This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in Medicine 2.0, is properly cited. The complete bibliographic information, a link to the original publication on http://www.medicine20.com/, as well as this copyright and license information must be included.
Many speak of the digital divide, but variation in the opportunity of patients to use the Internet for health (patient eHealth readiness) is not a binary difference, rather a distribution influenced by personal capability, provision of services, support, and cost. Digital divisions in health have been addressed by various initiatives, but there was no comprehensive validated measure to know if they are effective that could be used in randomized controlled trials (RCTs) covering both non-Internet-users and the range of Internet-users.
The aim of this study was to develop and validate a self-completed questionnaire and scoring system to assess patient eHealth readiness by examining the spread of scores and eHealth inequalities. The intended use of this questionnaire and scores is in RCTs of interventions aiming to improve patient eHealth readiness and reduce eHealth inequalities.
Based on four factors identified from the literature, a self-completed questionnaire, using a pragmatic combination of factual and attitude questions, was drafted and piloted in three stages. This was followed by a final population-based, cross-sectional household survey of 344 people used to refine the scoring system.
The Patient eHealth Readiness Questionnaire (PERQ) includes questions used to calculate four subscores: patients’ perception of (1) provision, (2) their personal ability and confidence, (3) their interpersonal support, and (4) relative costs in using the Internet for health. These were combined into an overall PERQ score (0-9) which could be used in intervention studies. Reduction in standard deviation of the scores represents reduction in eHealth inequalities.
PERQ appears acceptable for participants in British studies. The scores produced appear valid and will enable assessment of the effectiveness of interventions to improve patient eHealth readiness and reduce eHealth inequalities. Such methods need continued evolution and redevelopment for other environments. Full documentation and data have been published to allow others to develop the tool further.
The term eHealth is used in various ways, some (eg, World Health Organization) [
In developing the ideas for this study, literature was first reviewed in November 2010 and updated in May 2013 from Web of Knowledge, PubMed, and Google Scholar using the terms (1) [E-health OR ehealth OR telehealth* OR telemedicine OR (ICT AND health) OR (technology AND health <in topic>) AND (readiness OR preparedness OR (implementation AND measure*) <in title>], (2) E-health literacy, (3) (Digital divide OR digital inclusion OR digital exclusion OR e-health inequalities) AND health, and (4) Inequality AND measure <in title> AND health. Web of Knowledge was used to examine citations of this literature for further relevant studies.
There is evidence that direct use of the Internet by patients can benefit patients [
There are four domains of barriers to eHealth opportunity that were identified (1) provision of eHealth opportunity, (2) personal abilities of the patient, (3) the support from others they may have to use eHealth, and (4) economic barriers.
Provision of eHealth opportunity varies. For example, while some British general practices [
Physical and psychological attributes of patients contribute to digital divisions in health. Someone may have problems from sight or hearing impairment, arthritis, or lack of mobility in their hands. They may have no prior experience or find it difficult to learn Internet use, have limited literacy or health literacy, or lack confidence either in their Internet use or in making decisions using health information. They may distrust the Internet [
Some factors limiting personal use of eHealth may be diminished if people have support from others. For example, anonymous e-mail support may help people with long-term conditions use the Internet [
Finally, economic factors may affect digital divisions in health [
Others have examined barriers to eHealth use and the eHealth readiness of organizations or health services [
Should governments or health services address digital divisions in health? Some argue that it is just a matter of time before everyone has Internet access and that digital divisions will disappear. Others remind us that in the diffusion of technology [
Like the seven-year difference in life expectancy by social class in England [
Initially, physical access to eHealth received a good deal of attention. From the late 1980s, there were experiments with public access kiosks [
Various studies have addressed eHealth inequalities or tried to ameliorate their impact. For example, Kerr et al [
Digital divisions caused by physical disability have been subject to legislation. Web accessibility laws and regulations have encouraged developers to make websites accessible to those with visual, auditory, motor, neurological, or cognitive impairments. In Britain, the Disability Discrimination Act 1995 [
Projects and national initiatives such as those described above need to measure eHealth inequalities to know (1) if action is needed, (2) what is the main cause of inequality, and (3) if inequalities are addressed, if the intervention was successful and cost effective. But inequality cannot be directly measured; it has to be measured as a difference in another variable, namely eHealth readiness. To compare eHealth inequality over time, we need a measure of “patient eHealth readiness” that is comprehensive, valid, and reliable. A measure that is also “diagnostic,” allows development of interventions tailored to the needs of populations.
