Norgaard Furstrand Klokker Etal Thee-Healthliteracyframework 2015
Norgaard Furstrand Klokker Etal Thee-Healthliteracyframework 2015
net/publication/288823976
CITATIONS READS
178 12,095
7 authors, including:
All content following this page was uploaded by Ole Norgaard on 30 December 2015.
ISSN 2073-7904
Ole Norgaard
University of Copenhagen, Denmark
Dorthe Furstrand
University of Copenhagen, Denmark
Danish Cancer Society, Copenhagen, Denmark
Louise Klokker
Bispebjerg & Frederiksberg Hospital, Denmark
Astrid Karnoe
University of Copenhagen, Denmark
Roy Batterham
Deakin University, Melbourne, Australia
Lars Kayser
University of Copenhagen, Denmark
Richard H. Osborne
Deakin University, Melbourne, Australia
Recommended citation:
Norgaard, O., Furstrand, D., Klokker, L., Karnoe, A., Batterham, R., Kayser, L.,
& Osborne, R. H. (2015). The e-health literacy framework: A conceptual
framework for characterizing e-health users and their interaction with e-health
systems. Knowledge Management & E-Learning, 7(4), 522–540.
Knowledge Management & E-Learning, 7(4), 522–540
Ole Norgaard
Department of Public Health
University of Copenhagen, Denmark
E-mail: [email protected]
Dorthe Furstrand
Department of Public Health
University of Copenhagen, Denmark
Danish Cancer Society, Copenhagen, Denmark
E-mail: [email protected]
Louise Klokker
The Parker Institute, Department of Rheumatology
Bispebjerg & Frederiksberg Hospital, Denmark
E-mail: [email protected]
Astrid Karnoe
Department of Public Health
University of Copenhagen, Denmark
E-mail: [email protected]
Roy Batterham
Faculty of Health
Deakin University, Melbourne, Australia
E-mail: [email protected]
Lars Kayser*
Department of Public Health
University of Copenhagen, Denmark
E-mail: [email protected]
Knowledge Management & E-Learning, 7(4), 522–540 523
Richard H. Osborne
Faculty of Health
Deakin University, Melbourne, Australia
E-mail: [email protected]
*Corresponding author
1. Introduction
Today, problems and challenges are often met with digital or technological solutions. The
health domain is not an exception to this phenomenon, with the e-health industry rapidly
expanding in many directions. The World Health Organization (WHO) defines e-health
as “the use of information and communication technology (ICT) for health” (World
Health Organization, 2015). e-Health is considered a way to improve quality, capacity,
efficiency and access to healthcare services and information (Hernandez, 2009), and thus
holds the potential to promote health (Camerini & Schulz, 2012) and improve health
equity (Dodson, Good, & Osborne, 2015; Neter & Brainin, 2012).
Engagement in one’s own healthcare, having access to health services and
handling health information in an appropriate way is a complex task for many people,
where the complexity is further increased with the addition of e-healthcare options. The
challenges incorporated in finding, understanding, using and appraising health
information and health services have been labelled as health literacy (Nutbeam, 1998).
Health literacy has been found to be strongly correlated with socioeconomic factors such
as educational level (Beauchamp et al., 2015; HLS-EU Consortium, 2012). Health
literacy, often measured using tests of health-related reading ability and numeracy, has
also been associated with a wide range of behavioural and health outcomes (Berkman,
Sheridan, Donahue, Halpern, & Crotty, 2011). Inspired by the concept of health literacy
and the potential for it to be a health mediator, Norman and Skinner have proposed the
term e-health literacy as a way to capture this emerging field (Norman & Skinner, 2006).
The many diverse aspects of e-health, from being a tool for providers to deliver
better care, to a way for individuals to be informed about their own health (Oh, Rizo,
Enkin, & Jadad, 2005), require that the knowledge, skills and resources needed for people
to use e-health tools are well understood. This understanding should inform the
Knowledge Management & E-Learning, 7(4), 522–540 525
2. Methods
In order to develop a model of e-health literacy that is likely to resonate with IT users and
non-users, patients, healthcare providers, IT experts and managers, we conducted a series
of concept mapping workshops with this diverse range of stakeholders to capture a wide
range of perspectives. Concept mapping exists in many versions; we used a computer
assisted process originally developed by Trochim (Kane & Trochim, 2007; Trochim,
1989a; 1989b). This process has been used broadly for consulting with a wide range of
stakeholders to develop conceptual models for questionnaire development and is
described in detail elsewhere (Busija, Buchbinder, & Osborne, 2013; Klokker et al., 2015;
Osborne, Batterham, Elsworth, Hawkins, & Buchbinder, 2013; Osborne, Elsworth, &
Whitfield, 2007).
