influence the adoption of ePHRs. Future studies should confirm the effect of the factors included in this model and identify additional factors.

    J Med Internet Res 2019;21(7):e12373

    doi:10.2196/12373

    Keywords

    health records, personal; patient portal; electronic personal health records; technology acceptance; technology adoption; intention; unified theory of acceptance and use of technology; structural equation modelling 


    Background

    Electronic personal health records (ePHRs) refer to secure Web-based tools that enable individuals to access and manage their medical records and share them with trusted others [1]. More advanced ePHRs provide additional functionalities, such as scheduling appointments, requesting prescription refills, messaging providers, requesting referrals, and educational tools [2-4]. Benefits of using ePHRs include the following: enhancing patient empowerment [5,6], improving patient self-management and medication adherence [7,8], enhancing the relationships and communications between patients and health care providers [9,10], enabling patients to easily access health services [11,12], avoiding duplicated tests [9,11], and reducing adverse drug interactions and allergies [9,11,13].

    In 2015, the National Health Service in England launched a program called Patient Online, which requires general practices (GPs) to provide patients with Web-based services, such as booking appointments, requesting prescription refills, and viewing summary information from GP records [14,15]. GPs use one of the following systems to provide their patients with the abovementioned services: SystemOnline, Patient Access, Patient Services, The Waiting Room, Engage Consult, and Evergreen Life or i-Patient [14].

    Research Problem and Aim

    The overall adoption rate of Patient Online was 18.9% in April 2017 and reached 24.4% in April 2018 [16], and so adoption remains low. Identifying and understanding factors that affect patients’ use of ePHRs is crucial to develop interventions to increase patients’ adoption and improve the implementation success of ePHRs [17-22]. According to a systematic review conducted by Abd-alrazaq and colleagues [23], there are no published studies on factors affecting patients’ use of ePHRs in England. Although many studies have been conducted in other countries, they have several shortcomings, namely, (1) few studies were theory-based research [21,24-27], (2) many studies focused on factors that affect patients’ intention to use ePHRs instead of actual use [28-32], (3) many studies have assessed the factors that affect self-reported use rather than actual use [27,32-35], (4) almost all studies examined independent and dependent variables at one point in time using the same data collection instrument, so being at risk of common method bias [25,32,36], and (5) almost all studies did not differentiate between factors affecting initial use and continuing use of ePHRs.

    This study aimed to examine factors associated with patients’ adoption of ePHRs (Patient Online) in England. As 76% of patients in England have never used Patient Online [16], the study focused on factors associated with patients’ initial use of ePHRs. Therefore, it was more appropriate to investigate the factors that make nonusers become users (ie, initial use stage).


    Theoretical Foundation

    In total, 12 theories and models originated from various disciplines, such as psychology, sociology, and information systems, were reviewed to select the appropriate one for our study. Selection of the appropriate theory was based on predefined 6 criteria. Although 2 criteria were related to the applicability of the theory on the phenomena of interest (ie, population and type of behavior), the remaining 4 were related to goodness of the theory (ie, logical consistency, explanatory power, falsifiability, and parsimony). The unified theory of acceptance and use of technology (UTAUT) was the only theory that met all those criteria. Therefore, this study chose UTAUT as a theoretical lens to examine factors associated with patients’ use of ePHRs. More details about how the theories met or did not meet each criterion are explained in Multimedia Appendix 1.

    According to UTAUT, behavioral intention (BI) is affected directly by performance expectancy (PE), effort expectancy (EE), and social influence (SI) [37]. Both BI and facilitating conditions (FC) are hypothesized to affect use behavior (UB) directly [37]. UTAUT also proposes that most of these relationships are moderated by age, sex, experience, and voluntariness [37].

    In this study, the adoption of ePHRs is not compulsory. The UTAUT construct of voluntariness is only applicable in nonvoluntary contexts [38]. Thus, for this study, the moderator voluntariness was dropped from the model. This study focused on the factors that explained how nonusers become users of ePHRs (ie, preusage stage); the sample comprised only nonusers of ePHRs (ie, having no experience). For that reason, the moderator experience was also removed from the model.

    A review of the literature identified a consensus on the influential effect of the following factors on ePHRs adoption: PPS [26,39-48], internet access [11,28,39,49-53], income [26,28,39,49,51,54-58], and education level [26,28,39,44,49,51,56,59-63]. These 4 factors were not part of UTAUT but were included in our adapted model to make it more appropriate for the context of ePHRs adoption. Although PPS was proposed as an independent variable, the remaining 3 factors were hypothesized as a moderator. The research hypotheses and the proposed model are presented in Table 1 and Figure 1, respectively. Multimedia Appendix 2 shows the conceptual definitions of the constructs in the proposed model. Multimedia Appendix 3 shows the theoretical foundations for the new proposed relationships that were added to the UTAUT model.

    Table 1. The research hypotheses.
    Ha numberHypothesis
    H1PEb positively influences patients’ intention to use Patient Online.
    H2Age, sex, education, and income moderate the positive relationship between PE and patients’ intention to use Patient Online, such that the influence is stronger for younger males with lower level of education and higher income.
    H3EEc positively influences patients’ intention to use Patient Online.
    H4PE positively mediates the positive relationship between EE and BId.
    H5Age, sex, education, income, and internet access moderate the positive relationship between EE and patients’ intention to use Patient Online, such that the influence is stronger for older females with lower level of education and income and without internet access.
    H6SIe positively influences patients’ intention to use Patient Online.
    H7Age and sex moderate the positive relationship between SI and patients’ intention to use Patient Online, such that the influence is stronger for older females.
    H8PPSf positively influences patients’ intention to use Patient Online.
    H9PE positively mediates the positive relationship between PPS and BI.
    H10Age, sex, education, and income moderate the positive relationship between PPS and patients’ intention to use Patient Online, such that the influence is stronger for older females with higher level of education and lower income.
    H11FCg positively influences patients’ use of Patient Online.
    H12Age, sex, education, income, and internet access moderate the positive relationship between FC and UBh, such that the influence is stronger for older females with a lower level of education and income and without internet access.
    H13BI positively influences patients’ use of Patient Online.

    aH: hypothesis.

    bPE: performance expectancy.

    cEE: effort expectancy.

    dBI: behavioral intention.

    eSI: social influence.

    fPPS: perceived privacy and security.

    gFC: facilitating conditions.

    hUB: use behavior.

    Study Design and Setting

    The proposed model was examined empirically using data from a cross-sectional survey. The survey was conducted at 4 West Yorkshire (England) GPs, 3 practices in Bradford and 1 in Leeds. More details about these practices are shown in Multimedia Appendix 4. Health Research Authority approval for this study was granted before starting data collection (REC reference: 17/SC/0323).

    ‎
    Figure 1. The proposed model.
    View this figure

    Measurement

    Self-administrated questionnaires were used to measure all variables proposed in the model except UB. UB was measured objectively using system logs that recorded the use of PatientOnline. Questionnaires included 29 well-validated items adopted from previous studies (Multimedia Appendix 5). An introduction about Patient Online was included at the top of the questionnaire to ensure all participants had the knowledge necessary to answer questions about Patient Online. The questionnaire was validated by sending it to a panel of experts to assess the face validity and content validity of the questions. After modifying the questionnaire according to experts’ recommendations, it was pilot tested by sending it via email to 37 patients (members of patient and carer community) who were asked to fill in the questionnaire and answer questions regarding clarity or ambiguity of questions, clarity of instructions to answer questions, difficulty to answer questions, time needed to complete the questionnaire, clarity and attractiveness of the layout, missing of important topics, and sequence of questions. A few issues were reported by experts and patients, and the questionnaire was modified accordingly (Multimedia Appendix 6). System log data of the number of times that each participant logged into the system during 6 months after completing the questionnaire were the objective measure of use.

