Barriers to the use of personal health records by patients: a structured review

Eligibility criteria

The literature search identified publications providing evidence about barriers which might interfere with a patient’s decision to adopt a personal health record, or discourage continued use. Publications were included if they considered any stage of patient involvement with a PHR, from their willingness or ability to use the internet or health information technology in the context of PHR use, through to long term use of a PHR as a part of their healthcare.

Publications in English after 2003, in a refereed journal, or presented in a refereed conference or scientific meeting were considered for inclusion. Publications were excluded if they focused on barriers affecting healthcare providers or organisations rather than patients, or if the description of barriers was not based on objective evidence, for example white papers, opinion pieces or editorials.

The types of publication which were sought included:

• Comparative trials involving multiple participating sites;

• Evaluations which involved the collection of data from patients about PHR barriers (using focus groups, interviews, surveys or questionnaires);

• Observational studies; and

• Details of the attitudes and opinions of patients about possible future PHR use.

The review considers the type of study reported, the number of participants in the study, and whether any aspects of the methodology in each case could make the identification of barriers less likely.

A conventional systematic review seeks to provide some degree of quantitative rigour within the findings. However, this structured review has applied a more inclusive, wide-ranging approach to the identification of barriers. Although raw counts of identified barriers are included in the text, there has been no attempt (or intention) to provide an overall qualitative assessment of barriers, or to evaluate their likely impact in particular settings.

Study selection and data extraction

The review process followed published guidelines on Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Liberati et al., 2009). Full literature searches were conducted in PubMed, Embase, CINAHL and ProQuest databases between January and April 2014, with additional searches conducted in May 2014. Details were retrieved for all publications in English from January 2004 to the date of the search.

As an example, the search conducted in PubMed used the terms (personal health record OR personal electronic health records OR patient portal) AND (barrier OR barriers), retrieving 51 citations. Searches were also conducted in Embase, CINAHL and ProQuest using comparable search terms. Additional items were retrieved by tracking citations within publications, and from a small number of other sources.

All publications were initially screened to remove items which were considered to be out of scope, for example where the reference to PHRs was incidental (Bonacina & Pinciroli, 2010; Abimbola et al., 2012), where the barriers identified were exclusively those affecting healthcare providers and organisations (Hart, 2009; Gaskin et al., 2011), or where the focus was on PHR infrastructure issues (Hammond, 2005; Tejero & De la Torre, 2012). The screening process also removed items which made only incidental mention of PHRs (Stead, Kelly & Kolodner, 2005) or barriers (Burke et al., 2010). Publications were included if they provided specific evidence about barriers which might influence the intended or actual adoption of PHRs by patients, or their continued use of a PHR.

Data from the publications which remained after screening were extracted using an iterative process of reviewing full text publications. The data variables which were recorded included the phase of PHR implementation, the type of investigation undertaken, barriers which were identified, the location of the study and the PHR system in use. Details were also recorded where relevant of the number of individuals in the population being studied, and the number included in the study. For studies which obtained information or participation from individuals, aspects of the methodology which might discourage or exclude low capability subjects from seeking to enrol in the study, or reduce the likelihood of their selection as participants were noted. Following an initial review of the data from all in-scope publications, frameworks were developed for the phase of PHR implementation studied, the type of investigation, and the evidence it provided about barriers.

Summary

Searches in PubMed, Embase, CINAHL and ProQuest retrieved a total of 439 publications. Another 36 items were identified from citation tracking and other sources, giving a total of 475 publications. After removing 80 duplicates, 395 publications remained for initial screening. This resulted in the exclusion of 263 records, leaving 132 full text articles to be evaluated for eligibility. This evaluation removed 98 articles which provided no direct evidence about PHR barriers or did not address patient barriers to PHR adoption and use, and literature reviews. This left 34 articles for the synthesis of evidence. This process is outlined in Fig. 1.

Each of the included publications was coded in order to identify the particular phase of the PHR implementation which was being evaluated, the type of investigation conducted, and the barriers which were identified by the study.

Key features of selected studies

Publications were categorised according to investigation type (data from users and non-users; observational studies; patient attitudes and opinions; or mixed). The four tables which follow are grouped by investigation type, and provide details of the included publications, including method, size of target population and number of participants.

Table 1 summarises 20 studies involving the collection of data about barriers from PHR users, or participants who did not initiate or continue PHR use (using focus groups, interviews, surveys or questionnaires). One study in this category used semi structured interviews with health professionals about the characteristics of patients likely to use a PHR, as well as patient questionnaires. Barriers which were identified by both patients and clinic staff were included in the overall summation of barriers.

Table 2 outlines six observational studies which provide a qualitative or quantitative evaluation of demographic data and records of users and non-users, as well as patterns of activity for PHR users.

The six studies in the third category, which collected details of the attitudes and opinions of patients about barriers to possible future PHR use, and the demographic characteristics of those with particular usage intentions, are provided in Table 3.

Table 4 outlines two studies using more than one of the previous categories of investigation.

