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 the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Online access to all or part of their health records is widely demanded by patients and, where provided in form of patient portals, has been substantially used by at least subgroups of patients, particularly those with chronic disease. However, little is reported regarding the longer-term patient use of patient-accessible electronic health record services, which is important in allocating resources. Renal PatientView (RPV) is an established system that gives patients with chronic kidney disease access to live test results and information about their condition and treatment. It is available in most UK renal units with up to 75% of particular patient groups registered in some centers. We have analyzed patient use out to 4 years and investigated factors associated with more persistent use.
Our aim was to investigate RPV use by patients over time from initial registration in order to understand which patients choose to access RPV and the endurance of its appeal for different patient groups.
We analyzed an anonymized extract of the database underlying RPV containing information on patient registration and events including patient access and the arrival of new blood test results or letters that patients might wish to view.
At the time of the extract, there were 11,352 patients registered on RPV for 0-42 months (median 17). More than half of registrants became persistent users, logging in a median of 2.0 times each month over post-registration intervals of up to 42 months (median 18.9). Provision of assistance with first logon was strongly associated with becoming a persistent user, even at 3 years. Logons by persistent users occurred around the time of consultations/tests, strongly suggestive of patient engagement. While indices indicative of greater deprivation were the strongest determinants of non-participation, they had negligible influence on drop-out rates among established users.
In this mature patient portal system, a large proportion of patients made regular use of their online health records over protracted periods. The patterns and timing of use indicate strong patient interest in detailed information such as recent test results and clinic letters. Supporting patients through the first steps of establishing access to their online records is associated with much higher rates of long-term use of RPV and likely would increase use of other electronic health records provided for patients with chronic disease.
Patient-accessible electronic health records (EHRs) could have many advantages, but patient enthusiasm and use have been variable [
Renal diseases are mostly chronic conditions and sometimes progressive, so dialysis or renal transplantation is required. Patients with progressive or end-stage renal disease are under regular specialist supervision for the rest of their lives. They may therefore have a strong incentive to become interested in the monitoring and treatment of their condition. In particular, blood tests change in important ways in kidney disease and are typically undertaken every 3-12 months in CKD patients and patients with stable transplants, and monthly in patients receiving center-based hemodialysis.
Renal PatientView (RPV) was established in 2004 to provide information to patients with chronic kidney disease (CKD) (see the Renal PatientView website for a demo [
Patient and staff responses to RPV have been quite positive (in the case of staff, this is despite common early hesitancy) [
In order to better understand variation in RPV uptake and persistence of usage, we analyzed the registration and access log data from the RPV server log files collected over nearly 4 years.
RPV data extant on September 7, 2009 (the RPV census date), were stripped of patient identifiers and transferred to a research database. Subsets of the data were extracted to R for statistical analysis (the R Foundation for Statistical Computing [
The dataset comprised patient factors listed in
Center factors were defined to capture possible influences from the centers providing renal services to patients. Centers were classified according to the date RPV was first offered to patients, the proportion of the Renal Replacement Therapy (RRT) population recruited to use RPV, and their provision or not of extra support to patients registering with RPV for the first time (assisted start).
Patient factors.
Patient factors | Notes | ||
Age in years | <18, 18-34, 35-54, 55-74, >75 |
|
|
Gender | M/F |
|
|
|
|||
|
Hospital HDa | Includes most dependent patients | |
|
Home HD/PDb |
|
|
|
Transplant |
|
|
|
Not on RRTc |
|
|
|
|||
|
By decile | 1-10 | 1 is most deprived |
|
|
||
|
|
High | Deciles 1 and 2 |
|
|
Middle | Deciles 3-8 |
|
|
Low | Deciles 9 and 10 |
Access log data | Dates and times of every logon |
|
|
Blood test results | Sample dates and values |
|
|
Center | Unit code | UK renal registry code of treatment center |
aHD: hemodialysis.
bPD: peritoneal dialysis.
cRRT: renal replacement therapy, so HD, PD, or transplant.
dDeprivation measures: see text.
For some studies, patients were also classified by treatment in order to enable comparison with an appropriate overall UK population. This enabled comparison of adult RPV registrants in receipt of renal replacement therapy with the overall UK adult population in receipt of renal replacement therapy at the time of the RPV census as recorded at the UK Renal Registry report 2009 [
Classification of RPV users by logon analysis.
