1 Background

People across the globe use technology to conduct many essential and daily functions such as banking, shopping, social networking and information gathering. Increasingly, policy directives and technological innovation are also facilitating online engagement between patients and healthcare providers [1]. Patients are also demanding greater access to their own healthcare data [2]. Typically, this involves using a Patient Access Portal (PAP) that enables patients and/or carers to book appointments, make medication requests, view their medical records and send messages [3].

In the USA, two-thirds of Kaiser Permanente’s 3.4 million members have signed up for online appointment booking, email communication and test results retrieval [4, 5]. Similarly, 600,000 members of the USA’s Veterans Health Administration have registered for online services with over 20 million online visits [4]. In Sweden, recent figures indicate that national e-health services accounts have been set up by a third of the population [6, 7]. There has also been widespread uptake in Estonia and Denmark [8]. In the UK, progress in the National Health Service (NHS) has been somewhat slower, although in April 2015, the Department of Health mandated that all general practitioners (GPs) should give patients free online access to their medical records [9, 10].

A systematic review of online services in primary care [4] found that patients using PAPs reported increased convenience and satisfaction. Positive impacts on patient safety were noted. Professional concerns about privacy were unrealised, and those about workload were inconclusive. The review highlighted the paucity of studies in the British National Health System (NHS) and questioned the relevance of US-based research to the NHS while calling for further research on assessing the values patients place on online services [5]. The current study attempts to address that gap.

Egton Medical Information Systems (EMIS) is a leading primary care computer system which is used in over half (56%) of the general practices in the UK [11]. EMIS has a web-based PAP which went live in 2005 [12]. Subject to availability and approvals from their own practice, patients (or their carers) can register with EMIS to access the PAP to (a) book appointments, (b) submit requests of repeat medication prescriptions, (c) send/receive messages via the portal to clinical and administrative staff and (d) review their own medical record. As part of their ongoing evaluation of the PAP, EMIS designed and administered a survey to registered users, to elicit their perceived value of the PAP. EMIS conducted the initial analysis internally based on their own areas of interest. The portal is not available as a mobile application at present, and so no observations or conclusions can be made about the impact this might have on take-up and usage.

In this study, we undertook secondary analysis of the completed survey data to determine how people used the PAP and their views on its value compared to other online services such as banking and shopping. Perceived monetary value of the PAP was also assessed.

2 Methods

2.1 Design

A service user evaluation of the EMIS PAP was conducted. A cross-sectional survey was designed and administered by EMIS. Surveys were administered between 12 May 2015 and 27 June 2015 via the online Patient Portal feature in EMIS. Registered PAP users from the age of 16 upwards were invited to complete the voluntary (with no incentives) online survey. The survey could be completed by the users who were either the named patient or their carer.

The survey consisted of 17 multiple choice and two free text items (see “Appendix 1”). It captured demographic details including country of respondent, age range (see “Appendix 1”, Part 1, Questions 2 and 3 for ranges) and health conditions (informed by long-term conditions (LTC) listed in the NHS Outcomes Framework [13], see “Appendix 1” Part 1, Question 4). A survey responder was deemed to have an LTC—subsequently summed into four levels (nil, one, two, three+)—if they selected any of the conditions from the survey: being involved in a preventative programme of treatment (associated with any of the following—congestive heart failure, diabetes, asthma, angina, epilepsy, hypertension, influenza, pneumonia, chronic obstructive pulmonary disease (COPD), dehydration or gastroenteritis—all counting as one only), diabetes, epilepsy, breast cancer, lung cancer, colorectal cancer, any other cancer, cardiovascular disease, dementia, liver disease, other long-term conditions, serious mental illness, MRSA or asthma (see “Appendix 2”). Gender, in-country location and ability to make use of individual PAP features were not captured. For example, it was not clear whether individual respondents GP practices had “switched on” all four PAP features (online appointment booking, medication requests, patient messaging, and medical record viewer).

Responders were asked to report if, how and their frequency of PAP use (four possible uses included: appointment booking, medication request, patient messaging, medical record review), perceived value of the PAP versus other online services (banking, grocery shopping, shopping, social networks, search engines), and to assign a monetary value (on an ordinal scale ranging from £0 to £10,000) to the worth of the PAP per annum, and in general, via free text. Responders could also add additional comments in free text. (Free text responses are not reported in this paper.)

