The context for quality
Increasing complexity and a system under strain
Our combined data sources, including our empirical data and grey literature (Table 1), revealed a vastly changed and changing context. In the past 5 years, the complexity of general practice — clinical, technical, and operational — has greatly increased, with ever more components, actors, health conditions, and wider needs, technologies, stressors, and demands.
Financial austerity and loss of resilience across the system
Our period of data collection occurred during an extended period of government policy oriented to reducing public spending. Qualitative data from participating practices, along with official reports,51–53 revealed numerous examples of under-resourced and withdrawn services, leading to bottlenecks, blocks, and tensions at the interfaces between services. Secondary care services had tightened referral criteria, and introduced required pre-referral work-ups and trials of therapy. Secondary care provided ‘advice and guidance’ instead of seeing a patient, or saw them once only and asked the GP to follow them up; patients who missed their secondary care appointments were discharged. This GP voices the frustration:
‘I know that the hospital isn’t twiddling their thumbs. I know they’re busy. I know they’ve never been busier. But we certainly have a frustration that work is dumped on us, but it’s secondary [care] guys’ responsibility.’
(GP, Westerly)
When GPs sought to push back against such ‘shifted’ work, practice staff had to contact secondary care and explain decisions to patients. Likewise, loss of resilience in the social care sector (reduced overall funding, reduced capacity, closure of some services, and longer waits for assessment) and within wider social networks (reduced informal community support) meant that practice teams had to spend time signposting to, form-filling for, or compensating for gaps. As well as being time consuming, this was challenging because of lack of knowledge, training, and shared records.
More complex patterns of illness and need
Staff in many participating practices reported that the population they served was getting older, sicker, and more socially isolated, as well as consulting more frequently. Our study was not designed to verify these impressions quantitatively.
A changing workforce and changing roles
Our interviews and ethnographic observations revealed reports of high staff turnover, perceived unmet training needs, and high levels of stress and sickness absence (covered in more detail in separate articles 33,38). Senior receptionists confided that newly recruited support staff, in particular, sometimes became disillusioned and left quickly as the full demands of the role became apparent. In some practices, there were unfilled posts for GPs (or insufficient funds to employ additional qualified GPs). Stress on partners in particular was sometimes exacerbated by a high burden of supervision of other multidisciplinary team (MDT) members.52,53 GPs in some practices had shifted to seeing only the most complex cases, eroding continuity and creating decision fatigue.53
Infrastructure under strain
While novel digital technologies were evident in some practices, digital infrastructure sometimes lacked the required functionality, speed, or capacity, resulting in systems that were, in the words of one staff member, ‘not up to the job’ (Practice Manager, Fernleigh). Limitations in how information was stored and (re) presented sometimes made it difficult for MDT members to follow the trajectory of a patient’s illness or understand others’ contributions.53 Expansion in numbers of staff had intensified pressure on physical space (some staff described taking their administrative work home because there was nowhere to work).
Telephone systems were frequently a source of frustration. Patients often spent a long time trying to get through; not all systems made it clear that the patient was working their way up a queue; and calls were sometimes dropped after a fixed period of waiting. Some practices had subcontracted telephone answering to call centres, leading to an impersonal experience for patients.
Purchasing decisions made at locality level sometimes left practices with no choice but to use technologies that were unfit for purpose, with standardised procurement processes and cost pressures resulting in a limited choice of technologies or key functionality being excluded from the specifications. Some practices procured their own technologies, but this could prove challenging in terms of the specialised knowledge and skills required.
Loss of in-person interactions among staff
It was evident from our ethnographic visits that digital technologies allowed both individual and collaborative work to happen without staff being physically co-located or working synchronously in time. Some staff welcomed improved options for interacting with other MDT members, especially those (such as pharmacists) who were not physically co-located. However, even when staff were working in the same building, emails and instant messages often replaced telephone calls or spontaneous in-person conversations (such as when people passed in the corridor or met in the staff room), reducing real-time interactions between staff and thus the opportunity to build strong relationships.
Delivering on quality (balancing benefits and trade-offs of hybrid services)
Our dataset included many positive comments from staff and patients about the increased convenience and flexibility of a hybrid service. This was evident in ethnography, interviews, and many of the patient reviews:
‘The option of telephone really cheesed a lot of people off [initially], “how can I be … treated over the phone?” and things like that. But I think gradually … patients have got used to that and a lot of patients are actually liking the convenience of not having to attend the GP and wait in the waiting room and can just have a quick call.’