Eight workshops were completed between June and August 2012, four with
patients and four with professionals (see Table 1). All patients had a chronic health
condition and were recruited from a rheumatology clinic, a general practitioner and an
umbrella organisation for patients’ associations. The group of professionals included
health professionals, health informatics professionals, researchers in public health, and
computer scientists recruited through the authors’ networks. To maximise cultural and
system diversity two workshops were conducted in London, United Kingdom, with
526 O. Norgaard et al. (2015)
Kruskal’s algorithm (Kruskal, 1964) and cluster analysis applying Ward’s method (Ward
Jr., 1963).
Dendrograms were used to identify commonalities across concept maps. These
explore the hierarchical structure of the data from broad concepts to increasingly refined
sub-concepts down to the individual statements provided by workshop participants.
Hierarchical cluster analysis was applied to produce a dendrogram for each workshop
showing all cluster solutions from two to 20 clusters. With these diagrams the division of
clusters were explored, looking at the specific item content each time a cluster is split into
two smaller clusters. As the aim of this study was a detailed and complete understanding
of the key elements of e-health literacy, clusters were increasingly split until the
maximum number of clusters that made conceptual sense to the researchers was reached.
These clusters, 128 in total, were labelled according to the content of their statements (see
Table 2). Finally, an analysis of the concept labels generated in the workshops and the
analysis of the dendrograms were consolidated into one overarching model of the key
hypothesized domains of e-health literacy.
As a validation, a concept mapping was subsequently applied through an e-survey
of health professionals across the networks of the research group, by e-mail and posted
on social media (selected LinkedIn groups). This type of e-survey, previously described
by Klokker et al. (2015), involved a two-step approach; (1) statement gathering and (2)
statement sorting. In the first step, participants were asked to provide up to 20 brief,
narrative statements responding to the same seeding statement as used in the face-to-face
workshops (outlined above). Further, participants were asked to provide demographic
data including their country of residence, profession, academic qualifications, and current
work role. Statements were collated, and duplicates and very similar statements removed.
In the second step, the statement pool was sent back to the professionals for sorting and
rating, and then analysed as described above.
All data obtained during workshops was anonymised and neither biological
samples nor medical equipment was used. Consequently, specific ethical approval was
not acquired under the regulations of the Danish National Data Protection Agency or the
Danish National Committee on Health Research Ethics in force during the data collection
period.
3. Results
The multidimensional scaling of the data collected during each workshop and the
resulting two-dimensional map with collectively labelled clusters informed the
subsequent analysis. An example of the output, a concept map from the first workshop
with professionals, is shown in Fig. 1.
From the experience of the eight workshops and the content analysis resulting in
128 cluster labels, seven main themes were hypothesised initially through an inductive
analysis by the first author (ON) followed by extensive discussions with RHO, RB, LOK
and LK. The cogency of the content and labels for the themes was then further discussed
and refined (ON, RHO, LK, DF, AK). The conceptual independence of each domain was
carefully considered such that each could be suitable for later development into a scale
within a multidimensional e-health literacy questionnaire.
The labels of the smallest meaningful clusters, resulting from the dendrogram
analyses, and the domains they were attributed to, are reported in Table 2, to illustrate the
528 O. Norgaard et al. (2015)
ideas promoted by patients and professionals respectively. The content and distinctions of
each domain are derived from the statements belonging to these clusters.
Fig. 1. Example of a concept map from the first workshop held with professionals. Each
number represents a statement and each shape represents a cluster as calculated by the
software. The label of each cluster is proposed by workshop participants through
consensus.
Table 2
e-Health literacy domains and associated smallest meaningful cluster labels: Cluster
labels with similar content were merged to create greater overview of data, thus reducing
the number of cluster labels from 128 to 88.