Others have considered the “readiness” of practitioners, organizations, or communities to adopt telehealth or eHealth [
There are two groups of literature that exist at the “patient level” (1) the “digital divide” and (2) eHealth literacy. The digital divide-as the term implies-tends to be measured as a binary division. For example, whether someone has or does not have access to the Internet or has or has not used the Internet in the last three months [
Simple measures of whether or not someone has Internet access are insufficient as even among Internet users some may be more ready to make progress in using eHealth if they have access to support and are not struggling with the cost of access. In particular, interventions at patient and community levels need tools that can measure their impacts.
The aim of this project was to get the benefits of a scaled (rather than binary) approach (like eHEALS), but to include eHealth provision, support, and economics in the scale. In particular, the study aim was to develop and validate a self-completed questionnaire and scoring system for use in intervention studies hoping to improve eHealth readiness and reduce eHealth inequalities.
The Patient eHealth Readiness Questionnaire (PERQ) and related scores have been developed in two stages. First, four domains (1) provision (from the digital divide literature), (2) personal (from the eHealth literacy literature), (3) support, and (4) economic were used to draft a self-completed questionnaire and take it through three stages of piloting (January-March 2012). Second, a cross-sectional population survey was carried out (April-August 2012) and proposed scoring systems checked and iteratively refined to ensure construct validity.
Although there are good national statistics on home Internet access [
The most frequently used [
Prior to the Dutch studies [
Although interpersonal support to help people start using the Internet was a major component of the Race Online 2012 campaign [
Whether or not someone uses the Internet for their health depends on whether they are motivated to do so [
Three pilots (PERQ1-3) including repeatability were followed by a baseline survey. (PERQ4):
The first had 15 people (work colleagues and friends). Questions from eHEALS [
The second had 20 friends and family of a research assistant; 17 of these were subsequently asked to complete PERQ4 to assess repeatability (reported below).
The third had a convenience sample of 103 houses likely to have a high proportion of more elderly residents. This was used to test the survey method, response rate, completeness of data, and that non-Internet users would respond. The response rate was 44% and data were reasonably complete. It was found that 5 out of 43 (12%) respondents had not used the Internet.
After each stage, revised questionnaires were circulated among colleagues to check readability (see
As one intended use of PERQ was in a geographically based cluster randomized controlled trial (RCT), it was appropriate to test that mode of use. The aim was to use a sample representative of urban, suburban, and semirural postcodes and different levels of affluence to pilot the questionnaire, its distribution, and methods for construction of eHealth readiness and inequality scores. The 2001 census included 14,279 postcodes for the PL postcode area, with a total population of 510,223. There were seven postcodes (total population 3243) with very high populations (being either military camps or university halls of residence) that were excluded. The remaining postcodes were “ordered” by population and a 1% systematic sample (142 postcodes) was taken. Each postcode was “looked up” on the free Zoopla website service giving estimated property values to find current average property values and number of properties in each postcode. To achieve a “practical” sample of just fewer than 1000 properties, all postcodes further than 12 miles from Plymouth University were excluded, leaving 79 postcodes. These 79 postcodes were again ordered by Zoopla average property values, and a further systematic sample of 53/79 postcodes was taken with a total of 975 properties.
The final sample therefore comprised 53 postcodes within 12 miles of Plymouth University, with a total of 975 properties, population of 2126, and an average of 2.18 people per house. Average property prices (January 2012) per postcode ranged from British £78,163 to £459,360. The sample was representative of the range of property prices. Number of properties per postcode (a crude measure of rurality) varied from 1 to 53.