The publications retrieved for this review displayed a distinct geographic bias, with 32 out of a total of 34 studies reporting on PHR implementations in the USA (with one each from the United Kingdom and the Netherlands). Three particular PHR systems—MyHealtheVet, kp.org and MyChart—accounted for nine of the publications (with three each). Data about a possible bias in the selection of participants was retrieved during the data extraction, and evaluated as a secondary outcome.

Implementation phase

For the purposes of categorising publications, the following schema was developed in order to identify which phase of PHR readiness, adoption and use was being studied in each investigation.

1. Readiness to use a PHR, including evaluations of internet use:

1.1—Patient use of technology, including the internet;

1.2—PHR usage intentions;

1.3—Design of PHRs with User Centred Design (UCD), or usability studies.

2. Initial registration for an account within a PHR system;

3. Initial use of a PHR; publications which studied any use of a PHR at an unspecified time after registration were included in this category;

4. Continued use of a PHR, including long term use;

5. PHR benefits affecting the patient’s health and wellbeing.

Barriers

Each of the included publications provided statements about barriers to patient adoption and continued use of a PHR. This evidence was either: described by users in advance or anticipation of PHR use; reported by potential users as a reason for not commencing use of a PHR, or not continuing that use; or inferred from demographic differences between users and non-users. Thematic analysis was used to identify barriers and to categorise barrier types. This process involved a degree of simplification for some of the barriers described in publications.

Some concepts related to PHR barriers were unambiguous, and required little or no simplification. A reference to ‘Age’, for example, was taken as a straightforward description of a barrier, with no further interpretation required. However, some more diffuse concepts required a degree of interpretation. For example, “...problems due to reading, understanding and filling out forms, not due to poor vision” (Sarkar et al., 2010, p. e4) was recorded as a barrier resulting from poor health literacy, while an observation that “[p]articipants did not perceive the PHR as having added value for managing their existing self-care behaviors...” (Fuji, Abbott & Galt, 2014) was interpreted as a barrier related to ‘Lack of Motivation’. The analysis identified 21 distinct barriers, which are listed by barrier category in Tables 5–10.

Primary outcome: PHR barriers

Tables 5–10 summarise the barriers which were identified in each investigation type, and in each phase of PHR implementation for each barrier category. Most barriers were identified in most phases of PHR implementation, and in most types of study. Barriers which are likely to be associated with socioeconomic disadvantage are flagged. It should be noted that failure to identify a barrier within a particular publication does not provide evidence that the barrier was absent in the population studied, merely that it was not identified. It should also be noted that some of the publications report multiple phases or investigation types.

Each of the sections which follow provides additional information about a barrier category, and the barriers which were identified within that category.

Individual characteristics

Barriers related to age, sex and race or ethnicity are innate characteristics of an individual user, not amenable to change, and were categorised as Individual factors.

Demographic factors

Income, socioeconomic status, level of education, and internet and computer access were categorised as Demographic factors related to an individual’s circumstances.

Capabilities

Four barriers to PHR use were related to the skills and abilities of users and potential users. Functional or text literacy, numeracy, health literacy, and technical literacy and skills were assigned to the Capability factors category.

Health related

Barriers resulting from the individual’s health and wellbeing, including the presence of a chronic disease, disability generally, and specific physical, cognitive or visual limitations, were categorised as Health related.

PHR factors

Barriers associated with the usability of a PHR, the costs associated with access, or lack of information about the PHR were categorised as PHR factors.

Attitudinal factors

The remaining barriers—discomfort with computer use, concerns about privacy, security and confidentiality, and lack of motivation—were categorised as Attitudinal factors.

A focus on those already using technology

In five publications participation was restricted to subjects who already had experience using a web browser (Lin et al., 2005; Nijland et al., 2011), had an existing portal account (Byczkowski, Munafo & Britto, 2011) who had received training in the use of a PHR (Fuji, Abbott & Galt, 2014), or who were required to complete web based surveys during the study (Hilton et al., 2012). These studies did not report barriers related to Capability factors, or to disability.

Exclusion of participants with serious illness or infirmity

In two publications subjects were excluded if they were prevented from participating in an interview as a result of a serious comorbidity (Atreja et al., 2005) or if obvious cognitive deficits were observed (McCleary-Jones et al., 2013). These studies did not report any barriers associated with Individual or Demographic factors, and only health literacy was identified as a Capability factor.

Excluding participants on the basis of language and literacy

Selection of participants for four of the studies (Logue & Effken, 2012; Kruse et al., 2012; Patel et al., 2012; McCleary-Jones et al., 2013) required them to be able to speak, read or write English. These studies identified a wide range of barriers in all categories (11 in all).

Selection of subjects from within a population less likely to be disadvantaged

In these three publications data collection was restricted to participants with a landline telephone (Anderson, 2004), to university undergraduates in schools of business and information systems (Whetstone & Goldsmith, 2009), or to members of a community less likely to be disadvantaged (McCleary-Jones et al., 2013). Health literacy and privacy concerns were the only barriers to PHR adoption and use which were identified in these studies.