1. Ability to process information
Patients: Professionals: Ability to share information with
others
Have the mental resources to cope Cognitive abilities
with vast information Basic/computer literacy
Functional literacy
Ability to read, write and learn Understand the language used
Ability to learn how to use IT Basic skills in using IT Ability to utilize information and
system
Know one's limitations and be open Critical assessment skills
Know how to critically interact
IT may lower one's attention Health literacy
Have confidence in oneself
Feel confident in systems and data Feeling comfortable with what
information is used Know other people's (hidden)
Ability to use existing systems agenda
Basic computer skills Know when and how to get help
Understand confidentiality
Feel confident in using IT Information exchange and feedback
from health care professionals and Ability to access the systems
Know how to access the systems systems Accept non-human interface during
Be able to make sense of data and interaction with health care
Know how to read and write using
IT understand what it is used for Confidence in using IT
Ability to appraise information Being able to use the IT systems Incorporate use of IT into one's life
Ability to learn the context-specific Knowledge about your different Deeper understanding of how
stills (IT) options systems work
Cognitive abilities and training Being critical when using the
information one gets from the system
to the sense of security and control over data. “Feeling that your information is safe and
secure” (professional) appeared to be imperative to participants. Many statements
focused on the notion of trust: “Trust in the source of information” (professional) and
“You have to trust that the information you give is not misused” (patient). This reflects
the sensitive nature of health data and demonstrated that safety and security was a high
priority to participants. Further, it leads to the need of being in control and promotion of
user rights: “You should own your own data” (professional) and “You should know what
your data are used for” (patient), with transparency being a key value. The need for
personal contact was also raised for participants to feel safe: “Human contact is still
required” (patient). Commercial interests were raised but not frequently emphasised:
“Need to feel that the technology is not driven by commerce (being tracked – big brother
is watching you)” (professional).
5) Motivated to engage with digital services
Statements concerning the incentives for using e-health formed this domain, with a focus
on benefits aligned with use, but also including attitudes such as curiosity, courage,
enjoyment and feeling of closeness and comfort as motivational factors for use. The label
arose from statements such as “Citizen is motivated to use IT” (professional). The value
of the system to the user, the “Purpose or need to use the technology” (professional) was
a core feature, as was “Some people might need an incentive (a sense that they will get
something from it, e.g. might save money, link with others)” (professional). Some
statements included specific motivations such as “You should have courage”
(professional), “You should not be afraid to click around on the [web] pages” (patient),
“You should be curious” (professional) and “You should believe in yourself” (patient).
In this domain statements also touched on other aspects of the attitude towards
digital health, such as “Your age is of significance” (patient), that “Social isolation can be
a reason to use it” (professional) or a situation where it can be “Imposed upon you (you
will have to accept IT in the health system)” (patient), so you feel forced to use it,
resulting in “You can feel alienated” (patient).
6) Access to digital services that work
In this domain, the statements related to access to hardware and software. Statements
included “Access to devices (electronic and medical)” (professional), “You should have
access to solutions” (professional) and that “The system should be easily accessible (user
friendly)” (patient). A recurring point was that “You should be able to afford to buy and
use the technology” (patient). The need for flexibility was also evident, i.e. the need to be
“Able to have 24/7 access” (professional) and that “The systems should communicate
with each other (integration)” (patient). A broad understanding of access emerged,
including “Access to help/manuals” (patient). Not only the user needs access, there was
also an aspiration for “Health information kept in one place where you and relevant
health care professionals have access” (professional). The statement “There are fewer
errors when the patient helps to register drugs and test results” (patient) suggested why
this was important to participants and supports that “The solutions should work”
(professional).
7) Digital services that suit individual needs
This domain focused on the system matching the needs of the individual user. It includes
the user interface: “The user interface must match the user (the user must help to
develop)” (patient) and “There should be understandable feedback from the health
system” (patient). Also it should be “Available in my language” (professional) and “You
should be able to understand the messages you get (not medical language)” (patient).
Knowledge Management & E-Learning, 7(4), 522–540 533
Essentially, “The system should be able to be fitted to the citizen (an adaptive system)”
(professional) and “The system should support the citizen’s development (from novice to
experienced user)” (professional). Data should be adapted to fit the receiver, so that it
provides “Access to having results (e.g. a blood test) interpreted” (patient). An important
aspect of this domain was the systems adapting to disabilities, such as “Options for
blind/visually impaired” (patient) and the option of including personal networks when
needed, in the form of “Empowering relatives when citizens are not able” (professional).