    Recruitment

    We recruited a convenience sample of patients from August 21, 2017, to September 26, 2017. Patients were eligible to participate if they (1) lived in England and were registered at 1 of the 4 GPs, (2) were aged 18 years or older, and (3) had not used Patient Online before (nonusers). The researcher distributed the questionnaire to eligible participants visiting 1 of the 4 GPs. After 6 months from the completion of the questionnaire, data from the system log were extracted to ascertain participants’ use of Patient Online.

    Statistical Analysis

    Structural equation modeling (SEM) was used to test the theoretical model and hypotheses. Specifically, the measurement model was examined in terms of 3 aspects: model fit, construct reliability, and construct validity [64,65]. After ensuring the validity of the measurement model, the structural model was tested in terms of 3 aspects: model fit, predictive power, and strength of relationships [65-67]. The strength of relationships was examined using different methods depending on the type of the proposed effect. Specifically, direct effects were assessed by checking path coefficients [68]. Mediating effects were examined by assessing the indirect effect using bootstrapping. The moderating effect for the metric moderator (ie, age) was examined using the interaction effect method [64,69]. The moderating effects for nonmetric moderators were tested using multigroup SEM [64,69,70]. Analysis of moment structures (version 24; IBM SPSS) software was used for conducting all abovementioned analyses.


    Participants’ Characteristics

    Of the 800 eligible patients invited to take part in the survey, 624 participants returned a completed questionnaire giving a response rate of 78%. The mean age of participants was 44.2 years. The majority of participants were white (79.8%, 498/624) and had internet access (84.6%, 528/624; Table 2). Differences between participants and nonparticipants in terms of age, sex, and ethnicity were not significant (P=.21, P=.06, and P=.64, respectively). It was, therefore, concluded that the risk of nonresponse bias was minimal.

    Table 2. Participants’ characteristics (n=624).
    VariablesRespondents, n (%)
    Age (years)a

    		18-24107 (17.1)

    		25-34148 (23.7)

    		35-44116 (18.6)

    		45-5498 (15.7)

    		55-6465 (10.4)

    		65-7446 (7.4)

    		75 and older44 (7.1)
    Sex

    		Male293 (46.9)

    		Female331 (53.1)
    Ethnicity

    		White498 (79.8)

    		Asian73 (11.7)

    		Black20 (3.2)

    		Mixed26 (4.1)

    		Others7 (1.2)
    Income (£)

    		<20,000284 (45.5)

    		20,000-29,99980 (12.8)

    		30,000-39,99965 (10.4)

    		40,000-49,99943 (7.0)

    		50,000-59,99926 (4.2)

    		≥60,00012 (1.9)

    		Prefer not to say114 (18.2)
    Education

    		Up to secondary school69 (11.1)

    		Secondary school147 (23.6)

    		College/diploma165 (26.4)

    		Bachelor’s degree174 (27.9)

    		Master’s degree47 (7.5)

    		Doctoral degree22 (3.5)
    Internet access

    		Yes528 (84.6)

    		No96 (15.4)

    aMean 44.2 (SD 18.9).

    Measurement Model

    Model Fit

    All fit indices indicated a good fit of the initial model except the following 3 indices: goodness-of-fit index (GFI, 0.923), root mean square error of approximation (RMSEA, 0.053), and standardized root mean square residual (SRMR, 0.057; Table 3). The following 4 items were identified as a source of the poor fit of the measurement model as their factor loading was less than 0.70: FC 4, FC 5, PPS 3, and PPS 5. After deleting these 4 items from the model, all fit indices of the modified model improved and existed within their acceptable levels, indicating a good fit (Table 3).

    Table 3. Results of fit indices of the initial and modified measurement model.
    Fit indicesCutoff pointInitial measurement modelModified measurement model
    Relative chi-square (df)1-32.8 (215)1.4 (137)
    GFIa≥0.950.9230.969
    AGFIb≥0.900.9020.957
    RMSEAc<0.050.0530.026
    PCLOSEd≥0.050.1941.000
    SRMRe≤0.050.0570.017
    NFIf≥0.950.9640.988
    CFIg≥0.950.9770.995
    TLIh≥0.950.9720.996

    aGFI: goodness-of-fit index.

    bAGFI: adjusted goodness-of-fit index.

    cRMSEA: root mean square error of approximation.

    dPCLOSE: p of close fit.

    eSRMR: standardized root mean square residual.

    fNFI: normed fit index.

    gCFI: comparative fit index.

    hTLI: Tucker-Lewis index.

    Construct Reliability

    Results for the modified model of Cronbach alpha, composite reliability, and average variance extracted (AVE) for each construct were within their cutoff of ≥.70, ≥0.70, and ≥0.50, respectively (Multimedia Appendix 7). This indicates that the measurement items are consistent and reproducible in measuring what it is assumed to measure.

    Construct Validity

    The values of factor loading and AVE for all items considerably exceeded the thresholds of 0.70 and 0.50, respectively (Multimedia Appendix 8). These results indicate that items had good convergent validity. Similarly, items showed good discriminant validity according to 3 measures. Specifically, intercorrelation coefficients are located within the acceptable ranges (<0.85; Multimedia Appendix 9). With regard to the second measure, each value of square root of AVE for a construct (values on the diagonal) was higher than all intercorrelation coefficients between that construct and each other construct (off-diagonal values; Multimedia Appendix 9). With regard to the third measure, the loading of each item on its construct was higher than cross-loadings in rows and columns (Multimedia Appendix 10).

    Structural Model

    Model Fit and Predictive Power

    All fit indices of the structural model indices were within their cutoff levels, indicating a good model fit (Table 4). The structural model accounted for 51% of the variance in PE, 76% of the variance in BI, and 48% of the variance in UB (Figure 2).

    Table 4. Results of fit indices of the structural model.
    Fit indicesCutoff pointFitness of the structural model
    Relative chi-square (df)1-31.6 (157)
    GFIa≥0.950.962
    AGFIb≥0.900.949
    RMSEAc<0.050.032
    PCLOSEd≥0.051.000
    SRMRe≤0.050.036
    NFIf≥0.950.984
    CFIg≥0.950.993
    TLIh≥0.950.992

    aGFI: goodness-of-fit index.

    bAGFI: adjusted goodness-of-fit index.

    cRMSEA: root mean square error of approximation.

    dPCLOSE: p of close fit.

    eSRMR: standardized root mean square residual.

    fNFI: normed fit index.

    gCFI: comparative fit index.

    hTLI: Tucker-Lewis index.

    ‎
    Figure 2. Structural model estimates.
    View this figure
    Strength of Relationships

    Of the direct effects, BI was associated with PE (beta=.57), EE (beta=.16), and PPS (beta=.24; Table 5). The path from SI to BI was not significant (beta=.03, P=.18). Both FC and BI were significantly associated with UB (beta=.25 and beta=.53, respectively). Therefore, the following hypotheses were supported: H1, H3, H8, H11, and H13 (Table 5).