Statistical analysis used the R framework [
Survival (of RPV use) analysis employed the survplot program [
The overall intensity of use was computed by counting, for each patient who completed at least 1 logon, the total number of logons made and dividing by the number of months between registration and census to compute logons per month. The distribution of the logarithm of logon activity across the patient population was approximately normal (see Results), enabling patients to be classified by the quartile (Q1-Q4) that their individual log (logons per month) fell in the overall log (logons per month) distribution.
The timing of logon events in relation to test results was explored first with a plot devised to show the activity of individual patients as series of colored dots along a horizontal timeline. The activity of approximately 100 patients could be plotted without obscuring individual patient events, so plots show random selections of patients from appropriate subgroups ordered by the duration of logon activity. The overall tendency of logon events to occur close in time to test result events was explored by computing for every logon event the interval to the closest result event and constructing a histogram to show the proportion of logons by interval to closest test date. Significance was assessed by comparing histograms made with randomly shuffled logon times or test times.
At the time of the RPV census, 11,352 patients had been registered from 37 of the United Kingdom’s 73 adult centers. The characteristics of the 11,352 registrants are shown in
Description of the entire RPV-registered population (N=11,352).
Characteristic | Proportion, % | |
|
||
|
<18 | 1.6 |
|
18-34 | 11.6 |
|
35-54 | 37.9 |
|
55-74 | 38.9 |
|
>75 | 10.0 |
|
||
Male | 59.9 | |
|
||
|
Hospital HD | 19.4 |
|
Home HD/PD | 6.7 |
|
Transplant | 32.8 |
|
Not on RRT | 41.0 |
Description of the RRT-dependent subgroup (n=6646) of the RPV-registered population compared with the overall UK RRT population by age, gender, and treatment (N=44,649)a.
Characteristic | Proportion registered for RPV of all UK RRT patients, % |
|
|
|
|||
|
18-34 | 16.1 |
|
35-54 | 16.6 | ||
55-74 | 14.8 | ||
>75 | 8.4 | ||
|
|||
|
Male | 15 | Not significant |
Female | 14.6 | ||
|
|||
|
Hospital HD | 10.5 |
|
PD | 17.7 | ||
Transplant | 16.6 | ||
|
|||
|
|
14.8 |
|
aSignificance of comparisons estimated by Pearson’s chi-square test statistic.
RPV registration by patient deprivation. The proportion of adult RRT patients registered for RPV is shown by patient rank deprivation (1-10, 10 is least deprived) for patients with postal codes in Scotland (filled diamonds) or England & Wales (open diamonds).
The penetration (proportion of adult RRT patients registered) by center ranged from 6-75% (median 32) for centers enrolling at least 6 months. The variation was only in part explained by differences in the duration of recruitment effort (6-48 months) (
RPV registration by center. Number of registrants (top chart) and proportion (%) of available RRT patients registered for RPV is shown by renal center, ordered by the start date of patient enrollment. The duration of active RPV enrollment is superimposed on the bottom chart; duration ranged from 0.5-4 years.
About one-fifth (23.20%, 2634/11,352) of registered patients never logged in despite having signed up and having been sent details. Logistic regression indicated that the strongest registrant-specific effects (up to 2-fold) were age and treatment group (
The effect of deprivation was also very significant. Registrants from addresses associated with greatest deprivation were more likely to not complete first logon (OR 1.24, 95% CI 1.08-1.42), whereas registrants with addresses associated with low levels of deprivation were less likely to not complete first logon (OR 0.79, 95% CI 0.65-0.97), both compared with registrants of middle rank deprivation, indicating a negative influence of deprivation beyond that on becoming registered, as described in other chronic disease populations [
Patients at centers that began offering RPV in the 2 years prior to the census were twice as likely to not complete first logon (OR 2.04, 95% CI 1.41-2.93) as patients at centers that had been offering RPV for more than 2 years, which seemed greater than the lag to be expected for recently enrolled centers to build up recruitment. This is a subject of continuing investigation.