Finally, the system should be receptive to the user’s rights, such as “Having the right not
to know” (professional) and “Acknowledgement that there will be people who will not
use the technology” (professional).
Confirmatory e-consultation process
The e-consultation resulted in 67 unique statements, which were broadly consistent with
the seven domains, and all domains were represented in the data. The health aspect in the
second domain, “Engagement in own health”, was represented by statements such as
“Understand health concepts (bodily functions etc.)”. The fourth domain: “Feel safe and
in control”, was illustrated by the statement “Trust the source of information”. The fifth
domain on being motivated to engage with digital services was included in statements
such as “Have an incentive for using the digital health service.” Thus the e-consultation
confirms the results as presented above.
The e-health literacy framework (eHLF)
The seven domains were clearly related to each other and connect the individual and the
individual’s experience with the system acting as an e-health literacy framework. Fig. 2
represents the seven domains on a pair of axes: from domains largely dependent on the
individual to domains largely dependent on the system, on the horizontal axis; and from
domains that relate to externalized, observable actions to domains representing more
internalised concepts and feelings on the vertical axis. The first two domains concerning
ability to process information (domain 1) and engagement in own health (domain 2) were
largely dependent on the competences of the individual and thus placed on the
individual’s side in the model. Access to working systems (domain 6) and a system’s
ability to suit individual needs (domain 7) depend mainly on the characteristics of the e-
health systems, so these two domains were placed on the system side of the model. The
user skills and knowledge have little influence on whether there is hardware accessible or
an Internet connection when needed, or whether the system adapts to individual needs,
such as visual impairment, if the system is not providing this option. However, individual
capabilities and resources do influence perception of access to a system and to what
extent the system adapts to individual needs. Therefore, knowing about the user’s
perception of the technology contributes to an inclusive understanding of e-health literacy.
The interaction between the individual and the system is where unique aspects of
the concept of e-health literacy start to unfold. How a person might engage with
information in the context of a system (domain 3) is dominated by more than just
technical skills. Experiencing safety and control (domain 4), benefit and comfort, and
having the right attitude in approaching technology (domain 5) become just as relevant as
knowing the inner workings of the systems and having the skills to navigate it.
Moreover, the model includes the relative level of internalization of each ability.
Within each circle the more external skills, such as reading or having access to a working
system, were placed in the upper regions, whereas the more internalized concepts such as
motivation, engagement and individualisation were placed in the lower regions.
534 O. Norgaard et al. (2015)
Fig. 2. The e-health literacy framework (eHLF). The figure shows the interaction
between individual and system as illustrated by the domains.
4. Discussion
This study presents the development of a model of e-health literacy based on the
experiences of a wide range of stakeholders, by applying a systematic, inductive
approach. This has resulted in a seven-domain framework, the e-health literacy
framework (eHLF), which provides a new way to understand the interaction and relation
between individuals and the system.
The strength of the eHLF is that it includes not only the known domains of health,
information and basic digital skills, but also includes new elements with relevance to the
dynamics that occur when the system meets the individual, as exemplified by the
domains 3. Ability to actively engage with digital services, 4. Feel safe and in control,
and 5. Motivated to engage with digital services, placed in the interaction between these
two (see Fig. 2). Together with the domains 6. Access to digital services that work and 7.
Digital services that suit individual needs, bound to the system, this illustrates how the
eHLF supports an understanding of e-health literacy that is not only about the
individual’s abilities and resources but also strongly context dependent and related to the
complexity of systems.
Knowledge Management & E-Learning, 7(4), 522–540 535
All literacies included in Norman and Skinner’s Lily Model are contained in the
first three domains of eHLF that describe basic skills to process information, understand
health and use technology, although both eHLF domains concerning engagement are
broader. The content of the remaining four domains of eHLF are not included in Norman
and Skinner’s model.
The nine themes suggested by Koopman et al. are all represented in the eHLF
domains. Compared to the domains of eHLF, the themes from Koopman et al. tend to be
more specific. E.g. the theme “Asynchrony”, attentive to the digital influence on
increasing asynchrony in the access to healthcare professionals, but not regarding the
access to hardware, software, training and economic means, that are also included in the
eHLF domain.