We attempted to deliver questionnaires by hand to all 975 properties in April 2012. There were thirty houses no longer in use, leaving 945 occupied houses as our sample. The research assistant called at each house and if someone answered, she explained the purpose of the survey and if possible handed the resident a questionnaire and covering letter (24 refused to take the questionnaire). If there was no response at the house, the questionnaire and covering letter were posted through the letterbox. In June 2012, reminders were posted through the letterbox of 658 who had not responded. The instructions on the questionnaire, and explained by the research assistant, were for the person with the next birthday in the house to complete the questionnaire.
Survey data were used to assess the ability of PERQ to collect good quality data, and used to develop and iterate a scoring system suitable for use in intervention studies, particularly RCTs. This required that only questions that contributed to the score were included, collected data were complete, consistent, and valid, scales must not have floor or ceiling effects, and must reflect meaningful changes.
Scales that combine a number of “Likert style” attitude questions normally assess reliability using Cronbach alpha. In this study, the construction of the eHealth readiness scale relied on pragmatic combinations of factual questions with some ratings, so assessing scale properties such as repeatability, face, and construct validity, was also pragmatic. Questions were cross checked for consistency and reviewed for their contribution to eHealth readiness scores, face validation against other information sources, or description of sample demographics. Comparisons of scores between subgroups were made using Mann Whitney U tests for groups less than 100 and
The contribution of each question was checked. Not all questions made direct contributions to scores, some were asked to ground respondents to give them the “right frame of mind” for subsequent questions. Other questions were used as consistency checks.
Scores for each constructed variable were essentially arbitrary, but to have some way of measuring change before and after interventions, an overall score is needed that is at least ordinal, and if possible approximates to a cardinal scale. Similarly, overall scores need to combine component variables in a sensible manner. A pragmatic and iterative approach was taken to examine construct validity of scores by examining the scores of sampled individual respondents with a range of scores. If the order and difference in scores between individuals did not match with an understanding of the barriers to adopting eHealth, the weights of scores were adjusted.
Scoring was also adjusted after examining the repeatability of scores and to cope with occasional missing values. This process of tuning scoring weights continued until all components seemed internally consistent. Various methods of combining the four subscales to produce an overall readiness scale were tried, checking for construct validity by examining differences between Internet users and non-Internet users.
Once the scoring system was finalized, questionnaires and dataset were again reviewed to check that all questions and answers were useful either as contributors to the score, as “grounding” for other questions, or as consistency checks (see
Finally, data from the survey were used to model possible changes to participants “states” and to check the ability of the scoring system to measure those changes. Comparisons of scores between subgroups were made using Wilcoxon Signed Rank tests for groups less than 100 and
This provided a further check that the subscores and weights seemed sensible, and to allow an assessment of methods of analysis and estimation of sample size for possible RCTs.
The anonymized dataset from the cross-sectional survey is available from the author.
Despite care in design and three stages of piloting, PERQ still had missing and some inconsistent data (see
Sample response and characteristics.
Personal eHealth readiness questionnaire and scale.
Scores (0-10) representing “need for health information and support” were constructed from one multi-part factual question (A3), by adding 2 points for each professional contact and health-information seeking behavior in the last three months. Scores had face validity, for example, women had higher Need scores than men. Need was used as a modifier of Provision scores.
Personal use of the Internet in the last three months was similar to national figures (271/342, 79.2% versus 77% from Office for National Statistics, ONS) [
For Internet users, provision scores comprised two parts (section D) (1) General Internet Provision (4 points) ascertained by questions about what opportunities there are to access the Internet, and (2) Health Internet Provision (5 points) ascertained by questions on GP website (3.5 points), and Internet condition specific information and support (1.5 points). Provision of online information and support may vary by condition (eg, there are many resources for breast cancer, but fewer for stroke). However, not everyone has a need for health information or support and so may never have had reason to look for their GP website or for health information. So the Need score was used to moderate Provision; more was added to Health Internet Provision if Need was equal to zero (see
Component questions showed that nearly half (30/71) of non-Internet users had an Internet connected computer at home. Of 271 Internet users, 249 people had used it at home, but 3 said they had no home Internet connection, of these, 2 had used a mobile device and so it is possible that questions about “home Internet use” need to be clarified. There were 3 other people who had accessed the Internet at home, but did not apparently know if they had an Internet connection, may have not known about the “speed” of their home Internet, so some clarification may be needed for that question. A substantial minority (33/243,13.6%) thought their home Internet connection was not fast enough, a third of these said it was because they would need to pay more, a third because their provider did not offer a faster connection, and a third did not know.