    With regard to mediating effects, results of bootstrapping indicate that PE mediated significantly the effect of EE and PPS on BI (beta=.20 and beta=.28, respectively; Table 6). Accordingly, H4 and H9 were supported in this research.

    With regard to moderating effects, age moderated significantly 3 paths: PE→BI (beta=−.10), EE→BI (beta=.06), and FC→UB (beta=.16; Table 7). Sex moderated significantly only the relationship between PE and BI (P=.009; Table 8). In relation to moderating effect of education, the relationship between FC and UB was statistically stronger for secondary school or lower group than college group (beta=.39 vs beta=.30, P=.003; Table 9) and than bachelor or higher group (beta=.39 vs beta=.21; Table 10). The path from EE to BI was statistically weaker for bachelor or higher group than secondary school or lower group (beta=.01 vs beta=.14; Table 10) and than college group (beta=.01 vs beta=.13; Table 11). As shown in Table 12-14, the relationship between FC and UB was statistically stronger for patients with low income (beta=.43) than patients with moderate or high income (beta=.25 and beta=.10, respectively). Internet access moderated significantly 2 paths EE→BI (P=.01) and FC→UB (P<.001; Table 15). Accordingly, H10 was rejected, whereas the following hypotheses were partially supported: H2, H5, H7, and H12.

    Table 5. Results of direct effects.
    HaPathSE (beta)95% CIP valueSupported?
    H1PEb→BIc.570.51 to 0.64<.001Yes
    H3EEd→BI.160.11 to 0.21<.001Yes
    H6SIe→BI.03−0.03 to 0.10.18No
    H8PPSf→BI.240.18 to 0.29<.001Yes
    H11FCg→UBh.250.20 to 0.30<.001Yes
    H13BI →UB.530.48 to 0.58<.001Yes

    aH: hypothesis.

    bPE: performance expectancy.

    cBI: behavioral intention.

    dEE: effort expectancy.

    eSI: social influence.

    fPPS: perceived privacy and security.

    gFC: facilitating conditions.

    hUB: use behavior.

    Table 6. Results of mediating effects.
    HaIndirect effectEstimate (beta)95% CIP valueSupported?
    H4EEb→PEc→BId.200.15-0.25<.001Yes
    H9PPSe→PE→BI.280.23-0.33<.001Yes

    aH: hypothesis.

    bEE: effort expectancy.

    cPE: performance expectancy.

    dBI: behavioral intention.

    ePPS: perceived privacy and security.

    Table 7. Results of moderating effect of age.
    Interaction effectStandardized estimate (beta)P value
    PEa×Age→BIb−.10<.001
    EEc×Age→BI.06.03
    SId×Age→BI.01.06
    PPSe×Age→BI−.03.22
    FCf×Age→UBg.16<.001

    aPE: performance expectancy.

    bBI: behavioral intention.

    cEE: effort expectancy.

    dSI: social influence.

    ePPS: perceived privacy and security.

    fFC: facilitating conditions.

    gUB: use behavior.

    Table 8. Results of moderating effect of sex.
    Hypothesized pathSE (beta)P valueChi-square difference test, P value

    		MaleFemaleMaleFemale
    PEa→BIb.59.51<.001<.001.01
    EEc→BI.17.19<.001<.001.32
    SId→BI−.03.06.53.06.07
    PPSe→UB.27.20<.001<.001.65
    FCf→UBg.35.28<.001<.001.32

    aPE: performance expectancy.

    bBI: behavioral intention.

    cEE: effort expectancy.

    dSI: social influence.

    ePPS: perceived privacy and security.

    fFC: facilitating conditions.

    gUB: use behavior.

    Table 9. Results of moderating effect of education level (secondary school versus college/diploma).
    Hypothesized pathSecondary school or lowerCollege/diplomaChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEa→BIb.57<.001.62<.001.38
    EEc→BI.14.02.13.003.38
    PPSd→BI.17.005.29<.001.22
    FCe→UBf.39<.001.30<.001.003

    aPE: performance expectancy.

    bBI: behavioral intention.

    cEE: effort expectancy.

    dPPS: perceived privacy and security.

    eFC: facilitating conditions.

    fUB: use behavior.

    Table 10. Results of moderating effect of education level (secondary school versus bachelor or higher).
    Hypothesized pathSecondary school or lowerBachelor or higherChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEa→BIb.57<.001.57<.001.50
    EEc→BI.14.02.01.16.03
    PPSd→BI.17.005.24<.001.14
    FCe→UBf.39<.001.21<.001.02

    aPE: performance expectancy.

    bBI: behavioral intention.

    cEE: effort expectancy.

    dPPS: perceived privacy and security.

    eFC: facilitating conditions.

    fUB: use behavior.

    Table 11. Results of moderating effect of education level (college/diploma versus bachelor or higher).
    Hypothesized pathCollege/diplomaBachelor or higherChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEa→BIb.62<.001.57<.001.08
    EEc→BI.13.003.01.16.008
    PPSd→BI.29<.001.24<.001.54
    FCe→UBf.30<.001.21<.001.23

    aPE: performance expectancy.

    bBI: behavioral intention.

    cEE: effort expectancy.

    dPPS: perceived privacy and security.

    eFC: facilitating conditions.

    fUB: use behavior.

    Table 12. Results of moderating effect of income (low income versus middle income).
    Hypothesized pathLow incomeaMiddle incomebChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEc→BId.54<.001.52<.001.43
    EEe→BI.14<.001.22<.001.39
    PPSf→BI.26<.001.28<.001.99
    FCg→UBh.43<.001.25<.001.048

    aLow income: >£20,000.

    bMedium income: £20,000-39,999.

    cPE: performance expectancy.

    dBI: behavioral intention.

    eEE: effort expectancy.

    fPPS: perceived privacy and security.

    gFC: facilitating conditions.

    hUB: use behavior.

    Table 13. Results of moderating effect of income (low income versus high income).
    Hypothesized pathLow incomeaHigh incomebChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEc→BId.54<.001.68<.001.09
    EEe→BI.14<.001.12.048.67
    PPSf→BI.26<.001.25<.001.87
    FCg→UBh.43<.001.10.03.03

    aLow income: >£20,000.

    bHigh income: ≥£40,000.

    cPE: performance expectancy.

    dBI: behavioral intention.

    eEE: effort expectancy.

    fPPS: perceived privacy and security.

    gFC: facilitating conditions.

    hUB: use behavior.

    Table 14. Results of moderating effect of income (middle income versus high income).
    Hypothesized pathMiddle incomeaHigh incomebChi-square difference test, P value

    		SE (beta)P valueSE (beta)P value
    PEc→BId.52<.001.68<.001.06
    EEe→BI.22<.001.12.048.27
    PPSf→BI.28<.001.25<.001.88
    FCg→UBh.25<.001.10.03.02

    aMedium income: £20,000-39,999.

    bHigh income: ≥£40,000.

    cPE: performance expectancy.

    dBI: behavioral intention.

    eEE: effort expectancy.

    fPPS: perceived privacy and security.

    gFC: facilitating conditions.

    hUB: use behavior.