Two of the Koopman themes mapped into domains 5 and 7 were only partially
covered in eHLF while most other themes are partially but not fully contained by the
eHLF. For example, the theme “Looking for information for and about others” is about
an individual’s need for information (domain 7), but it also contains a specific need for
looking up information on conditions not affecting one self.
Interestingly, the elements incompletely covered by the eHLF are already well
defined in broader understandings of health literacy. Our model is limited to the users
interacting with technology and digital services in a health context while the current
understanding of the concept of health literacy has multidimensional characteristics
(HLS-EU Consortium, 2012; Osborne et al., 2013), and includes aspects of social support
networks and engagement with health professionals.
As the only e-health literacy model for almost a decade, Norman and Skinner’s
Lily Model became the foundation for another two expansions of the original model
(Chan & Kaufman, 2011; Gilstad, 2014): Chan and Kaufman applied Bloom’s taxonomy
to each of the six literacies adding a cognitive dimension to the existing model (Chan &
Kaufman, 2011). The eHLF does not contain this taxonomic construct, most likely due to
its origin from a concept mapping-based development process. However, this could be
incorporated into the development of instruments to measure e-health literacy based on
eHLF.
In 2014, Gilstad suggested an extension of Norman and Skinner’s Lily Model
(Gilstad, 2014). Through an interdisciplinary review of technology studies, human and
social sciences and health studies she applied elements of communicative expertise, the
bodily experience, cultural, social and institutional context and propositional and
procedural literacy. These additions are not covered by eHLF but the context, relationship
and personal values may be covered in the understanding of health literacy or should be
included in a third dimension placed outside the interaction between the individual and
the system.
The concept mapping process that we undertook included a broader range of
stakeholders than was reported for the other models. The other models focus on
individual skills and none include the system domains that are represented in our model.
The two domains attributed to the system: 6. Access to digital services that work and 7.
Digital services that suit individual needs, include important features of what is needed
for the user to use and benefit from a digital service. The grounded approach with
inclusion of a wide range of stakeholders across settings is a strength of the eHLF
development that has generated new and highly relevant aspects of e-health literacy that
have the potential to advance this field.
Knowledge Management & E-Learning, 7(4), 522–540 537
these and similar setting, the framework may require refinements in other contexts, such
as cultures from outside the European continent or in developing countries.
The background of the participants was wide ranging. The recruitment was
through convenience sampling, mainly via our own networks (professionals and e-survey
participants) and patients were recruited from clinics but also through patient associations.
Importantly, the structure of the concept mapping workshops ensured that all participants
felt safe to contribute their own view, with careful attention to not let the facilitators’
perspectives influence participants. A wider demographic spread among the participants
may have generated a wider set of responses. While the concept mapping provides some
reassurance that the eHLF domains are comprehensive, further work in this area is
warranted, particularly with individuals from non-European backgrounds.
The process of constructing a concept like e-health literacy will always involve
multiple analytic and decision-making stages. The systematic approach provided by the
concept mapping method and software contributed structure to the data, that supported
the inductive work of identifying themes among the statements. Being qualitative work,
the prior understanding of the researchers will influence the analysis. Thus the authors’
background in public health as well as medicine, education, and medical informatics, will
have an impact on the results. The validity and rigor of the qualitative process was
strengthened by repeated immersion in the content of the statements in multiple sessions
with different groups of authors as well as the confirmatory process of the e-survey. In
this work, the method facilitated the emergence of domains not previously recognized
from the literature that were clearly relevant to participants in the workshops.
5. Conclusion
This paper presents a new e-health literacy framework (eHLF) which provides a
comprehensive scaffold for investigation of an individual’s ability to use and benefit from
e-health technologies at the micro, meso and macro level. The seven domains provide
novel insight into e-health literacy from the perspective of the system and the individual
as well as the interaction between the two. The eHLF provides a starting point to assist
researchers, practitioners, funders and policymakers to explore e-health literacy from a
development, research, policy or user point of view in order to obtain a better match
between the demands of e-health systems and the knowledge, skills, resources and
motivation of users.