Of 271 Internet users 89/271 (32.8%) had looked at their GP’s website, 51/271 (18.8%) thought their GP had a website, but had not seen it, 7/271 (2.6%) thought their GP did not have a website, and nearly half (122/271, 45.0%) did not know. Of the 89 who had looked at their GP’s website, 64 knew they could order repeat prescriptions online, 6 said their GP did not offer this service, and 14 did not know. Only 1 person knew they could see their medical record online, 20 knew they could not, but 66 did not know. Half (143/271, 52.8%) of Internet users had tried to find information on health topics with all but 9 having found what they wanted, but only 27/271 (10.0%) people had tried to contact an organization or forum or other people online connected with health.
The internal consistency of Provision scores was addressed by comparison of answers to sections B (where people had used the Internet) and D (home Internet provision and use for health).
Four sub-scales of Provision, Personal, Support, and Economic presented as "short scales" of 0-5, showing Internet-users and non-Internet-users, including mean and standard deviation (SD) for full (0-9) scales.
Personal scores comprised moderated confidence scores for Internet users (section E) and the willingness to try using the Internet for non-Internet users. Internet users rated their skills on four tasks from which skills scores (0-12) were constructed. Users then rated their overall Internet confidence (0-10). This sequence of questions aimed to ground their confidence rating in the reality of their ability and to provide a consistency check on their confidence rating. Skills scores correlated with confidence self-ratings (Spearman’s Correlation=0.60,
Component questions showed that just under half of non-Internet users said they would try using the Internet if they had help, would have a home Internet connection if they had help and it was cheap, and would use the Internet for health at some other place (most frequent choice public library). Questionnaires asked about disabilities. Six non-Internet users and six Internet users said they had disabilities (including arthritis, eye problems, hearing impairment, learning difficulty, and dyslexia) that made using computers difficult, but this information was not used in Personal score calculations on the assumption that respondents would themselves make that adjustment.
The face validity of Personal scores was assessed by exploring associations with frequency, range, and ubiquity of Internet use. As expected there were strong associations between frequency of use and moderated confidence (χ2
12=81,
Support scores were largely based on factual questions. There were 22 out of 271 Internet users that did not complete the section on support, half of these (10/22) said (H1) that they had no barriers to Internet use and were confident in using the Internet for health (E3).
Component questions showed that among Internet users, only 58/271 (21.4%) had been given information by health professionals to help them use the Internet. Just under half (117/271, 43.2%) knew where they could find help locally in using the Internet; many of these (78) cited their local library. A quarter (68/243, 28.0%) said there had been times when help would have been useful, and of these, 50 had someone they could ask, of which 47/50 could ask about health. Nearly three-quarters (49/71, 73%) of non-Internet users had someone use the Internet for them. There were 40/71 (65%) that had someone that could help if they wanted to try using the Internet.
That Support scores were less differentiated between Internet users and non-Internet users “made sense” in the way that questions were asked and answered. Exploration of how Support and Personal scores were associated led to a pragmatic combination, using the Personal score to moderate the Support score in the overall Readiness score (see below). There were no ceiling effects on Support.
The Economic subscale was constructed slightly differently to the other three subscales, relying on comparison of perceptions of the cost of using the Internet compared to other health activities such as visiting their GP or local hospital. Internet users and non-Internet users answered the same questions.