Odds of not persisting with RPV use at proposed hurdlesa,b.
|
Odds not completing initial logon (N=9552), |
Odds lapse early (N=822 vs 7427), |
Odds lapse late (N=1401 vs 6023), |
|
|
||||
|
<18 | 1.81 (1.27-2.55) | 1.02 (NS) | 0.76 (NS) |
18-34 | 1.32 (1.12-1.56) | 1.21 (NS) | 1.03 (NS) | |
55-75 | 0.98 (NS) | 1.02 (NS) | 1.11 (NS) | |
>75 | 1.41 (1.19-1.66) | 1.46 (1.12-1.90) | 1.58 (1.27-1.96) | |
|
||||
|
Pre RRT | 1.0 (NS) | 0.83 (NS) | 0.37 (0.31-0.44) |
Home HD/PD | 0.88 (NS) | 0.96 (NS) | 0.68 (0.53-0.86) | |
Transplant | 0.60 (0.52-0.69) | 0.61 (0.49-0.75) | 0.45 (0.38-0.52) | |
|
||||
|
Greatest | 1.24 (1.08-1.42) | 1.73 (1.40-2.13) | — |
Least | 0.79 (0.70-0.89) | 0.79 (0.65-0.96) | — | |
Unit offering RPV <2 years | 2.04 (1.41-2.93) | — | — | |
Unit registration rated in Q3 (Q2 also significant) | 0.79 (0.65-0.97) | — | — | |
Unit offering routine assisted start | 0.31 (0.21-0.46) | — | — |
aSummaries of models obtained by logistic regression for the likelihood that patients choose not to persist with use of RPV at the three decision points: (1) choosing not to complete first logon, (2) having made an initial logon choosing not to logon on again beyond 1 month (lapse early), and (3) discontinuing logons at some later time (lapse late).
bOdds ratios are shown for the influential factors followed in brackets with 95% confidence intervals or NS if the interval spans 1.0. Factors marked with a dash (—) were removed because of insignificant effects in the indicated model. Gender had insignificant effects in all models.
cAnalysis restricted to the 9552 (of 11,352) registrants with complete data. The major reason for exclusion was missing treatment type as a result of this parameter not being recorded by one major center (1009 registrants). Alternative analysis including these 1009 registrants and excluding treatment as a factor did not for any other factor change the assignment of significance and only slightly altered the ORs.
dCenters were grouped in quartiles of percentage enrollment of RRT patients as a measure of a center’s effectiveness in recruiting patients to RPV.
Interval from registration and likelihood of completing first logon. A: the median interval in days between registration and initial patient logon by renal center. The size of markers is proportional to total number of registrants at each center. Two centers are remarkable for completing the process within a day for most patients (87% of patients at the larger center). B: Probability of completing first logon by age. A logistic (completed first logon = T/F) general additive model (non-parametric) was constructed using the mgcv package to model the likelihood of patients completing first logon by age at time of registration without assumption as to the shape of any relationship.
The interval between registration and first logon was investigated because it was thought likely to be a marker of patient enthusiasm. However, the interval was dominantly influenced by center (
There was substantial variation in both the frequency and persistence of RPV logon activity subsequent to first logon. As some registrants had been registered only shortly before the RPV census such that their logon activity was too brief to be assessed, analysis of subsequent logon activity was restricted to the 8249 registrants who completed first logon more than 3 months prior to the census. Three broad patterns were distinguished (
Factors that might influence subsequent logon behavior were investigated first by logistic regression. Early lapse was associated with age over 75 and greater deprivation and was less likely in transplant recipients (
Survival analysis was used to better understand lapses in use of RPV over time, plotting probability of continuing use by time for users partitioned by relevant factors (
The substantially greater early use observed in users at centers offering assisted starts was succeeded by a slightly greater rate of attrition out to 2.5 years leading to some convergence of the survival curves, but the beneficial influence remained highly significant.
Probability of persisting use of RPV at intervals after registration shown as survival (of RPV use) classified as shown. Shading indicates 95% confidence limits.
An immediate peak of very frequent logons was observed, which settled to a slowly declining average rate of logons. This concealed a highly variable rate of logons for individual patients, with periods of intense activity associated with periods of more frequent test results. These periods are likely to be times of medical uncertainty; for instance, some could be around transplantation or other hospital admissions.