2.2 Statistical analysis

Based on findings and/or insights from previous studies [1, 14,15,16,17], we posited that survey responses were likely to be influenced by the reported age group (< 21 years, 21 to < 65 years, 65 to < 75 years, 75+ years), number of LTCs (nil, one, two, three+) and services used in the PAP (online appointment booking, medication request, patient messaging and medical record viewing). We tabulated all the results by the number of LTCs.

To examine the relationship between binary response variables and predictors (age group, number of LTCs and PAP functions used), we used trees which are a class of data mining tools, especially suited to identifying homogenous subgroups based on searching for nonlinear interactions [18]. In particular, we used conditional tree models using the party package [19] in R [20] by specifying p < 0.001 for statistical significance, along with a Bonferroni correction for multiple testing. Tree models are shown in “Appendix 3”. Due to limitations within the survey, we were unable to report the results by gender, UK location or the number of PAP functions available in each practice and patient uptake.

3 Results

3.1 Characteristics of responders

There were 73,044 responses. Responders were excluded from analysis if they lived outside of the UK (3%, UK citizens who maintain registration with a UK GP practice, and PAP, despite country of residence) or had not completed the survey fully (12%, defined as pressing the “Finish” button to submit their answers), resulting in a total number of 62,486 included respondents. Table 1 describes the characteristics of included responders. Included responders were aged 21 to < 65 years of age (66.4% 41,393/62,486), had one LTC (46% 28,718/62,486) and completed the survey themselves (95.6% 59,798/62,486) (see Table 2 for aggregate overview and “Appendix 3” for details stratified by number of LTCs and conditions). Remaining participants with missing data for other survey questions were retained for subsequent analysis.

Table 1 Characteristics of responders
Full size table
Table 2 Overview of uses and frequency of use
Full size table

3.2 Descriptive analysis

Most responders used the PAP monthly (66.3%) and had been using it for one to two years (39%). Medication requests were the most frequently cited function used by responders (86.3%); appointment booking was the second (78.4%). Most responders reported their favourite online service was banking (29.2%), PAP was the third (20%), behind search engines (22.8%). Most responders found PAP to be worth about the same to them as online banking (43.3%) and search engines (30.8%), and worth more than online grocery shopping (56%), online shopping (52.7%) and social media (57.1%). Most responders considered the PAP to have had a positive impact on their life (87%), and most considered it to have monetary value from at least £10 up to £10,000 (69.5%).

Tables 2 and 3 present an aggregate overview, and “Appendix 3” provides details stratified by a number of LTCs.

Table 3 Worth of PAP compared to other online services
Full size table

3.3 Tree models

The trees, all with the same covariates (age group, number of LTCs and PAP service used), are shown in “Appendix 3”.

We analysed the UK-based responders who had completed the survey (86% = 62,486/73,044, Table 1) stratified by the number of LTCs.

3.3.1 Self-completion of the survey

Age and the number of LTCs were significant predictors of self-completion; however, PAP service used was not (see tree model one). The lowest self-completion rate was amongst those under 21 (58.8% n = 1440), and the highest was of those aged 21 to < 65 years (98.8% n = 41,328) and 65 to < 75 years (96.9% n = 13,920.

3.3.2 Usage of the PAP: frequency and purpose

The reported usage included booking appointments online, medication requests, messaging and medical record viewing. Patient age, the number of LTCs and purpose of usage were significant predictors of usage patterns and age and the number of LTCs affected types of usage (see tree models two to six in “Appendix 3”).

3.3.3 Favourability and positive impact

Favourite online service was modified by PAP service used and age group. Users with two or more LTCs and using the messaging service were most likely (30.4%) to rate the PAP as their favourite online service, whereas users with nil or one LTC not using the messaging service were least likely (17.9%) to report the PAP as their favourite (see tree model 7).

Perceived positive impact of the PAP was modified by which service was used and the number of LTCs (see tree model 8). The positive impact was reported at its highest for those using the PAP for medication requests and online appointment booking (range 71.7–77.6%) though this was dependent on the number of LTCs, with responders with more than one LTC being more likely to report a positive impact. Those least likely to report a positive impact were not using medication requests (51.6%).