Considering the component questions, there were significant differences on the two Internet questions and on the cost of visiting the hospital between Internet users and non-Internet users. For the two Internet questions, this was dominated by the “don’t knows” among non-Internet users; 45% (29/64) of non-Internet users did not know about the cost of home access and 57% (35/61) about the cost of mobile access compared to 4.8% (13/269) and 32.8% (87/265) of Internet users. There was no difference between Internet users and non-Internet users in perceptions of cost in getting to the local library or GP. Most (195/325, 60.0%) did not think it cost much to get to a public library, but a large minority (95/325, 29.2%) did not know. The vast majority (317/337, 94.1%) agreed that visiting their GP cost nothing or very little. Non-Internet users were more likely to think that visiting their nearest hospital cost nothing or little (55/67, 82% versus 185/268, 69.0%; χ2
4=15.7;
Question H1 sought to identify the most important issue in using or not using the Internet for health. The original intention was to use this question to weight subscale scores in their combination to produce an overall readiness score. This idea was abandoned when it was realized that there was a close relationship between the Support and Personal subscales and an alternative combination method was developed. However, H1 remained a useful consistency check on the subscale scores.
Further breakdown of the groups in
To test the construct validity of subscales, constructed variables were compared to answers to question H1 (
The initial intention was to create an overall eHealth readiness score by taking the mean of the four subscales, that is (Provision + Personal + Support + Economic)/4. However, exploration of the data led to recognition that Support was much more important for non-Internet users. Those who were already competent Internet users for health needed little support and scored low on Support. This reduced their overall eHealth readiness score and was misleading. Support was therefore added to eHealth readiness in inverse proportion to that person’s Personal score, (ie, people with a higher Personal score had less weight given to their Support score). Through a process of iteration considering whether the impact on overall eHealth readiness made sense, the term 3*Support/(Personal+Support) was added as a “Modified Support” term. This Modified Support score can range from 0 to 3 and the sum of Personal and Modified Support can range from 0 to 10.5. The Economic score also seemed less important in being “eHealth ready” than Personal and Provision scores, so the short score (range 0-5) was used as the contribution to eHealth readiness. So,
eHealth readiness= Provision (0-9) + (Personal + Modified Support) (0-10.5) + Short economic (0-5)
It was then multiplied by 9/24.5 to scale to the range 0-9.
Numbers choosing statements (in section H) that best summarized their view of using the Internet for health and Mann Whitney
Overall View | Non-Internet user | Internet user | Total | “Nearest” variable | Mean score for those who chose this item versus rest ( |
||
|
Home access | No home access | |||||
|
|
|
|
|
|
|
|
(H11) No need for health information. | 1 | 3 | 34 | 38 | NEED | 1.8 versus 3.3 |
|
(H12) No interest in using the Internet. | 11 | 29 | 12 | 52 | PERSONAL | 1.1 versus 6.2 |
|
(H13) Would use the Internet more for health if could get a good Internet connection. | 0 | 2 | 3 | 5 | PROVISION | 2.8 versus 3.8 |
|
(H14) Don’t understand the Internet that much. | 9 | 4 | 17 | 30 | PERSONAL | 2.7 versus 5.7 |
|
(H15) Would use the Internet more for health if could get someone to help. | 0 | 0 | 10 | 10 | SUPPORT | 2.1 versus 2.1 |
|
(H16) Would use the Internet more for health if money were no object. | 1 | 2 | 2 | 5 | ECONOMIC | 0.8 versus 2.5 |
|
(H17) Uses or would use the Internet for health and have no real barriers to that use. | 4 | 0 | 185 | 189 | PERSONAL | 7.2 versus 3.2 |
|
ECONOMIC | 3.0 versus 1.9 |
||||||
PROVISION | 4.0 versus 1.9 |
||||||
SUPPORT | 3.1 versus 2.3 |
||||||
READINESS | 4.4 versus 2.3 |
||||||
Total | 26 | 40 | 263 | 329 |
|
|
|
a
bnsd=no statistically significant difference
Seventeen of the 20 people who completed the second-stage pilot (January 2012) completed PERQ4 in September 2012. Of these, four non-Internet users were excluded as the questions in section C had changed too much between the earlier and later version of PERQ to be comparable. For 13 Internet users their January data was converted to the September version of the questionnaire to allow a comparison and some assessment of “repeatability” and change over time. Each pair of questionnaires was examined for changes to answers and the impact on the scoring system to see if it made sense and if the scoring system was appropriate. This check resulted in some changes to the scoring system. With the final scoring system there was reasonable consistency in scores between January and September 2012 (see