Frequency of use overall was assessed from the number of logons made by patients divided by the interval between first and last logon. This analysis was limited to the 5808 users who had made at least three logons over at least 3 months after first logon; 87.55% (5085/5808) were persistent users and 12.45% (723/5808) late lapsers. The median number of logons per month was 1.8 (2 for persistent users, 0.94 for lapsers) with a markedly skewed distribution (range 0.08-84) (
The patients exhibiting the most frequent logon activity (top 5%) were also analyzed separately. None of the factors available was associated with this extreme pattern of use.
Most logon events occurred on weekdays and between 08:00-22:00; in particular, Sunday logon was not popular (
The time of logons to RPV was also examined in relation to the most proximate blood tests because surveys had suggested that accessing blood test results was a major reason for using RPV. First logon events were plotted against interval since first logon to visualize the patterns of logon activity. Examples are shown in
Logon activity. A-F: Patients classified by quartile of log (count of logons per month) (defined in panels A and B as described in the Methods), and comparisons made of the partitioning of Q1-Q4 activity patients by the factors shown. The even spread by gender contrasts with that by age (C) and treatment type (E). G-H: Histograms of counts of logons made by day of week (G) and time of day (H).
Relationship between times of patient logons and blood test results. A, B: Individual logon events shown by interval (in weeks) since first logon for randomly selected lower (A) and higher (B) frequency transplant recipient users where each selected patient’s activity is depicted by a horizontal series of colored dots. Most logon events are red indicating they occurred less than 1 week after a new blood result event. Superimposed is the proportion (% on the right hand axis) of patients in the respective activity groups that log in each week post first logon. C: Histogram (1 day bins) showing the predominance of logons occurring 1-3 days after tests is strikingly different from that observed with randomly shuffled data.
It is striking and encouraging for advocates of patient-accessible EHRs that over half of registrants became persistent users (median of 2 logons per month) with only a small proportion (about 4% per year) discontinuing regular use out to 3.5 years. More persistent use was expected as CKD is a chronic condition, but the rate is much higher than the 26-33% reported in other studies that have assessed use by patients with chronic diseases (eg, [
Blood test results are important in the continuing management of CKD informing, for example, the need for and effectiveness of dietary prescriptions as well as the progression of kidney failure or transplant performance, and many patients take a close interest encouraged by their doctors. Provision of timely blood test results was a core design goal for RPV and was identified as a key attraction for users in a small survey [
Analyzed as a group, after an early phase of frequent logons, during which patients may be gaining familiarity with the system and learning from the information links provided, the frequency of logons by persistent users settles to a median of 2 logons per month with most logons during usual waking hours and the working week (
It is well recognized that initial authentication of new users is a substantial barrier for some patients. In the current data, 20% of patients never made a single logon despite having completed a registration form, returned it to the center administration, and become registered. The first logon procedure for new RPV users is explained in the registration letter and requires navigation of an internet browser to the RPV website, entering the temporary credentials and creating a new password. An important observation from the current data is that providing additional support targeted just at this first logon hurdle has a profound and long-lasting positive influence on patient use. Two centers adopted the practice of providing initial support in the form of face-to-face or over-the-phone guidance through the first logon procedure, with about a 20% increase in the probability of patients’ continuing use of RPV at 3 years (
Inclusiveness is a concern in digital service provision [
Level of deprivation was strongly associated with RPV registration and initial use. This is a new observation with regard to RPV but was anticipated from the results of many studies of access to EHRs and health care information on the Internet [
Renal PatientView attracts strong and sustained use by many renal patients in the United Kingdom. The current data indicate that prevalent center and patient factors underlie considerable variation in RPV uptake and use. These require further study, but it is already clear that centers should consider reviewing their approach to recruiting and supporting patients and possibly identify patients that are not using RPV post-registration to offer targeted help.
chronic kidney disease (irreversible, often progressive kidney damage)
hemodialysis
electronic health record
peritoneal dialysis
Renal PatientView
renal replacement therapy (such as dialysis or kidney transplant)
This investigation was partly supported by the SUSTAINS (Support Users to Access Information and Services) project to enhance patient access to electronic health records in Europe.
None declared