A 'telephone first' approach to demand management in English general practice: a multimethod evaluation

Support offered by the companies

It is important to note that the two providers do not just provide a ‘telephone first’ appointment system but seek to work with practices to understand and manage demand and provide support in the implementation of and transition to their specific ‘telephone first’ approach. Doctor First has a three-stage approach for transitioning to a ‘telephone first’ approach over an 18-week period: (1) data gathering, to help practices to gain a clear understanding of activity, backlog, capacity and demand in order to identify how many clinical sessions are required per day to meet the demands of patients (takes 4–8 weeks); (2) an implementation phase, to reduce backlog of appointments and provide training to all staff to help with the smooth transition to Doctor First, inform patients of upcoming changes and clinical system configuration (takes 10 weeks); and (3) following ‘going live’, 4 weeks of patient feedback and monitoring to help gather views on the system and communicate with patients. 9

The other provider, GP Access, has a five-stage approach over 12 weeks; this is described as (1) coming to a consensus among partners and appointing a change leader (week 1), (2) preparing the whole staff team, patients and the system [includes an e-learning session for GPs on telephone consulting (weeks 1–3)], (3) support on the launch day (weeks 3–4), (4) access to rapid support by telephone following changeover (weeks 5–11) and (5) confirming decision to continue with approach 12 weeks after implementation. GP Access also offers practices a 3-month subscription to an analytical support service, which provides measurement, feedback and advice to address problems as they arise. 10

Reported benefits of the approach

Both providers report substantial benefits of the system to practices, in terms of both reduced stress for practice staff and cost savings as a result of time saved and more efficient use of time. The providers also report that practices that have switched to the new system have increased patient satisfaction. The nature of the claims made by Doctor First and GP Access on their websites are summarised in the following sections. As noted, there is no independent evaluation to date to support or refute these suggested benefits; our study seeks to address this gap in the evidence base.

Impact of patient and public involvement

The study team benefited from input from a number of patient and public involvement (PPI) members during the study; for example, the PPI members recommended the inclusion of out-of-pocket patient expenses in the economic analysis. We also had patient input on written documents, such as patient information leaflets and consent forms, ensuring that information disseminated to participants was suitable and understandable to a lay audience. Rather than repeating this information in each chapter, we have summarised the contribution of PPI in Chapter 11, Impact of patient and public involvement.

Analysis of administrative data from general practices (see Chapter 3)

Patient experience survey (see Chapter 4)

Patient experience: analysis of data from the national GP Patient Survey (see Chapter 6)

Secondary care utilisation: analysis of Hospital Episode Statistics data (see Chapter 7)

Economic analysis (see Chapter 8)

Interviews with patients and staff (see Chapters 9 and 10)

Analysis of disenrolment data (not conducted)

The original protocol included examining disenrollment data. However, because of the delays in obtaining HES data, we decided not to request additional data on disenrolment as it would have caused further delays in the delivery of the project.

Key studies and bodies of work

The majority of studies reviewed were based in the UK (53), and among these it is worth noting two dominant bodies of work, that is, papers relating to the same study or involving the same authors.

The first is the ESTEEM trial, to which nine papers relate. 6,7,21–27 The ESTEEM trial was a cluster RCT comparing nurse-led and GP-led telephone triage systems with usual care for patients requesting same-day consultations in general practice in England. 28 The core intervention in the ESTEEM trial consisted of a number of steps. Following an initial telephone conversation with a receptionist, patients requesting a same-day face-to-face appointment with a GP were called back by either a nurse or a GP. The phoning clinician discussed the patient’s condition and chose from a range of management options, such as dispensing self-care advice, booking a same-day or future face-to-face appointment or referral to other appropriate services. 6 In addition to presenting evidence around clinical effectiveness and cost-effectiveness of GP- and nurse-led triage, papers resulting from the trial also explore patient and staff experience, patient satisfaction and impact of telephone triage on GP workload.

The second dominant body of work is a series of five papers by McKinstry and colleagues. 29–33 These include qualitative research around GPs’, other staff members’ and patients’ views on the safety and appropriateness of telephone consultations compared with face-to-face consultations, as well as a two-site RCT investigating the use of doctor time and subsequent service use for telephone consultations compared with face-to-face consultations (as a way to manage requests for same-day appointments). 31 The RCT excluded urgent cases and patients asking to speak to the GP specifically for advice. In addition to the original journal articles, the RCT also accounts for seven letters to journal editors,34–39 written in response, and one editorial40 included in our review.

Terminology

The literature refers to both ‘telephone triage’ and ‘telephone consultation’, but it can be challenging to differentiate what is meant by each. In many cases, the terms appear to be used interchangeably or are ill-defined or poorly described. 41 The term ‘telephone triage’ is generally applied to an approach in which patients requesting a face-to-face appointment are asked to first speak to a doctor by telephone before a decision is made regarding whether or not the patient needs to attend and how quickly that should happen. ‘Telephone consultation’ can be understood as a more general term, which encapsulates consultations undertaken as part of triage or consultations that are undertaken by telephone for other purposes, such as scheduled check-ups or follow-up for the management of chronic conditions. In this chapter, we use the term ‘telephone consultation’ to encompass both triage and more generic consultation systems. It was not clear if any of the systems covered in the literature related directly to ‘telephone first’ approaches as presented in this study.