Identification of these potential sources of bias is not intended as a criticism of the studies, or of the authors. However, inadvertent bias within the methodologies of studies may mean that any evaluation of barriers within publications (such as that provided by this review) is likely to underestimate the prevalence and significance of barriers, particularly if those barriers are related to exclusion criteria which have been applied in the selection of participants.

Barriers

This literature review has identified evidence for 21 barriers, categorised as Individual, Demographic, Capability, Health related, PHR related and Attitudinal factors, which could interfere with or prevent a patient’s adoption or continued use of a personal health record. The evidence is consistent, with ten of the barriers being identified in six or more publications. However, the frequency with which a particular barrier is identified provides little indication of that barrier’s overall significance, or of its importance in particular settings. The low incidence (four publications or fewer) of reports identifying text literacy, numeracy, generalised disability, or physical and visual impairment as barriers is more likely to result from the research methodology and from the relative invisibility of disadvantaged participants, rather than from the insignificance or absence of these barriers. The complex interrelationship which exists between many of the barriers is suggested in some publications, and emerges more clearly from this review. Socioeconomic status and educational attainment are closely related, and associated with text, technical and health literacy, and with numeracy; internet and computer access, computer skills and discomfort with the use of a computer are closely intertwined; and lastly PHR usability is likely to have a greater impact on users with lower capabilities, and PHR costs will be more challenging for poorer patients. Furthermore, socioeconomic disadvantage is likely to be statistically more prevalent among older citizens, and within non-Caucasian communities. The review identified predominantly US studies, which identify specific issues for elderly, African-American, and Latino communities.

Bias

A number of the included studies chose participants in a way that might result in a lower proportion of disadvantaged and low-capability users, compared with the overall population, resulting in a probable underestimate of PHR barriers. Some degree of bias may be unavoidable. Acquiring evidence about PHRs, including evidence about barriers, must rely on subjects who are able to participate: studies of PHR usage must rely on PHR users, participants must read a written questionnaire in order to respond, and it can be difficult to ethically engage research subjects with cognitive limitations. On the other hand, PHRs are intended for users who are unwell, not just healthy, educated, well-off patients. One study (Zarcadoolas et al., 2013) (not included in the evaluation of bias) deliberately introduced an inverse bias by seeking out participants with a low socioeconomic status.

Principal findings

This review has found evidence of a range of barriers which interfere with the adoption and continued use of PHRs, with 111 instances of 21 distinct barriers identified across 34 publications. This evidence was found in all types of investigation, and in all phases of PHR adoption. Further research may find other as yet unidentified barriers, as well as variants of barriers identified in this review. A close relationship is evident between socioeconomic status and PHR barriers, with 13 of the 21 barriers being associated with socioeconomic disadvantage. This confirms that the use of a PHR is likely to be harder for disadvantaged patients; PHRs as they are currently implemented may not provide a universal solution for problems with healthcare delivery or communication. The relative importance of a PHR barrier cannot easily be deduced from the number of times that it appears within the research literature. Rather, there is an obligation during PHR design, and during PHR implementation, to make a careful assessment of the likelihood of each barrier being present within the population being considered as users. In the USA, the Meaningful Use Stage 2 compliance criterion for 2017, which requires that 5% of patients access their record (Centers for Medicare and Medicaid Services, 2017), is more likely to measure record access by competent PHR ‘early adopters’ than by disadvantaged users. Despite the problem of a growing ‘ehealth divide’ (Cummings, Chau & Turner, 2008) this criterion as currently defined provides little impetus for health professionals or hospitals to encourage PHR enrolments among disadvantaged patients.

Future research priorities

While this review has identified a broad range of PHR barriers, there was insufficient consistency across multiple studies to provide a comprehensive picture of the effect of barriers during PHR implementation and use. If those barriers affecting the population of potential users are to be addressed early in the process of design and implementation, there is a need for better identification and characterisation of both barriers and users. As Kushniruk and Turner have observed, “...greater consideration of who the user is and how the user is involved and their inputs mediated needs to be further articulated. To address these issues it is useful to try to be more precise about who the users are, when and where they are engaged, what expectations we have about our users and why.” (Kushniruk & Turner, 2011, p. 281). Developing a better understanding of the impact of barriers on PHR users will help to ensure more effective use of the resources allocated to PHR systems. There is also a need for a better appreciation of how barriers can affect PHR adoption and use, and how that effect can be countered. Simply being aware of the possibility that a particular barrier may inhibit PHR use for some patients should be enough to ensure that this barrier is taken into account during PHR design and implementation. However, the apparent bias evident in a number of the studies suggests that the existence and significance of barriers is not universally recognised, and that further research may be warranted in order to provide stronger evidence. Finally, the results of this literature review raise a number of interesting questions which may suggest possibilities for future research:

• What does a PHR designed specifically for ‘low functional literacy’ users look like?

• What assistive options within a PHR could help to reduce the negative impact of poor health literacy?

• How can attention to PHR design minimise the impact of cognitive limitations for older patients?

The response to these questions may help to identify a path towards PHRs designed for specific groups of disadvantaged patients, or with an interface which is sufficiently simple, and adaptable to meet the needs of all users.