3.3.4 Comparability with other online services

The percentage of responders who found the PAP more valuable than other online services increased with the number of long-term conditions (see Table 4). Patient age, number of LTCs and purpose of usage predicted value ratings (see tree models 9–13 in “Appendix 3”).

Table 4 Monetary value (£) of the PAP per year
Full size table

3.3.5 Monetary value

A monetary value of more than £0 per annum was higher for those with medication requests and more than three LTCs (see tree model 14). This was with or without online booking (yes 82.1% n = 30,748; no 75.1% n = 9885). Also awarding it high value were those with no LTCs but using medication requests and online booking (76.7%, n = 10,424). Those least likely to value the PAP at more than £0 per annum had no LTCs, were not using online booking but still having online medication requests (67.1% n = 2851), and those not having online medication requests aged 21 to < 65 years (67.1%, n = 6466).

3.3.6 Changing doctors and the PAP

The likelihood of the availability of a PAP influencing the decision to change GP practice is shown in tree model 15. This comprised 72.3% (45,176/62,486) of respondents. This increased with the number of LTCs (range 68.7% nil LTCs to 77.4% three+ LTCs). Those most likely (82.6%, n = 8984) to respond that they would be influenced by PAP availability were those who used online booking, medication requests and medical record viewing. Those who were least likely (59.5%, n = 3268) to be influenced did not use online booking and had no LTCs.

3.3.7 Overview of significant/non-significant factors for binary response variables

Table 5 shows which factors are significant and non-significant predictors of the range of binary response variables considered in the fourteen tree models. The number of LTCs was a significant predictor in all bar one tree models (use of PAP for online messaging) as was the use of PAP for online booking in all bar two (self-completion and frequent use). The use of the PAP for medication requests was a significant predictor in all bar four tree models and patient age group in all bar five. The use of patient messaging was a significant predictor in four tree models and the use of the PAP for medical record viewing in three tree models.

Table 5 Binary response variables
Full size table

4 Discussion

In a relatively large cross-sectional survey of NHS users of an online PAP in the UK, we found that they use it mainly for medication requests and online appointment bookings, and, to a lesser extent, medical record viewing and messaging. The vast majority of responders reported a positive impact from using the PAP. One in five rated the PAP as their favourite online service second only to online banking. Almost three out of four responders stated that availability of online access would influence their move to another general practice. Nonetheless, responders were reluctant to award a high monetary value to the PAP although this varied according to the number of LTCs: those users with a higher number being more willing to award value.

While the survey does not ask respondents to detail or elaborate on what they count as positive impact, we would suggest that such an articulation may be consistent with previous studies which report that users view online access to medical records as a positive development [4, 5, 21,22,23,24,25,26]. Overall, users see online access as “convenient, useful, usable and flexible” [23]. A 2014 audit of two UK primary care practices [23] who provided online access to a PAP found that users (n = 226) reported savings in time and money (from reduced travel and telephone calls) and being more active in the management of their health. This related to reviewing records online and recalling past consultations, often resulting in identifying omissions and discrepancies of information or outcomes of a consultation, and feeling more prepared for future consultations. Kruse et al. [27] in their systematic review of the effect of PAPs on quality outcomes found that patient satisfaction and patient retention were increased while Sorondo et al. [28] found improvements in mental wellbeing and lesser healthcare utilisation for patients with chronic conditions. Otte-Trojel et al. [29] suggested that PAPs—as a supplement to existing services—can result in improvements in patient behaviour and experiences, as well as clinical outcomes.

We found that responders did not place a high monetary value on the PAP. This finding aligns with studies from the US which also found that users were unwilling to pay for online services or were unwilling to pay more than a small sum [4, 5, 30]. This is despite a fundamental difference in the way in which healthcare services are paid for (cost to the patient (USA); free at the point of need (UK)). So, while users report increased satisfaction as a result of access to online services, this does not translate into a willingness to pay. We see this purely as a symbolic expression regarding value rather than one based on real monetary exchange.