(Receptionist, Westerly)
Some staff talked about improved data capture from patient self-monitoring:
‘So she actually put in the blood pressure readings [on FootFall] that she’s done at home because she had a big list of them. So it means she can go away now and do that herself, and that I’ll upload on to our system on to her records … And the good thing with the blood pressure is it actually gives us the average return at the bottom. … So that’s been good.’
(Healthcare assistant, River Rd)
However, there were also numerous, sometimes troubling, examples of quality in one domain achieved at the expense of compromised quality in (an)other domain(s). In the quote below, a GP talks about possible improvements to patient safety and variety of consultation modalities achieved at the expense of increased workload:
‘So electronic system[s], that’s the big key [thing] is patient safety, risk management … but not workload. It’s increased workload. And you know, it’s meeting other needs and it’s giving us other capacity, other tools to help consult the patients. So there are benefits, but not in terms of capacity. It’s reduced our capacity.’
(GP, River Rd)
Such trade-offs were evident across all the RBD2 practices, regardless of digital maturity, suggesting that they may be inherent. Below, we offer four illustrative examples of key quality trade-offs that were evident in our data.
Example 1: Care and compassion versus efficient transactions
Patients’ highest priorities involved the human and relational aspects of general practice including familiarity, warmth, empathy, compassion, and effective communication, as the following quote illustrates:
‘I’ve been with this practice for nearly 10 years, along with my kids. I am really happy with them. The doctors listen to me, share their expertise, make me feel involved, and give me choices about my care. The nurses are friendly and professional, and put us at ease, even with scary things like jabs for the kids. Reception are always very kind, explain things clearly, give useful options for appointment times, and don’t make me feel rushed.’
(Online patient review, Ogden East, February 2023)
Our ethnographic findings showed that both clinicians and support staff drew on their relational knowledge and human qualities to adapt care to patients’ needs. They often knew, for example, when a person lived alone or had a disabled relative. Some practices had a ‘usual doctor’ or ‘buddy group’ system to optimise relational continuity and this was highly valued by patients; others used a ‘most appropriate clinician’ arrangement, reducing relational continuity (as we have reported in more detail elsewhere35).
Despite the contextual pressures described in the previous section, staff in participating practices were strongly committed to delivering the human elements of quality (captured in the quality domains in Box 1 as patient-centredness, whole-patient care, equity, and relational continuity), and expressed frustration when they were unable to:
‘If somebody rings and they’re expecting [a baby] and they’re bleeding, we’re signposting them straight to the early pregnancy assessment unit but it’s pressure … and I just think that if somebody is expecting and they’re bleeding, I just think it’s nicer to come from the GP.’
(Receptionist, Fernleigh)
The receptionist in the above quote articulates a crucial tension — between what high-quality general practice would look like in an ideal system (the patient’s own, known GP would witness their suffering and guide them through the personal crisis of potential early pregnancy loss) and the transactionally framed compromises that are currently happening because of ‘pressure’. Note, however, that this was a staff member’s perspective; the patient may have been very happy to omit the GP step in this illness journey.
Example 2: Access and triage: matching supply with need and demand
All RBD2 practices had some kind of triage system designed to allocate patients to the most appropriate staff member and prioritise needy and urgent patients. These systems varied considerably, both in terms of technologies (including apps and web forms, usually with the facility for sending attachments, automated telephony systems, and receptionists asking questions of the patients, often from a flowchart or pro forma) and in terms of who did the triaging (receptionist, back-office support staff, pharmacist, nurse, allied health professional, GP trainee, or established GP), where it was done (designated triage room, variable office space, clinical room, home), how the patient was contacted (email, telephone call-back, text message), how other members of the MDT were informed (intranet messages, direct entry on to record, shared lists, paper artefacts such as sticky notes), and what rules and heuristics were used (for example, in one practice, a patient was allowed to book an in-person appointment only if the triage doctor allowed it).
Despite these differences, demand for general practice care often (although not always) exceeded supply. Practice staff made strong arguments that if demand went unchecked, systems would quickly become overwhelmed, leading to urgent problems and vulnerable patients being lost in the sheer volume of requests. In the quote below, a pharmacist reflects on how the introduction of electronic consultations led to a deluge of requests for minor (and usually self-limiting) problems:
‘I think to be honest with you, e-Consult has really left a scar on the practice [laughs] because at first it sounded great, you know, we had this system that was going to triage all these patients, and whatever, but I think the problem was it was just there … it was being abused … and perhaps, yes, we do need to educate patients to explain what the system is, how it should be used, you know.’