    Table 15. Results of the moderating effect of internet access.
    Hypothesized pathSE (beta)P valueChi-square difference test, P value

    		Internet accessNo internet accessInternet accessNo internet access
    EEa→BIb.12.28<.001.005.01
    FCc→UBd.18.44<.001<.001<.001

    aEE: effort expectancy.

    bBI: behavioral intention.

    cFC: facilitating conditions.

    dUB: use behavior.


    Principal Findings

    This study found that PE was positively associated with BI. This means that patients are more likely to intend to use Patient Online when they perceive it to be very useful and advantageous. This relationship is consistent with other studies investigating the uptake of ePHRs [18,26,71,72-75]. Our results suggest that this relationship is stronger for younger males, meaning that younger males who perceive the system as more useful are more likely to intend to use it. This study is one of the very few studies that successfully assessed the moderating effect of age and sex in explaining the use of ePHRs. A similar moderating effect has been demonstrated for the use of consumer health information technologies (CHITs) [76,77].

    These results showed that EE was positively associated with BI, that is, patients are more likely to intend to use Patient Online when they perceive it as an easy-to-use system. This finding is consistent with studies investigating the use of ePHRs outside of England [26,72,78-80]. This study showed that the relationship between EE and BI was mediated by PE. So, patients who perceive Patient Online as easy to use are more likely to perceive it as a useful system, thereby, they are more likely to intend to use it. This finding is in line with findings of 2 CHIT studies [22,81]. Furthermore, our results showed that the relationship between EE and BI was stronger among older patients with lower level of education and without internet access. The moderating effect of age has also been found in studies investigating the use of CHITs [76]. Ours is the first study to examine the moderating effect of education and internet access to explain the use of ePHRs or CHITs.

    We found that SI and BI were not statistically associated. This means that opinions and beliefs of people who are important to the patient do not affect their intention to utilize ePHRs. This nonsignificant relationship could be attributed to the use of Patient Online being voluntary. The literature suggests that the effect of SI is significant only in contexts where using the technology is mandatory [37,82-84]. The presence of PE in a model may weaken the direct effect of SI on BI [85,86] as SI affects BI indirectly through PE [22,84]. The nonsignificant effect of SI may also be attributed to the fact that the questionnaire measures perceptions of SI rather than actual SI.

    This study demonstrated that PPS was positively associated with BI, that is, patients are more likely to intend to use Patient Online when they perceive that it is secure and will maintain their privacy. This relationship is documented elsewhere in the literature [30,45,46,48,87,88]. This study showed that the relationship between PPS and BI was mediated by PE, that is, patients who perceive Patient Online to be secure and able to maintain their privacy are more likely to perceive it as a useful system and, therefore, are more likely to intend to use it. Although several studies in the context of ePHRs and CHITs examined the direct influence of PPS on PE and on BI, no previous study has tested the indirect effect of PPS on BI through PE.

    The statistical analysis showed that FC was positively associated with UB. This means that patients are more likely to use Patient Online when they feel that they have the resources and knowledge enough to use it. This effect of FC was supported by several studies in the context of ePHRs [43,44,89]. In our study, this relationship was stronger for older patients with lower level of education and income and without internet access. In other words, these groups of people tend to place more importance on the presence of sufficient support and resources to use Patient Online. Although the moderating effect of age was supported in 1 CHITs study [76], this is the first study to investigate the moderating effects of education, income, and internet access in the context of ePHRs.

    This study showed that BI positively associated with UB, that is, patients are more likely to use Patient Online when they intend to use it. This finding is consistent with findings of several studies in the context of ePHRs and CHITs [22,80,90-93].

    Overall, the model accounted for 48% of the variance in UB. This moderate predictive power of the model indicates that there are other factors yet to be identified, which would account for the unexplained variance. Although the predictive power of the proposed model is comparable with the predictive power of the original UTAUT model (48%), it is higher than the predictive power of models proposed by other studies in the context of ePHRs: Hsieh [90] (42.7%) and Tavares and Oliveira [80] (26.8%).

    Theoretical and Practical Contributions

    This is the first theory-based study to examine factors associated with patients’ use of ePHRs in England. Very few studies have utilized theories or models to understand the factors that impact patients’ use of ePHRs [21,24-27]. Furthermore, UTAUT was not employed in those few studies. Accordingly, this study contributes to the ePHR literature by adopting and testing UTAUT in the context of ePHRs, which can be used by the future studies in the context of ePHRs and CHITs.

    This research and a study conducted by Tavares and Oliveira [80] are the only studies in the area of ePHRs that included both BI and UB in 1 model, and this is the best practice to study technology adoption [20,25]. Furthermore, our study contributes to the existing ePHR literature by being the first theory-based study to measure the UB objectively. In addition, this study is one of the few theory-based studies in the context of ePHRs that endeavored to minimize the common method bias by ensuring a gap in time between the main dependent variable (ie, use of Patient Online) and other variables. This research is the first study to shed light on the important role of moderators and mediators that explain the use of ePHRs, and this extends our understanding of factors that affect the adoption [27].

    With respect to practical contributions, we have identified that PE and EE play a crucial role in forming patients’ intention to use Patient Online. Accordingly, developers should involve patients in the process of designing the system to consider functions and features that fit patients’ preferences and skills. Developers should pilot test the system with potential users before implementation [80,94]. As PPS is an influential predictor, developers should keep patient records as private as possible by protecting the platforms using security measures, such as strong firewalls, complex and long passwords, regular security reviews, and regular website updates.

    To ensure that patients perceive the system as useful, easy to use, and secure, marketers should conduct promotional campaigns about functions and features of the system, its advantages, its ease of use, availability of different sources to support the use of the system, the security measures, the laws and regulations protecting patient privacy, and how patients can use it safely. As face-to-face communication may be one of the most effective channels in marketing to persuade potential adopters to adopt an innovation [95-97], physicians, nurses, and receptionists can play an important role in improving the publicity of Patient Online by informing patients about it in their communications. Marketers should focus more on younger males when conducting promotional campaigns regarding the benefits of the system, whereas they should concentrate more on older and less educated patients without internet access when initiating advertising campaigns regarding the ease of use of the system.

    Patients who believed that organizational and technical infrastructure existed to support the use of Patient Online were more likely to use it. Therefore, to raise awareness of the infrastructure available, GPs could provide patients with manuals, Web-based assistance, technical support, and practical training sessions. This strategy is likely to be most effective with older patients and those with lower level of education and income.

    Allowing patients to try a beta version of ePHRs could create a positive personal experience that may enhance their perceptions of usefulness and the ease of use of the system [26,75,78]. Thus, GPs should assist patients in using a beta version of Patient Online through a computer in a waiting room.

    Research Limitations

    This study has limitations that need to be considered. Data were collected from 4 GPs, all implementing the same ePHRs (ie, SystemOnline); therefore, the findings of this study may not be applicable to other ePHRs (ie, Patient Access and Patient Services). However, the findings may still be generalizable to other systems because all the systems mentioned provide the same services to the patients, and all participants had not used any of them before. Therefore, they would be unlikely to have different perceptions about the different systems.

    This study focused on assessing factors that affect patients’ initial use of ePHRs; therefore, the findings are not generalizable to the context of continuing use. This research focuses on initial use of ePHRs because Patient Online is still a new system in England and has a low adoption rate; therefore, it is better to focus on the initial use in this period. Furthermore, as this study is cross-sectional, the associations identified with patients’ initial use of ePHRs do not imply causality, and so further longitudinal research is required.