Modelled results of interventions, showing impact on subgroup and whole sample on eHealth readiness score and eHealth inequality and Wilcoxon signed ranks test (
Assumed changes | Impact on sub group mean readiness score | Impact on whole sample | |
Mean readiness score | Inequality (standard deviation of readiness) | ||
|
|
|
|
20 non-Internet users are supported in getting online. They have not looked at the GP website, but have found other health information online. | Increase 1.8 to 5.0 |
Increase 4.24 to 4.44 |
Decrease |
20 Internet users who previously used computer at home and at work got mobile access and became aware of GP services including patient access to records. | Increase 4.9 to 6.2 |
Increase 4.24 to 4.32 |
Increase |
20 Internet users who said their Internet connection was too slow who got a faster connection and many of who used it to contact someone about health. | Increase 4.2 to 4.6 |
Increase 4.24 to 4.27 |
No change |
80 Internet users who previously had not had advice on using Internet from HCP, recommended websites by GP. | Increase 4.5 to 4.7 |
Increase 4.24 to 4.29 |
No change |
Distribution of eHealth readiness scores on possible scale 0-9 showing Internet-users and non-Internet-users.
A questionnaire (PERQ) and scoring system has been developed comprising four components of patient eHealth readiness (1) provision of Internet and Internet for health, (2) personal capacity to use it, (3) support in using it, and (4) economic barriers to use. The scoring system appears consistent, to have face and construct validity, and to produce a score that can be used to assess interventions that improve eHealth readiness. By examining the standard deviation of scores, eHealth inequalities can be reviewed to ensure that interventions have not worsened inequalities. The questionnaire is being used in two studies locally [
Although national cross-sectional data from the ONS and Oxford Internet Survey (OIS) show the uptake of the Internet, and sometimes include questions on health, there is a need for a tool to measure the impact of interventions in the context of RCTs. Although there are widely used measures of eHealth literacy, these were not sufficiently comprehensive in their scope; in particular they only “worked” for Internet users. Initially the eHEALS questionnaire on eHealth literacy was included within the PERQ questionnaire, but was then dropped as the single self-efficacy question seemed adequate and took less space. There was a need for a measure that covered the full range of individuals from non-Internet users through to frequent Internet users.
This pragmatic scoring system has many limitations, but is published with full details and data so that others can refine or continue to develop it. One problem with any measure of eHealth readiness is continually changing technology such as the current shift to smart phones [
Second, the scale and scoring rely on self-report. Although one dimension of the proposed scale is provision of eHealth services, and although this could be measured fairly objectively [
Third, having four dimensions makes having one summative number for eHealth readiness difficult. As described, the original plan to make the four components additive either in equal proportion or, using the ideas of Paterson et al [
Fourth, the eHealth readiness scale is at best ordinal and not cardinal. This compares to, for example, a difference in mortality that can be expressed as a difference in years of life. While some may argue that 10 years of life at age 20 is “worth more” than 10 years of life at age 70, “years of life” is essentially a cardinal scale. Self-reported questions used to construct an ordinal scale will always have limitations and should be used cautiously.
Fifth, the weights used for individual items were arbitrary. For example, Internet Health Provision included questions about whether patients could access their medical records (weighted 1.5) and could order repeat prescriptions online (weighted 1.0). These weights reflect the judgements of the author in the “difficulty” or “sophistication” of provision. Clearly other weights could be used and the dataset and analysis syntax are provided for others to explore, but this pragmatic approach seems to provide a way of scoring and measuring change.
Sixth, people whose opportunities to use the Internet are less because of limited English will not be identified by this approach, using an English language questionnaire.
