Consistent with a realist evaluation the findings were conceptualised as CMO configurations; the circumstances and ways in which the EMOS was used were perceived to lead to a number of medication safety outcomes. These CMOs were organised into three groups based upon the ways the system was utilised: access; engagement or disengagement with the system; the monitoring of prescribing; and work practices. Within each group we identified mechanisms, and contexts within which these mechanisms were activated, that led to given medication safety outcomes such as patients’ electronic health records being screened to identify potentially hazardous prescribing events.
Engagement and disengagement
The first group of CMOs concerned access, engagement and disengagement (Tables
2 and
3). In the first of these CMOs, the EMOS focused healthcare users’ attention on medication rather than on disease.
Table 2
Context-Mechanism-Outcome configurations concerning access and engagement
General Practitioner monitoring individual patients | Focuses attention on medications | Attention focused on patients most in need of review |
General Practitioner prescribing audited and monitored in practices | Proactively conducting own audits | Practice prescribing patterns benchmarked against each other across the Clinical Commissioning Group |
Communication between Clinical Commissioning Group and General Practitioner | Real time feedback | Patients reviewed to ensure appropriate monitoring, to optimise medications, or to avoid dangerous combinations of drugs |
Clinical Commissioning Group conducting searches of prescribing based upon “projects” and “initiatives” | Prescribing patterns and trends benchmarked against national targets and guidelines |
Clinical Commissioning Group encouraging clinicians to be engaged in more proactive safety management | Engagement of practices in using the system for feedback Voluntary engagement by clinicians Audits conducted as a means of support to General Practitioners | The effectiveness of safety initiatives audited more quickly Improved engagement with safety monitoring of prescribing |
Table 3
Context-Mechanism-Outcome configurations concerning disengagement
Communication between Clinical Commissioning Group and general practitioners | Feedback on alerts requires logging in | Potential delays in patients being reviewed |
Reliance on alerts being sent out centrally |
Information technology use in General Practice | Lack of use/not logging in to the system | Potential delays in review of patients |
Community pharmacists conducting medicine use reviews with patients | No access to additional information | Opportunity for more appropriate and directed medication review lost |
Community Pharmacy | Perceived conflict and lack of ownership | Limits potential improvements in quality of care for patients |
Patients using the electronic medicines optimisation system | Facilitated use by healthcare professional | Lack of direct access to information to benefit shared care and self-management |
Difficulties obtaining passwords and logging on |
“Say you are monitoring renal function and you look and the eGFR [patient’s filtration rate] has gone down to 29 and it was 31 the month before. You’re thinking, well that’s okay, we’ll just monitor that, you fail sometimes, […] one fails to think, ah, I need to review the allopurinol, I need to renew the metformin, because it is so, so, easy to focus on a disease and that’s, I think, where Eclipse can come in. (GP1-INT)
Engagement with the system by GPs could therefore lead to more focused patient reviews. The system could be used for feedback, giving them “some idea as to who’s perhaps even more engaged than others” (CCGP2). If activated, this mechanism could “inform the CCG about how well safety initiatives are happening” (CCGP2) and lead to a speedier audit and feedback of safety initiatives rolled out centrally.
Increased engagement with safer prescribing could be sustained by voluntary engagement with the EMOS on the part of the practices; this was said to reduce a
“big-brother” (GP1-INT) relationship with the CCG, challenge the belief that it was a tool primarily for the CCG pharmacists, and give a greater sense of ownership of the system within general practice. However, GPs could instead end up relying upon the medicines management team to send out alerts, disengaging them from proactively using the system and reinforcing CCG ownership. Engagement was to be encouraged financially in the future by building a requirement to use the EMOS into the
“prescribing incentive scheme” (CCGP2). In contrast, engagement was discouraged by blocking mechanisms in the context of IT use in general practice. One GP (GP2) stated that they and only one other colleague used the EMOS. Such task allocation meant that within their practice they operated as a prescribing lead where they took responsibility for auditing and monitoring the prescribing within their practice, and therefore were the only ones expected to use the system. Another GP remarked, in terms of seeing alerts in the system,
“I don’t commonly open the software full stop” (GP4) a barrier that was related to time pressures:-
“a third of my time (is) seeing patients, two-thirds of my time doing paperwork and an extra mystical 10 or 20% of time […] Eclipse fits into that last 10, 20% of time that doesn’t really exist.” (GP4)
Other stakeholders were also disengaged from the system. Community pharmacists were aware of it and perceived potential benefits involving increased information through access to care records that could inform medicines use reviews, improve communication with GPs and “influence a decision to sell a medicine or supply a medicine.” (CP3). However, they had not been given access by the CCG, nor had access through the “Patient Passport”, though such access had been planned. This was attributed to perceived difficulties with sharing information, issues of confidentiality and a perception that “GPs often see themselves as the custodian of the patient record” (CP3) which meant “historically a barrier to sharing that information” (CP3). Community pharmacists had been involved in patient passport initiatives that could have given them access to the EMOS but issues of confidentiality, delays and poor communication with the CCG and general practices had led to them being denied access.