(Pharmacist, Caerleon)
Some support staff described being able, through triage, to offer on-the-day appointments to people judged as needing them, and some clinicians gave examples of time freed up to spend with complex patients. But digital triage systems also had significant unintended consequences across the quality domains (Box 3) — so much so that some practices had begun to discontinue or limit their use.
Loss of organisational efficiency
Skilled staff members were needed to do the triaging, reducing capacity elsewhere in the system. If unskilled staff were used, patients sometimes ended up seeing the wrong person Pre-assessment questions generated large amounts of information (much of it irrelevant) that staff did not have time to read Patients were not always available when called back. Sometimes practices cancelled patients after two failed call-backs, requiring re-booking Over-protocolised work processes led to double-handling (for example, if an annual review did not include reauthorisation of medication, the patient needed a second appointment) Triaging software and formalised flowcharts usually required patients to prioritise one problem, so a patient with multiple problems had to submit multiple requests and see multiple people. A GP could usually deal holistically (and more efficiently) with all the problems Some algorithms were risk-averse, misclassifying patients as emergencies when they knew they were not (possibly resulting in a flawed instruction to ‘dial 999’ or ‘go to A&E’), while others lacked granularity (sending patients to services such as optometry or pharmacy which were inappropriate to their needs) When triage was subcontracted to a call centre, lack of local knowledge (such as which nurse does the baby jabs) could lead to patients booked in to see the wrong clinician Patients sometimes spent a proportion of the consultation complaining about how hard it was to access services, and staff spent time explaining or apologising In practices where web forms were used as a common pathway for all patients, support staff time was spent converting telephone or walk-up requests into web format The rapid pace and high risks associated with triage increased staff stress, contributing to burnout, sickness absence, and staff turnover
Loss of accessibility
Many patients found the digital triage system time consuming, difficult, and stressful to navigate Some less digitally enabled patients, including those from vulnerable groups (such as those who are older, sicker, or who had more complex needs) experienced the digital front door as impassable (although some, including patients with hearing impairments, welcomed text-based interaction) Some practices did not proactively inform patients about new ways of accessing the practice (such as via an app) because they were already overwhelmed by existing demand
Loss of timeliness
Loss of patient-centredness
Assessment by structured symptom checklist and decision algorithm was impersonal and disheartening, as well as time consuming Call-backs occurred at inappropriate times (for example, when in a meeting, teaching a class, or driving a bus)
Inequities
Disadvantaged groups (see Box 4) had particular problems navigating the system ‘Gaming’ of web forms by patients with high health and digital literacy to achieve a desired outcome (for example, an in-person appointment with a doctor) meant there were even fewer slots available for less enabled patients
Threats to safety
Loss of continuity
|
Box 3. Unintended consequences of remote and digital triage
Structural and poverty-related factors
Geographical variation in broadband availability and strength (some remote areas had limited service) No (or entry-level) digital devices Low data package Device(s) shared among several family members No private space in the home No home Weak social networks (nobody to help or support)
Capability factors
Low health literacy, digital literacy, or system literacy Inability to read or understand text Cognitive conditions (such as learning difficulties, brain fog, memory loss) Neurodiversity Reflexivity and mastery (for example, the ability to assess a social situation and plan accordingly) Visual or hearing impairment Low fluency or confidence in the language used in the practice
|
Box 4. Challenges to equity in remote and digital general practice
Example 3: Long-term conditions: protocolised versus personalised care
Most RBD2 practices suspended routine LTC monitoring during the pandemic; many took some time (in one case, 3 years) to re-start. Practices varied in how they did LTC reviews; some had mostly reinstated in-person appointments. Others used structured web-based forms completed by the patient or a support staff member (with the patient either present in person or on the phone), automatically populating the Quality and Outcomes Framework template, which triggered payment. LTC reviews could thus be marked as completed without seeing or speaking to the patient.
Staff raised concerns that while web forms could capture data on patients who had previously defaulted from LTC reviews, patients did not always fill them out accurately. Patients sometimes failed to declare symptoms either because of low health literacy (for example, assuming cough and breathlessness to be normal in asthma) or not wanting to bother the practice. In some cases, staff doubted that patients were competent in using the devices they had been supplied with; some suspected that certain patients invented the readings.
Nurses expressed concern that heart failure reviews (mostly in older patients) conducted by telephone or web form were likely missing some cases of deterioration because a key component of assessment was a visual overview of the patient’s general condition (such as speed of walking, breathlessness at rest, or swollen ankles). The in-person assessment had also been an opportunity to reinforce patient education through conversations held in the context of an existing, ongoing therapeutic relationship. Nurses worried that not all patients would be able to learn what they needed to know from leaflets sent as email attachments or links to web resources.