    This research is subject to a sampling bias because of using convenience sampling technique to recruit the participants [36,98]. This study found that there was no statistically significant difference between the participants and the nonparticipants in terms of age, sex, and ethnicity. Consequently, it can be said that the findings are generalizable to practices similar to the 4 practices in this study.

    It might have been appropriate to control for the effects from practices within the SEM. However, we found no differences in the demographics between the practices, indicating no evidence for practice level clustering. As these individual level factors are already included in the SEM, including clustering terms could lead to potential over adjustment.

    Recommendations for Future Research

    Further studies are required to examine the applicability of the adapted model to other contexts. For example, research could investigate the applicability of the model to other providers of Patient Online (eg, Patient Access), specific platform (eg, mobiles, tablets, and computers), other settings (eg, hospitals), and other cities or countries.

    Determining the factors that may influence the continuing use is important because long-term viability and eventual success of information technology depend on continued use [32,99-101]. Therefore, further primary studies and systematic reviews should be carried out to assess factors that affect the continuing use of ePHRs.

    Further research is needed to explain the nonsignificant effect of SI demonstrated in this study. Previous studies demonstrated that the effect of SI depended on the type of processes of SI that people considered in their decisions (internalization, identification, and compliance) [37,84,102]. Thus, researchers may consider these 3 types of processes when assessing the effect of SI. Furthermore, researchers should develop new measures to assess the actual SI, such as the number of times a patient has been informed about the system by doctors, receptionists, friends, leaflets, posters, videos, and/or automated messages.

    Although this study examined the effect of 4 moderators on most of the direct relationships, it did not examine their effects on the 2 indirect relationships (ie, EE→PE→BI and PPS→PE→BI). The effect of moderators on indirect relationships is called moderated mediation or conditional indirect effect [103-105]. To the best of our knowledge, the moderated mediating effect has not been examined in the context of ePHRs or CHITs. For this reason, future studies are required to test such an effect.

    Finally, to increase the predictive power of the proposed model, future studies should consider adding other factors to the proposed model, such as patients’ satisfaction, patient activation level, health status, perceived severity, perceived susceptibility, awareness of Patient Online, compatibility, and results demonstrability.

    Conclusions

    This study examined the main factors that affected patients’ use of ePHRs in England. The proposed model accounted for 48% of the variance in UB, indicating the existence of other, as yet unidentified, factors that influence adoption of ePHRs. Future studies should confirm the effect of the factors included in this model and identify additional factors. This study suggests that adoption rates are affected by key factors that should be taken into account for the successful implementation of ePHRs. For example, developers of ePHRs should involve patients in the process of designing the system to consider functions and features that fit patients’ preferences and skills, thereby, create a useful and easy to use system.

    Acknowledgments

    The authors would like to thank Dr Hamish Fraser for his help in the initial stages of the study, especially for his contribution to the development of the study design and setting up of links with practices.

    Conflicts of Interest

    None declared.

    ‎

    Multimedia Appendix 1

    Selection of theory.

    DOCX File, 69KB
    ‎

    Multimedia Appendix 2

    Conceptual definitions of constructs.

    DOCX File, 15KB
    ‎

    Multimedia Appendix 3

    Theoretical foundation.

    DOCX File, 79KB
    ‎

    Multimedia Appendix 4

    Characteristics of the GP practices.

    DOCX File, 16KB
    ‎

    Multimedia Appendix 5

    Measures of constructs.

    DOCX File, 18KB
    ‎

    Multimedia Appendix 6

    Questionnaire.

    DOCX File, 70KB
    ‎

    Multimedia Appendix 7

    Results of construct reliability.

    DOCX File, 12KB
    ‎

    Multimedia Appendix 8

    Results of convergent validity.

    DOCX File, 13KB
    ‎

    Multimedia Appendix 9

    Intercorrelation coefficients and squared roots of AVE.

    DOCX File, 14KB
    ‎

    Multimedia Appendix 10

    Item loadings and cross-loadings.