A limited number of patients had access to the EMOS through the patient passport. They saw this as potentially valuable in giving access to information about medications and their conditions, which would in turn have a positive bearing on self-management and shared care. However, this was prevented by a blocking mechanisms concerning access,
“The first problem I had was I couldn’t log in at all” (P2). Patients also felt that they would get best use out of the system if this was facilitated and interpreted by a health professional.
“I think that’s why it’s important to, it’s not just to be used on its own, it’s to be used with, to be used with a clinician of some kind to actually help you to interpret some of that stuff, because some of it is, I mean when you look at high haemoglobin levels or the glucose levels, […] Which are the bad ones? Which is this? What does this mean? “(P1)
The monitoring of prescribing
The monitoring of prescribing across general practices (see Table
4) was undertaken by pharmacists and GPs placed centrally at the CCG.
Table 4
Context-Mechanism-Outcome configurations concerning the monitoring of prescribing
Clinical Commissioning Group engagement with prescribing alerts | Alerts designed and results fowarded to practices | Prescribing patterns and trends benchmarked against national targets and guidelines |
Identify specific patients | Pre-emptive or timely review of individual patients |
Monitoring prescribing by conducting searches based upon local “initiatives” | Efficient use of time | Prescribing patterns benchmarked across the Clinical Commissioning Group |
Highlight suboptimal prescribing | Reduction of knowledge gaps to optimise use of medicines |
Reward good practice | |
We identified two contexts within which mechanisms were activated. The first of these concerned the engagement with prescribing alerts issued by the CCG. Alerts that related to the implementation of national guidance were designed and disseminated to general practices. These allowed for bespoke searches of prescribing data to be run across all general practices within the CCG. This in turn allowed for benchmarking against criteria set by national guidelines. One respondent (GP1-INT) acknowledged that the existing alerts embedded within the system could be used, but that they were unwieldy because of their large number so were not commonly used. Similarly, one CCG Pharmacist (CCGP2) said there was a lack of confidence in these alerts, because of a lack of knowledge about the content of the underlying algorithms used to generate the existing alerts, so they were seldom used. The engagement with prescribing alerts also allowed the activation of a mechanism for identifying specific patients, which was seen as more likely to lead to a timely review of patients.
“You [can] pin [the alert] to [specific patients]. So if you say […] metformin shouldn’t be prescribed with an eGFR less than 30 and these are the patients who you need to consider in this category it’s such a more meaningful event.” (GP1-INT)
The second context concerned the CCG setting up their own searches based upon local initiatives. Within this context one mechanism allowed for searches to be conducted speedily across all practices within the CCG. This was a change in working, where in the past “trawling round all […] practices” (CCGP2) had “[taken] us about three to four weeks” (GP1-INT). Since the introduction of the intervention, “we ran the same search and literally […] 90 min without actually leaving your desk, you’ve got the results” (GP1-INT). Using the system helped to identify prescribing patterns and “to have the ability to look at the prescribing by practice […] so we could compare […] the prescribing of a drug one practice to another” (GP1-INT). Participants saw this as leading to prescribing patterns being benchmarked across the CCG. Additionally, the EMOS was seen as an educational tool that could reduce knowledge gaps and change prescribing behaviour by highlighting suboptimal prescribing within and across practices “because we could identify those patients receiving whatever strength, notify GP within the system and […] got 100% adherence to this safety thing”(GP1-INT). This educational outcome was further enhanced by rewarding good practice: “if there are some practices that are demonstrating very good prescribing, then we’ve picked those out as well and highlighted those” (CCGP2).