We found higher usage of the PAP for medication requests and online booking of appointments compared with messaging and medical record viewing (although again it must be noted that we did not have access to data that identified enabled functions in each practice or of patient uptake). This ranking was also seen in a recent national UK GP Patient Survey (July 2017) which found that 11.8% of patients used a portal for medication requests, 8.9% for online appointment booking and 1.6% for viewing medical records (electronic messaging was not included in that survey) [31]. The same survey reports that 46.3% of patients are unsure of what online services their practices offer. The uptake of online services and the differential uptake of specific services raise important questions about access to such services. The general notion that users are not motivated or are incapable of using such services is not credible [24], suggesting, therefore, that other factors are constraining access to online services which need to be addressed. If a key constraining factor is usability from the users’ perspective, then a way forward could be to involve them in the future design of online access systems [22, 24, 32]. If, however, the constraint emanates from the healthcare provider, perhaps as a consequence of concerns about increased workloads, then this may be misplaced as the evidence is both scarce and contradictory [7, 33,34,35].

4.1 Strengths and limitations

Although this was a large study and one of the few that has taken place in the UK, the findings are based on the experience of those people using the PAP offered by one, albeit leading supplier of primary care computer systems, who also designed and administered the survey. In addition, although we did not have the total number of registered users or the number of GP practices that used this supplier during the time period surveyed, we used data reported by Kontopantelis et al. [11] to note that EMIS has a 56% market share and that this has remained the same since 2011. This is equivalent to 4199 practices and 32.2 million users (although obviously not all users are signed up to use the portal [36]). It must be noted, however, that this is clearly only a small proportion of the potential patient population: approximately 46 million [37]. We did not have access to data about the extent of PAP functions available, ease of use and patient uptake. This means that we do not know what services are enabled in each GP practice or of the influences which may have affected patient decisions about just what aspects of the PAP they chose to use or not use. Gender and other social characteristics were not available to us, yet evidence from other studies [4, 21, 24, 25, 38] suggests that these are important and that such factors—socioeconomic status, gender, race, social vulnerability—may affect access and usage. There is evidence to suggest that IT usage and acceptability may vary between participants within this category [33,34,35] most specifically in those aged 45–65 years [39]. However, trends may be changing with the proliferation of technology; for example, the latest Oxford Internet Institute study on internet usage indicates that gender is no longer a significant determinant [39].

We were, however, able to consider responses in terms of LTCs as well as the services used in the PAP. This is important as evidence from Newhouse et al. [34] suggests that as health conditions worsen the use of email by patients—where available—for healthcare communication increases. They found that email use, poor health, multimorbidity and number of physician visits are positively correlated. In addition, Palen et al. [33] found that where access to clinicians and to medical records was available, this resulted in an increase in usage of clinical services generally. This was consistent across age and chronic conditions.

Our findings will be of interest to a wide range of people—users, policy makers and healthcare professionals—and offer a timely patient-based perspective to inform ongoing attempts to enhance online services in the NHS [3]. The extent to which increased online access could mitigate the downward trend in patient satisfaction with access to GPs [31] also merits consideration, especially as 80% of general practices in England had the capability of giving the general population access to their medical records in 2014, but by 2017 only 8.9% were aware they could access their records and only 1.6% reported doing so [31]. By contrast, internet banking usage has increased from 30% (2007) to 60% (2016) [40]. Without trivialising the differences between banking and healthcare (for example the extensive advertising of online services in the banking sector), or overlooking negative consequences of online access, this striking difference in uptake of online services highlights how the former has embraced online services while the latter has lagged behind. The range of factors which influence adoption of online services is broad and includes: (1) concerns about privacy and security of data (although this does not seem to a barrier in the general practice context [4, 22]); (2) the notion that online access may increase health inequalities through uneven digital access and literacy especially amongst the elderly [3]; (3) the extent to which online access can support self-care and integration with other health and care data; (4) the extent to which online access is easy to obtain and use; (5) the extent to which online access is integrated into care processes and mitigate possible negative consequences for some patients [3, 4].

5 Conclusions

Most users place a relatively high value, though not monetary value, on the PAP and report a positive impact from using it. The potential for PAPs to enhance patient experience, especially for those with long-term conditions, appears to be largely untapped. Research and development addressing the reasons for non-use are also required.