Acknowledgements
Parts of the project took place when LK and ON were funded by the Danish National
Strategic Research Council. RHO was funded in part by a National Health and Medical
Research Council of Australia Senior Research Fellowship. DF is a PhD fellow supported
by the Tryg Foundation. Author AK was funded by The Health Foundation in Denmark
for a six month project regarding e-health literacy. Mr. Jacob Ø stberg Hansen provided
assistance in collating statements gathered through the e-survey. Thank you to all who
contributed to this work by participating in the workshops or the e-consultation. Thanks
to Ms Lena Sundby Jensen for assisting in editing the paper.
Knowledge Management & E-Learning, 7(4), 522–540 539
References
Beauchamp, A., Buchbinder, R., Dodson, S., Batterham, R. W., Elsworth, G. R., McPhee,
C., Sparkes, L., Hawkins, M., & Osborne, R. H. (2015). Distribution of health literacy
strengths and weaknesses across socio-demographic groups: a cross-sectional survey
using the Health Literacy Questionnaire (HLQ). BMC Public Health, 15(1): 678.
Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. (2011).
Low health literacy and health outcomes : An updated systematic review. Annals of
Internal Medicine, 155(2), 97–107.
Busija, L., Buchbinder, R., & Osborne, R. H. (2013). A grounded patient-centered
approach generated the personal and societal burden of osteoarthritis model. Journal
of Clinical Epidemiology, 66(9), 994–1005.
Camerini, L., & Schulz, P. J. (2012). Effects of functional interactivity on patients’
knowledge, empowerment, and health outcomes: An experimental model-driven
evaluation of a web-based intervention. Journal of Medical Internet Research, 14(4):
e105.
Chan, C. V., & Kaufman, D. R. (2011). A framework for characterizing eHealth literacy
demands and barriers. Journal of Medical Internet Research, 13(4): e94.
Cresswell, K., & Sheikh, A. (2009). The NHS care record service (NHS CRS):
Recommendations from the literature on successful implementation and adoption.
Informatics in Primary Care, 17(3), 153–160.
Dodson, S., Good, S., & Osborne, R. (2015). Health literacy toolkit for low- and middle-
income countries: A series of information sheets to empower communities and
strengthen health systems. New Delhi: World Health Organization, Regional Office
for South-East Asia. Retrieved from http://apps.searo.who.int/PDS_DOCS/B5148.pdf
Gilstad, H. (2014). Toward a comprehensive model of eHealth literacy. In Proceedings of
the 2nd European Workshop on Practical Aspects of Health Informatics. Retrieved
from. http://ceur-ws.org/Vol-1251/paper7.pdf.
Greenhalgh, T., Hinder, S., Stramer, K., Bratan, T., & Russell, J. (2010). Adoption, non-
adoption, and abandonment of a personal electronic health record: Case study of
HealthSpace. BMJ, 341: c5814.
Hernandez, L. M. (2009). Health literacy, eHealth, and communication: Putting the
consumer first: Workshop summary. Washington, DC: Institute of Medicine of the
National Academies, National Academies Press.
HLS-EU Consortium. (2012). Comparative report on health literacy in eight EU member
States - The European health literacy project 2009-2012. Vienna: Ludwig Boltzmann
Institute for Health Promotion Research.
Kane, M., & Trochim, W. M. K. (2007). Concept mapping for planning and evaluation.
SAGE Publications, Inc.
Kayser, L., Kushniruk, A., Osborne, R. H., Norgaard, O., & Turner, P. (2015). Enhancing
the effectiveness of consumer-focused health information technology systems through
eHealth literacy: A framework for understanding users’ needs. JMIR Human Factors,
2(1): e9.
Klokker, L., Osborne, R., Wæ hrens, E. E., Norgaard, O., Bandak, E., Bliddal, H., &
Henriksen, M. (2015). The concept of physical limitations in knee osteoarthritis: As
viewed by patients and health professionals. Quality of Life Research: An
International Journal of Quality of Life Aspects of Treatment, Care and
Rehabilitation, 24(10), 2423–2432.
Koopman, R. J., Petroski, G. F., Canfield, S. M., Stuppy, J. A., & Mehr, D. R. (2014).
Development of the PRE-HIT instrument: Patient readiness to engage in health
information technology. BMC Family Practice, 15(1): 18.
Kruskal, J. B. (1964). Multidimensional scaling by optimizing goodness of fit to a
540 O. Norgaard et al. (2015)