    DOCX File, 18KB
    1. Markle Foundation. 2003. The Personal Health Working Group: Final Report   URL: https://www.markle.org/sites/default/files/final_phwg_report1.pdf [accessed 2019-06-03]
    2. Nazi KM, Hogan TP, McInnes DK, Woods SS, Graham G. Evaluating patient access to electronic health records: results from a survey of veterans. Med Care 2013 Mar;51(3 Suppl 1):S52-S56. [CrossRef] [Medline]
    3. Pai HH, Lau F, Barnett J, Jones S. Meeting the health information needs of prostate cancer patients using personal health records. Curr Oncol 2013 Dec;20(6):e561-e569 [FREE Full text] [CrossRef] [Medline]
    4. Tulu B, Trapp AC, Strong DM, Johnson SA, Hoque M, Trudel J, et al. An analysis of patient portal utilization: what can we learn about online patient behavior by examining portal click data? Health Syst 2017 Dec 19;5(1):66-79. [CrossRef]
    5. Laranjo L, Rodolfo I, Pereira AM, de Sá AB. Characteristics of innovators adopting a national personal health record in Portugal: cross-sectional study. JMIR Med Inform 2017 Oct 11;5(4):e37 [FREE Full text] [CrossRef] [Medline]
    6. Ose D, Kunz A, Pohlmann S, Hofmann H, Qreini M, Krisam J, et al. A personal electronic health record: study protocol of a feasibility study on implementation in a real-world health care setting. JMIR Res Protoc 2017 Mar 2;6(3):e33 [FREE Full text] [CrossRef] [Medline]
    7. Rice RC. CiteWeb. 2014. Healthcare Leaders' Lived Experiences Regarding the Implementation of Electronic Personal Health Records   URL: http://citeweb.info/20141288190
    8. Vermeir P, Degroote S, Vandijck D, van Tiggelen H, Peleman R, Verhaeghe R, et al. The patient perspective on the effects of medical record accessibility: a systematic review. Acta Clin Belg 2017 Jun;72(3):186-194. [CrossRef] [Medline]
    9. Alyami MA, Song YT. Removing Barriers in Using Personal Health Record Systems. In: Proceedings of the 15th International Conference on Computer and Information Science. 2016 Presented at: IEEE'16; June 26-29, 2016; Okayama, Japan. [CrossRef]
    10. Ochoa 3rd A, Kitayama K, Uijtdehaage S, Vermillion M, Eaton M, Carpio F, et al. Patient and provider perspectives on the potential value and use of a bilingual online patient portal in a Spanish-speaking safety-net population. J Am Med Inform Assoc 2017 Nov 1;24(6):1160-1164. [CrossRef] [Medline]
    11. Morton AA. Examining Acceptance of an Integrated Personal Health Record (PHR). Baltimore: University of Maryland; 2011.
    12. Pagliari C, Detmer D, Singleton P. The Nuffield Trust. 2007. Electronic Personal Health Records: Emergence and Implications for the UK   URL: https://www.nuffieldtrust.org.uk/files/2017-01/electronic-personal-health-records-uk-web-final.pdf [accessed 2019-06-10]
    13. Pagliari C, Detmer D, Singleton P. Potential of electronic personal health records. Br Med J 2007 Aug 18;335(7615):330-333 [FREE Full text] [CrossRef] [Medline]
    14. National Health Service. 2016. Start Using GP Online Services   URL: https://www.nhs.uk/using-the-nhs/nhs-services/gps/gp-online-services/ [accessed 2019-06-10]
    15. National Health Service England. 2017. About GP Online Services   URL: https://www.england.nhs.uk/gp-online-services/about-the-prog/ [accessed 2019-06-10]
    16. Power BI. 2018. Patient Online Management Information (POMI): April 2017-May 2019   URL: https://tinyurl.com/ybptnhsu [accessed 2019-06-10]
    17. Fung V, Ortiz E, Huang J, Fireman B, Miller R, Selby JV, et al. Early experiences with e-health services (1999-2002): promise, reality, and implications. Med Care 2006 May;44(5):491-496. [CrossRef] [Medline]
    18. Huygens MW, Vermeulen J, Friele RD, van Schayck OC, de Jong JD, de Witte LP. Internet services for communicating with the general practice: barely noticed and used by patients. Interact J Med Res 2015 Nov 24;4(4):e21 [FREE Full text] [CrossRef] [Medline]
    19. Kaelber DC, Jha AK, Johnston D, Middleton B, Bates DW. A research agenda for personal health records (PHRs). J Am Med Inform Assoc 2008;15(6):729-736 [FREE Full text] [CrossRef] [Medline]
    20. Logue MD, Effken JA. Modeling factors that influence personal health records adoption. Comput Inform Nurs 2012 Jul;30(7):354-362. [CrossRef] [Medline]
    21. Najaftorkaman M, Ghapanchi AH, Talaei-Khoei A. Analysis of Research in Adoption of Person-Centred Healthcare Systems: The Case of Online Personal Health Record. In: Proceedings of the 25th Australasian Conference on Information Systems. 2014 Presented at: ACIS'14; December 8-10, 2014; Auckland, New Zealand p. 8-10   URL: http:/​/openrepository.​aut.ac.nz/​bitstream/​handle/​10292/​8104/​acis20140_submission_55.​pdf?sequence=1&isAllowed=y
    22. Or CK, Karsh B, Severtson DJ, Burke LJ, Brown RL, Brennan PF. Factors affecting home care patients' acceptance of a web-based interactive self-management technology. J Am Med Inform Assoc 2011;18(1):51-59 [FREE Full text] [CrossRef] [Medline]
    23. Abd-Alrazaq AA, Bewick BM, Farragher T, Gardner P. Factors that affect the use of electronic personal health records among patients: a systematic review. Int J Med Inform 2019 Dec 21;126(2019):164-175. [CrossRef] [Medline]
    24. Andrews L, Gajanayake R, Sahama T. The Australian general public's perceptions of having a personally controlled electronic health record (PCEHR). Int J Med Inform 2014 Dec;83(12):889-900. [CrossRef] [Medline]
    25. Assadi V. MacSphere: McMaster University. 2013. Adoption of Integrated Personal Health Record Systems: A Self-determination Theory Perspective   URL: https://macsphere.mcmaster.ca/bitstream/11375/13530/1/fulltext.pdf
    26. Emani S, Yamin CK, Peters E, Karson AS, Lipsitz SR, Wald JS, et al. Patient perceptions of a personal health record: a test of the diffusion of innovation model. J Med Internet Res 2012 Nov 5;14(6):e150 [FREE Full text] [CrossRef] [Medline]
    27. Or CK, Karsh BT. A systematic review of patient acceptance of consumer health information technology. J Am Med Inform Assoc 2009;16(4):550-560 [FREE Full text] [CrossRef] [Medline]
    28. Abramson EL, Patel V, Edwards A, Kaushal R. Consumer perspectives on personal health records: a 4-community study. Am J Manag Care 2014 Apr;20(4):287-296 [FREE Full text] [Medline]
    29. Nambisan P. Factors that impact patient web portal readiness (PWPR) among the underserved. Int J Med Inform 2017 Dec;102:62-70. [CrossRef] [Medline]
    30. Ozok AA, Wu H, Gurses AP. Exploring patients’ use intention of personal health record systems: implications for design. Int J Hum-Comput Int 2017 Jan 6;33(4):265-279. [CrossRef]
    31. Razmak J, Bélanger C. Using the technology acceptance model to predict patient attitude toward personal health records in regional communities. Inform Tech People 2018 Apr 3;31(2):306-326. [CrossRef]
    32. Gebauer L, Söllner M, Leimeister JM. Towards Understanding the Formation of Continuous IT Use. In: Proceedings of the Thirty Fourth International Conference on Information Systems. 2013 Presented at: ICIS'13; December 15-18, 2013; Milano, Italy.
    33. Legris P, Ingham J, Collerette P. Why do people use information technology? A critical review of the technology acceptance model. Infor Manag 2003 Jan;40(3):191-204. [CrossRef]
    34. Rodman DM. Meaningful Use: The Utilization of Patient Portals. Buffalo: D'Youville College; 2015.
    35. Turner M, Kitchenham B, Brereton P, Charters S, Budgen D. Does the technology acceptance model predict actual use? A systematic literature review. Inf Softw Technol 2010 May;52(5):463-479 [FREE Full text] [CrossRef]