Work practices
The final group of CMOs concerned the effect of the EMOS on work practices (Table
5). This involved a number of different stakeholders in general practices: GPs; practice managers; and practice-based pharmacists.
Table 5
Context-Mechanism-Outcome configurations concerning work practices
Multiple administrative work practices | Logging on, responding to alert, and reviewing patients through the system | Patients reviewed to ensure appropriate monitoring, to optimise medications, or to avoid hazardous combinations of drugs |
Pre-existing division of labour within General Practices | Task allocation |
General Practice workload | Task Prioritisation | Pre-emptive or timely review of individual patient |
Pharmacist workload | Existing work practices developed and adapted | Can result in a more focused medication review |
Pharmacist undertaking reviews in care homes | Accessing easily readable and informative data |
Necessary workarounds to overcome technical issues |
Necessary workarounds to find patient details |
The first context here concerned administrative work practices. Some practices relied on alerts being sent to them by email rather than proactively seeking the alerts by logging on to the EMOS. The process of responding to alerts varied, but often involved transferring information from email to paper in addition to logging on to the system, causing a delay.
“The alert is printed off on a piece of paper which [then] sits in my in tray with 500 other items of equal urgency, and […] it might be that I have to work my way down through that pile over a period of a few months.” (GP4)
Reviewing the patient through the system was a more successful mechanism that gave immediate feedback to the CCG, avoided the delays, and provided clear and speedily accessible information in a readable form where: “you can plot the graphs [and] quickly eyeball 100 patients in a couple of minutes.” (GP1-INT)
Within the context of pre-existing divisions of labour within practices, EMOS was seen to require a specific task allocation which would be
“certainly led by a clinician and most likely performed by a clinician” (GP4). There was variation in the ways the EMOS was used by either practice managers or GPs. One practice manager said that once an alert was received they took responsibility for it:
“I pass it on to the GP and get them to respond to me, and then I update Eclipse […] the doctor’s don’t access it at all” (GPM2)
Whereas in another practice the responsibility for accessing the system was the GP’s:
“
The GP actions it, I don’t have any more responsibility for it after that […] They go into Eclipse, they do it, […] I had to remind one GP today, I just wanted to check they had actually reviewed this patient” (GPM1)
If the system was used effectively then patients would be reviewed but, as noted by the general practice manager above, it was possible that the task allocation could act as a blocking mechanism (that is, inhibiting the effect of the system) if GPs had to be reminded to review patients.
Within the GP workload context, mechanisms associated with task prioritisation could lead to the timely review of patients. To utilise the system effectively, GPs had to juggle competing tasks and prioritise. If GPs were “getting pertinent alerts that they feel are relevant” these alerts were seen with “virtually no negativity.” (GP1-FG)
For pharmacists undertaking medication reviews in care homes, the system saved time by giving more speedy access to information,
“there and then in front of you” (CCGP1) allowing for a more focused review. The system gave the pharmacist the opportunity to send recommendations to the GP based on information about medications, test results, conditions and demographic factors. This information was easily accessed through the EMOS and findings easily interpreted.
“The benefit of Eclipse is you can log on and look at the graph and you can see the basic trend of blood pressure, of cholesterol, of weight et cetera, on a beautiful graph which is so easy to read with the red/amber/green bits, it’s so clear what’s going on.” (CCGP1)
Effective use of the system required some adaptations and improvisation on the part of the users. For example participant CCGP1 whilst carrying out tasks in a care home, had to adapt ways of obtaining passwords for the system to deal with limited internet access. Pharmacists “beforehand were trying to look up all the stuff on Eclipse whilst we were in the care home” (CCGP1) but had adapted their activities in order to have “more information to start off with (and) use Eclipse for less time in the care home, but in a more directed manner” (CCGP1). Limitations to the information available in the system, necessitated workarounds in order to obtain further patient details; “because it doesn’t list actual allergies” (CCGP1) and “we can’t look at letters” (CCGP1). This meant finding out more information from the general practices before the visit to the care home or returning to general practices to obtain “any relevant letters from consultants or anything like that” (CCGP1).