Lastly, it is not possible to directly measure eHealth inequalities, but instead to examine the spread of eHealth readiness scores. Those implementing eHealth interventions are interested in improvements in mean score (eHealth readiness) over time, but should also consider the standard deviation of scores (eHealth inequality). In other fields, for example, income inequality, measures such as the Gini coefficient directly measure inequality. Health inequalities are typically measured by differences in mortality, expressed in absolute numbers of life expectancy at birth or some other age between two groups. In this case, we hope that an intervention will improve eHealth readiness, but also reduce the standard deviation. It was important therefore in the construction of this score that a reduction in standard deviation was not artificially induced by a “ceiling effect” on the score. Given the natural progression of the Internet we are unlikely to see eHealth readiness reduce, so “floor effects” are less important. The modelling of possible interventions suggests that the eHealth readiness score is quite sensitive to relatively modest changes in Internet use for health. In determining sample size and setting significance levels, therefore, it is suggested that
Clarity is needed about the role of social determinants of eHealth inequalities. Should associations between eHealth use and demographic and social variables be explored, or should the focus be on the immediate “cause” of eHealth inequality? Demographics are clearly important in use of the Internet [
That the nonresponders in the baseline survey were more likely to live in lower value houses will have biased this sample towards Internet users [
The baseline survey response rate was fairly poor (36%) compared to the 59% achieved by ONS [
More detailed analysis of PERQ results could indicate the most appropriate interventions for individuals or subgroups. For example, groups that would most benefit from faster access, or support, or for whom economics was the main barrier could have interventions chosen appropriately. It is possible that a “stages of change” approach to classifying individuals might be useful, although the different dimensions (personal, provision, economic, support) need to be taken into account.
The support section of the questionnaire was the least successful. This had proved difficult throughout piloting. In particular, we had sought ways of getting those people who had never needed or sought help to answer the questions by wording the questions about “people in general,” and by stressing that we wanted everyone to answer this section. Nevertheless, 21/271 (7.8%) Internet users failed to answer this section. The second part of the “support section” (F) of PERQ probably did not collect particularly useful information, and given the desire to shorten the questionnaire, could possibly be dropped in further developments.
There was previously no measure of personal eHealth readiness or eHealth inequalities. The concept of a patient eHealth readiness based on provision, personal ability, support, and economic considerations with eHealth inequality as the standard deviation seems to “work” and be acceptable in a British context. The scores produced appear valid and sufficiently sensitive to enable assessment of the effectiveness of interventions to improve eHealth readiness and reduce eHealth inequalities. With suggested modifications PERQ is now being used in two other local studies. It could also be used to help identify interventions addressing eHealth readiness. Such methods need continued evolution; full documentation and data have been published to allow others to develop the tool further. In particular with changes to the “provision section,” PERQ might be adapted for use in nonBritish settings.
Initial piloting, more details on baseline survey, questionnaire development (versions 1-4), variables, SPSS syntax, dealing with inconsistent and missing data.
General Practice
Oxford Internet Survey
Office for National Statistics
Patient eHealth Readiness Questionnaire
randomized controlled trial
The author wishes to thank the following people for their assistance with this paper: the reviewers for their helpful comments on the original submission, Liz Baker who delivered questionnaires, entered the data from the questionnaires, as well as recruiting family and friends for one round of the piloting, Paul Hewson for statistical advice, Jeremy Hobart and Michael Hyland for advice on scale construction, Lesley Goldsmith, Emily Ashurst, Tess Lomax, Jane Grose, Julia Frost, Anita Slade, Phil Abbott-Garner, and Jo Atkey for comments on drafts of the paper, and finally various colleagues, family members, residents of Wesley Court retirement apartments, students of Central Saint Martins, and others for the completion of and comments on early drafts of the questionnaire. The literature review, research for this paper, development and early piloting was unfunded. Sentinel Community Interest Company (GP-Co-operative for Plymouth) funded the costs of data collection for the pilot study of the questionnaire. Analysis and write-up was funded by Plymouth University (in terms of RJ’s time).
None declared.