    36. Bhattacherjee A. Social Science Research: Principles, Methods, and Practices. Scotts Valley, California , US: CreateSpace Independent Publishing Platform; 2012.
    37. Venkatesh V, Morris MG, Davis GB, Davis FD. User acceptance of information technology: toward a unified view. MIS Q 2003;27(3):425-478. [CrossRef]
    38. Venkatesh V, Thong JY, Xu X. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Q 2012;36(1):157-178. [CrossRef]
    39. Ancker JS, Osorio SN, Cheriff A, Cole CL, Silver M, Kaushal R. Patient activation and use of an electronic patient portal. Inform Health Soc Care 2015;40(3):254-266. [CrossRef] [Medline]
    40. Baird A. Extending Adoption of Innovation Theory With Consumer Influence the Case of Personal Health Records (PHRS) and Patient Portals. Arizona: Arizona State University; 2012.
    41. Dontje K, Corser WD, Holzman G. Understanding patient perceptions of the electronic personal health record. J Nurse Pract 2014 Nov;10(10):824-828. [CrossRef]
    42. Goel MS, Brown TL, Williams A, Cooper AJ, Hasnain-Wynia R, Baker DW. Patient reported barriers to enrolling in a patient portal. J Am Med Inform Assoc 2011 Dec;18(Suppl 1):i8-12 [FREE Full text] [CrossRef] [Medline]
    43. Mishuris RG, Stewart M, Fix GM, Marcello T, McInnes DK, Hogan TP, et al. Barriers to patient portal access among veterans receiving home-based primary care: a qualitative study. Health Expect 2015 Dec;18(6):2296-2305 [FREE Full text] [CrossRef] [Medline]
    44. Patel VN, Dhopeshwarkar RV, Edwards A, Barrón Y, Sparenborg J, Kaushal R. Consumer support for health information exchange and personal health records: a regional health information organization survey. J Med Syst 2012 Jun;36(3):1043-1052. [CrossRef] [Medline]
    45. Rao M. Factors affecting health care technology use in baby boomers: a quantitative study. California: Northcentral University; 2014:1.
    46. Tieu L, Sarkar U, Schillinger D, Ralston JD, Ratanawongsa N, Pasick R, et al. Barriers and facilitators to online portal use among patients and caregivers in a safety net health care system: a qualitative study. J Med Internet Res 2015 Dec 3;17(12):e275 [FREE Full text] [CrossRef] [Medline]
    47. Turner AN, Osterhage K, Joe J, Hartzler A, Lin L, Demiris G. Use of patient portals: personal health information management in older adults. Stud Health Technol Inform 2015;216:978. [CrossRef] [Medline]
    48. Whetstone M, Goldsmith R. Factors influencing intention to use personal health records. Int J Pharm Healthc Mark 2009 Apr 3;3(1):8-25. [CrossRef]
    49. Cho AH, Arar NH, Edelman DE, Hartwell PH, Oddone EZ, Yancy WS. Do diabetic veterans use the internet? Self-reported usage, skills, and interest in using My HealtheVet web portal. Telemed J E Health 2010 Jun;16(5):595-602. [CrossRef] [Medline]
    50. Patel VN, Abramson E, Edwards AM, Cheung MA, Dhopeshwarkar RV, Kaushal R. Consumer attitudes toward personal health records in a beacon community. Am J Manag Care 2011 Apr;17(4):e104-e120 [FREE Full text] [Medline]
    51. Lyles CR, Harris LT, Jordan L, Grothaus L, Wehnes L, Reid RJ, et al. Patient race/ethnicity and shared medical record use among diabetes patients. Med Care 2012 May;50(5):434-440. [CrossRef] [Medline]
    52. Roblin DW, Houston TK, Allison JJ, Joski PJ, Becker ER. Disparities in use of a personal health record in a managed care organization. J Am Med Inform Assoc 2009;16(5):683-689 [FREE Full text] [CrossRef] [Medline]
    53. Ronda MC, Dijkhorst-Oei L, Gorter KJ, Beulens JW, Rutten GE. Differences between diabetes patients who are interested or not in the use of a patient web portal. Diabetes Technol Ther 2013 Jul;15(7):556-563 [FREE Full text] [CrossRef] [Medline]
    54. Noblin AM, Wan TT, Fottler M. Intention to use a personal health record: a theoretical analysis using the technology acceptance model. Int J Health Technol Manag 2013;14(1/2):73-89. [CrossRef]
    55. van der Vaart R, Drossaert CH, Taal E, van de Laar MA. Patient preferences for a hospital-based rheumatology interactive health communication application and factors associated with these preferences. Rheumatology (Oxford) 2011 Sep;50(9):1618-1626. [CrossRef] [Medline]
    56. Cahill JE, Lin L, LoBiondo-Wood G, Armstrong TS, Acquaye AA, Vera-Bolanos E, et al. Personal health records, symptoms, uncertainty, and mood in brain tumor patients. Neurooncol Pract 2014 Jun;1(2):64-70 [FREE Full text] [CrossRef] [Medline]
    57. Hibbard JH, Greene J. Who are we reaching through the patient portal: engaging the already engaged? Int J Pers Cent Med 2011;1(4):788-793. [CrossRef]
    58. Jhamb M, Cavanaugh KL, Bian A, Chen G, Ikizler TA, Unruh ML, et al. Disparities in electronic health record patient portal use in nephrology clinics. Clin J Am Soc Nephrol 2015 Nov 6;10(11):2013-2022 [FREE Full text] [CrossRef] [Medline]
    59. Kim E, Stolyar A, Lober WB, Herbaugh AL, Shinstrom SE, Zierler BK, et al. Challenges to using an electronic personal health record by a low-income elderly population. J Med Internet Res 2009 Oct 27;11(4):e44 [FREE Full text] [CrossRef] [Medline]
    60. Laugesen JD. MacSphere: McMaster University. 2013. Adoption of Electronic Personal Health Records by Chronic Disease Patients: Integrating Protection Motivation Theory and Task-Technology Fit   URL: https://macsphere.mcmaster.ca/bitstream/11375/13343/1/fulltext.pdf
    61. Davis SE, Osborn CY, Kripalani S, Goggins KM, Jackson GP. Health literacy, education levels, and patient portal usage during hospitalizations. AMIA Annu Symp Proc 2015;2015:1871-1880 [FREE Full text] [Medline]
    62. Martinez M, Baum A, Gomez SA, Gomez A, Luna D, González BF. Predictive variables of the use of personal health record: the hospital Italiano de Buenos Aires study. Stud Health Technol Inform 2013;192:1171. [CrossRef] [Medline]
    63. Mikles SP, Mielenz TJ. Characteristics of electronic patient-provider messaging system utilisation in an urban health care organisation. J Innov Health Inform 2014 Dec 18;22(1):214-221 [FREE Full text] [CrossRef] [Medline]
    64. Hair JF, Black WC, Babin BJ, Anderson RE. Multivariate Data Analysis: A Global Perspective. Seventh Edition. Essex, UK: Pearson Education; 2010.
    65. Kline RB. Principles and Practice of Structural Equation Modeling. Fourth Edition. New York, USA: Guilford Press; 2015.
    66. Byrne BM. Structural Equation Modeling With AMOS: Basic Concepts, Applications, and Programming. New York, USA: Routledge; 2016.
    67. Kamarulzaman Y. ORCA - Cardiff University. 2006. The Adoption of Internet Shopping for Travel Services   URL: https://orca.cf.ac.uk/55606/1/U584036.pdf
    68. Chin WW. The partial least squares approach to structural equation modeling. In: Marcoulides GA, editor. Modern Methods for Business Research. Mahwah, USA: Lawrence Erlbaum Associates; 1998:295-336.
    69. Field A. Discovering Statistics Using SPSS. Fifth Edition. London, UK: Sage Publication; 2017.
    70. Schumacker RE, Lomax RG. A Beginner's Guide To Structural Equation Modeling. Third Edition. New York: Routledge; 2019.
    71. Archer N, Cocosila M. Canadian patient perceptions of electronic personal health records: an empirical investigation. Commun Assoc Info Syst 2014;34(1):389-406. [CrossRef]
    72. Assadi V, Hassanein K. Consumer adoption of personal health record systems: a self-determination theory perspective. J Med Internet Res 2017 Dec 27;19(7):e270 [FREE Full text] [CrossRef] [Medline]
    73. Lazard AJ, Watkins I, Mackert MS, Xie B, Stephens KK, Shalev H. Design simplicity influences patient portal use: the role of aesthetic evaluations for technology acceptance. J Am Med Inform Assoc 2016 Apr;23(e1):e157-e161 [FREE Full text] [CrossRef] [Medline]