Algorithmic, remote management of LTCs is built on the implicit assumption that all symptoms in a particular organ system are a result of the known LTC. Worsening respiratory symptoms in patients with chronic obstructive pulmonary disease, for example, could trigger a prescription for steroids and antibiotics on the assumption that this indicates an exacerbation, but the opportunity to examine a patient and detect an early lung cancer will be missed. Another concern was that some patients who would benefit from in-person assessment on clinical grounds might insist on a telephone appointment on convenience grounds.
Variability in quality of care for diabetes was a particular concern of practice staff and national stakeholders. This new, life-changing diagnosis was sometimes delivered by telephone, making it difficult to establish the therapeutic relationship on which ongoing care could be built. Most practices conducted interim reviews by telephone and some even did annual reviews this way, raising major concerns about omission of key examinations (such as of the feet or insulin injection sites). In one or two practices, annual reviews were now mostly completed by healthcare assistants and with most information supplied by the patient in advance; practice nurses spent much less time with patients so were less able to gain a holistic perspective and answer patients’ questions. Positive changes to diabetes care included the use of pharmacists to review complex drug regimens, group consultations which provided opportunities for peer support and vicarious learning, and (in some practices) a reallocation of clinician time to undertaking visits to housebound patients.
The Care Quality Commission (CQC) has raised concerns about clinical quality and equity in LTC management, commenting that insufficient support is provided, especially to patients from minority ethnic groups.51,52 Both CQC and patients expressed concerns about annual LTC reviews for people with autism and learning difficulties; problems included the review not happening at all, carer not invited or included, and patient not being offered an in-person appointment, thereby missing a visually obvious deterioration.
Example 4: Equity and inclusivity: mitigating the digital inverse care law
Like all UK public services, general practices are required to accommodate people with variable digital set-up and capability.55 Box 4 lists the challenges to equity that were evident in the RBD2 practices. Some of these (such as poor broadband and widespread poverty) were beyond practices’ direct control, while others (such as how well patients could read) were not always known.
Intermediation roles to help patients use technology — variously termed digital navigator, support worker, and patient liaison officer — were used increasingly by practices as our fieldwork progressed. These individuals were usually reception staff who had learned new digital navigation skills on the job,33 and were performing this new role in addition to their traditional one. While these individuals were often perceived as helpful, their dual role meant a high workload, high cognitive demands, and consequently increased stress. A single patient with complex needs could take a significant proportion of the time available (one interviewee described helping a patient navigate the complex online benefits system). In practices where most patients had markers of disadvantage, the need for digital navigator support significantly exceeded supply.
Theorisation
Socio-technical theorists remind us that technologies are not merely tools that we use for certain tasks.48,49,56 Rather, ‘work practices are conceptualized as networks of people, tools, organizational routines, documents and so forth’.56 Technologies, and their affordances (the things we can and cannot do with them), constitute our work and change who we are in the workplace. If LTC management occurs using symptom checklists and data fields populated remotely by the patient, for example, this aspect of care becomes impersonal and transactional, and the person checking those data becomes a data processor.
Several phenomena observed in this study — persistence of various remote and (especially) digitally mediated interactions, subcontracting the digital front door of the practice to a call centre, use of monitoring protocols which require the patient to submit data for subsequent (semi-automated) processing by support staff, and a tendency for staff to interact asynchronously — can be analysed through a sociological lens. These profoundly changed ways of working are examples of two wider phenomena originally described by Giddens back in the 1980s: distanciation,57 defined as ‘[t] he stretching of social systems across time–space’47 and disembedding,58 in which social activities occur increasingly at a distance, ‘removed from the immediacies of context’47 on the basis of technologically mediated and abstracted forms of information.
Distanciation and disembedding replace the warmth of a here-and-now, in-person interaction with words on a screen, tick-boxes, algorithmic pathways, and mounting task lists. Social interactions, instead of being contextually grounded and richly meaningful, become ‘emptied out’, dehumanised, and deprofessionalised.47 The example above of reception staff signposting a patient with early pregnancy bleed directly to a specialist diagnostic unit epitomises this shift. In this and similar examples, the very essence of a GP’s expertise — as a clinical generalist, an expert in the patient as a person, a professional witness to suffering, and someone who accompanies the patient on their illness journey — has disappeared from the aspects of care that are being acknowledged, measured, and reimbursed.
As complexity theory predicts, the multiple elements of quality and their various interdependencies generate emergent tensions and paradoxes that are not amenable to simple fixes but which require creative responses and ongoing adaptation.59