    74. Liu CF, Tsai YC, Jang FL. Patients' acceptance towards a web-based personal health record system: an empirical study in Taiwan. Int J Environ Res Public Health 2013 Oct 17;10(10):5191-5208 [FREE Full text] [CrossRef] [Medline]
    75. Majedi A. Semantic Scholar. 2014. Consumer Adoption of Personal Health Records   URL: https://pdfs.semanticscholar.org/7d1e/27c3e792b2e5431a576e4293ca54f3d9a924.pdf
    76. Alaiad A, Zhou L. The Moderating Effect of Demographic Characteristics on Home Healthcare Robots Adoption: An Empirical Study. In: Proceedings of the Tenth Midwest Association for Information Systems Conference. 2015 Presented at: MWAIS'15; May 14-15, 2015; Pittsburg, Kansas p. 14-15.
    77. Rho MJ, Kim HS, Chung K, Choi IY. Factors influencing the acceptance of telemedicine for diabetes management. Clust Comput 2014 Mar 12;18(1):321-331. [CrossRef]
    78. Noblin AM. Semantic Scholar. 2010. Intention To Use A Personal Health Record (PHR): A Cross Sectional View Of The Characteristics And Opinions Of Patients Of One Internal Medicine Practice   URL: https://pdfs.semanticscholar.org/6fd2/9bd3c3b4cc3291297943977f887037bb0b1c.pdf
    79. Wu H. Exploring Healthcare Consumer Acceptance of Personal Health Information Management Technology Through Personal Health Record Systems. Baltimore County: University of Maryland; 2013.
    80. Tavares J, Oliveira T. Electronic health record patient portal adoption by health care consumers: an acceptance model and survey. J Med Internet Res 2016 Mar 2;18(3):e49 [FREE Full text] [CrossRef] [Medline]
    81. Hsu C, Lee M, Su C. The role of privacy protection in healthcare information systems adoption. J Med Syst 2013 Oct;37(5):9966. [CrossRef] [Medline]
    82. Davis FD. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q 1989 Sep;13(3):319-340. [CrossRef]
    83. Mathieson K. Predicting user intentions: comparing the technology acceptance model with the theory of planned behavior. Inf Syst Res 1991 Sep;2(3):173-191. [CrossRef]
    84. Venkatesh V, Davis FD. A theoretical extension of the technology acceptance model: four longitudinal field studies. Manag Sci 2000 Feb;46(2):186-204. [CrossRef]
    85. Carlsson C, Carlsson J, Hyvonen K, Puhakainen J, Walden P. Adoption of Mobile Devices/Services - Searching for Answers with the UTAUT. In: Proceedings of the 39th Annual Hawaii International Conference on System Sciences. 2006 Presented at: HICSS'06; January 4-7, 2006; Kauia, HI, USA. [CrossRef]
    86. de Veer AJ, Peeters JM, Brabers AE, Schellevis FG, Rademakers JJ, Francke AL. Determinants of the intention to use e-health by community dwelling older people. BMC Health Serv Res 2015 Mar 15;15:103 [FREE Full text] [CrossRef] [Medline]
    87. Daglish DE. MacSphere: McMaster University. 2013. Electronic Personal Health Records: A Matter of Trust   URL: https://macsphere.mcmaster.ca/bitstream/11375/13325/1/fulltext.pdf
    88. Arauwou JA. Older Adults' Perceptions of the UTAUT2 Factors Related to Intention to Use a Patient Portal for Engagement in Their Healthcare. California: Northcentral University; 2017.
    89. Luque AE, van Keken A, Winters P, Keefer MC, Sanders M, Fiscella K. Barriers and facilitators of online patient portals to personal health records among persons living with HIV: formative research. JMIR Res Protoc 2013 Jan 22;2(1):e8 [FREE Full text] [CrossRef] [Medline]
    90. Hsieh H, Kuo Y, Wang S, Chuang B, Tsai C. A study of personal health record user's behavioral model based on the PMT and UTAUT integrative perspective. Int J Environ Res Public Health 2016 Dec 23;14(1):pii: E8 [FREE Full text] [CrossRef] [Medline]
    91. Jian W, Syed-Abdul S, Sood SP, Lee P, Hsu M, Ho CH, et al. Factors influencing consumer adoption of USB-based personal health records in Taiwan. BMC Health Serv Res 2012 Aug 27;12:277 [FREE Full text] [CrossRef] [Medline]
    92. Hoque R, Sorwar G. Understanding factors influencing the adoption of mhealth by the elderly: an extension of the UTAUT model. Int J Med Inform 2017 Dec;101:75-84. [CrossRef] [Medline]
    93. Klein R. Internet-based patient-physician electronic communication applications: patient acceptance and trust. E-Service Journal 2007;5(2):27-52. [CrossRef]
    94. Bjerkan J, Hedlund M, Hellesø R. Patients' contribution to the development of a web-based plan for integrated care - a participatory design study. Inform Health Soc Care 2015 Mar;40(2):167-184. [CrossRef] [Medline]
    95. Andrews JC, Shimp TA. Advertising, Promotion, and Other Aspects of Integrated Marketing Communications. Tenth Edition. Boston: Cengage Learning; 2017.
    96. Kotler P, Armstrong G. Principles Of Marketing. Seventeenth Edition. Upper Saddle River, USA: Pearson; 2017.
    97. Rogers EM. Diffusion of Innovations. Fifth Edition. New York, USA: Free Press; 2003.
    98. Zikmund WG, Babin BJ, Carr JC, Griffin M. Business Research Methods. Ninth Edition. New York, USA: Cengage Learning; 2013.
    99. Bhattacherjee A. Understanding information systems continuance: an expectation-confirmation model. MIS Q 2001 Sep;25(3):351-370. [CrossRef]
    100. Forquer HA, Christensen JL, Tan AS. Predicting continuance-findings from a longitudinal study of older adults using an ehealth newsletter. Health Commun 2014;29(9):937-946 [FREE Full text] [CrossRef] [Medline]
    101. Nijland N, van Gemert-Pijnen JE, Kelders SM, Brandenburg BJ, Seydel ER. Factors influencing the use of a web-based application for supporting the self-care of patients with type 2 diabetes: a longitudinal study. J Med Internet Res 2011 Sep 30;13(3):e71 [FREE Full text] [CrossRef] [Medline]
    102. Malhotra Y, Galletta DF. Extending the Technology Acceptance Model to Account for Social Influence: Theoretical Bases and Empirical Validation. In: Proceedings of the 32nd Annual Hawaii International Conference on Systems Sciences. 1999 Presented at: HICSS'99; January 5-8, 1999; Maui, HI, USA p. 5-8. [CrossRef]
    103. Aguinis H, Edwards JR, Bradley KJ. Improving our understanding of moderation and mediation in strategic management research. Organ Res Methods 2016 Jan 27;20(4):665-685. [CrossRef]
    104. Preacher KJ, Rucker DD, Hayes AF. Addressing moderated mediation hypotheses: theory, methods, and prescriptions. Multivariate Behav Res 2007;42(1):185-227. [CrossRef] [Medline]
    105. Sardeshmukh SR, Vandenberg RJ. Integrating moderation and mediation: a structural equation modeling approach. Organ Res Methods 2016 Jan 6;20(4):721-745. [CrossRef]

    ‎
    AVE: average variance extracted
    BI: behavioral intention
    CHIT: consumer health information technology
    EE: effort expectancy
    ePHR: electronic personal health record
    FC: facilitating conditions
    GFI: goodness-of-fit index
    GP: general practice
    PE: performance expectancy
    PPS: perceived privacy and security
    RMSEA: root mean square error of approximation
    SEM: structural equation modeling
    SI: social influence
    SRMR: standardized root mean square residual
    UB: use behavior
    UTAUT: unified theory of acceptance and use of technology

    Edited by G Eysenbach; submitted 03.10.18; peer-reviewed by K Nazi, J Tavares, K Fiscella, B McMillan, M Nieroda; comments to author 02.02.19; revised version received 26.03.19; accepted 29.05.19; published 31.07.19

    Copyright

    ©Alaa Abd-Alrazaq, Bridgette M Bewick, Tracey Farragher, Peter Gardner. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 31.07.2019.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.