Prevalence of telephone consultations

There are few empirical studies that give an indication of the level of use of telephone consultations in general practice. A survey of general practices in Devon,42 published in 2001, showed that 19% of practices (n = 15) surveyed were offering telephone consultations for patients seeking same-day appointments that were not deemed urgent. A survey of practice managers in Wales,43 published in 2008, reported that 42% of those who responded (n = 167) had introduced GP telephone consultations more systematically since 2003, when a new General Medical Services contract was introduced. More recently, a survey of the use of different forms of alternatives to face-to-face appointments30 showed that two-thirds of practices (211/318, 66%) surveyed, across south-west England and Scotland, reported that they were using telephone consultations ‘frequently’, although this was not further defined.

Before the publication of a retrospective analysis of 100 million consultations in England (2007–14)44 in 2016, information on the proportion of consultations that take place by telephone within a general practice was mainly restricted to information provided by individual GPs in letters to journals. In these,34,37,45 GPs make the claim that telephone consultations form an important part of appointment provision, accounting in one case for 43% of all appointments. 45 The much larger retrospective analysis by Hobbs et al. 44 included analysis of GP and nurse consultations for registered patients at 398 English general practices from April 2007 to March 2014. The largest change observed in patterns of consultation over this period was a doubling of GP telephone consultation rates. This increase compared with a 5.2% increase in GP face-to-face consultations over the same period, although face-to-face consultations still accounted for 90% of all consultations despite the increase in use of telephone consultations.

Impact on access to primary care

One potential advantage of telephone consultations is that they improve access, either by negating the need to travel or by increasing convenience for patients; however, there is limited research addressing this directly. A qualitative study in Scotland33 reported that GP telephone consultations were generally seen to improve patient access. For patients who worked full time, had reduced mobility, valued not needing to travel to the surgery or lived in rural and remote areas, telephone consultations were seen as a helpful way to overcome physical distance and the need to travel. In urban practices, greater emphasis was placed on using telephone consultations for acute presentations as a way of managing demand.

The ESTEEM trial of telephone consultations for patients requesting same-day appointments23 reported that, although the introduction of the telephone consulting system was associated with improved access in terms of getting through on the telephone, there was no overall difference in ease of access to prompt care, comparing patients receiving the GP-led telephone system and those receiving usual care. The study also found that the introduction of telephone consultations did not seem to alleviate the challenges faced by working patients in accessing flexible and convenient care. 23 Patients unable to take time away from work or who could do so only with difficulty reported lower satisfaction than those who did not have these challenges, and this did not vary depending on whether patients were receiving GP-led telephone consultations or usual care. 27

Perceived appropriateness of telephone consultations

Patient and doctor perceptions of the appropriateness of using telephone consultations to deliver care in general practice appear to be given most focus among earlier published studies. Two studies46,47 found little agreement between doctors and patients in perceived appropriateness of GP telephone consultations for a given complaint. Stevenson et al. 46 found that, after face-to-face consultations, doctors and patients were in agreement about only 5.5% of 1067 face-to-face consultations in terms of whether or not the issue discussed could have been dealt with by telephone. Doctors showed a greater inclination towards telephone use than patients, with consultations considered in hindsight to be appropriate for management by telephone in 13.9% of cases for GPs, as opposed to 11.4% of cases for patients. 46 The single-practice study by Kernick et al. 47 also found little agreement between patients and doctors, who were separately interviewed before booked face-to-face consultations regarding whether they felt that the patient could have been appropriately managed by a telephone consultation with the GP or a consultation with a specially trained nurse. The study reported that both patients and doctors considered that only a small number of cases were suitable for management by telephone (5.8% and 5.1%, respectively). Among GPs interviewed by Hallam,48 there was broad agreement on the types of complaint that could be appropriately handled by telephone, including minor, self-limiting conditions and certain recurring conditions, whereas those thought to be inappropriate for telephone consultation included chest pain, abdominal pain, breathing difficulties, illness in a young child or new patient, non-traumatic bleeding and high fever.

More recently, the ESTEEM trial process evaluation6 suggests that patients are more open to telephone consultations and have greater confidence in the GP assessing their problem on the telephone. Patients appreciated the convenience offered by GP telephone consultations and felt that it made sense for the doctor, rather than the patient, to decide whether or not an appointment was needed and to weed out ‘time wasters’.

Experience of patients

Experience of staff

Education and training

The importance of education regarding the proper use of telephone consultations for clinicians and staff, as well as for patients, was highlighted by some authors, although it was not the main focus of any study reviewed.

Three papers55,59,60 from the grey literature highlighted concerns regarding a lack of specialised training for clinicians delivering telephone consultations, with GPs having to learn by trial and error. 55 The lack of training for receptionists and their key role in distinguishing urgent from non-urgent cases was also seen as a potential risk to patient safety. 60

Patient education and understanding of telephone consultation systems was also a factor explored in the literature, although to a small degree. A published survey of four practices in England52 linked significant differences in levels of patient awareness of telephone consultation systems between practices to the practices’ approaches to publicity. These included receptionists telling patients, use of leaflets or posters or reliance on word of mouth. 52 The study found that only half of the total number of patient respondents (n = 1025) knew that they could speak to a doctor by telephone. More recently, the ESTEEM trial process evaluation6 found that some patients were confused about how their new consultation system worked.

Impact on the nature of consultations

Two studies in the UK that compared the nature and content of patient–doctor interactions in telephone and face-to-face consultations61,62 found that telephone interactions were shorter and largely focused on single-issue or biomedical concerns. Telephone consultations emphasised biomedical information exchange over psychosocial or affective communication, and doctors used closed questions much more commonly than they used open ones;62 however, Hewitt et al. 61 point out that brief telephone consultations are appropriate when telephone consultations for new problems would lead to a face-to-face meeting. The ESTEEM trial25 also compared nurse and doctor telephone interactions with patients in a sample of video- and audio-recorded consultations. They reported that, although the length of calls was similar, nurses asked patients more questions [mean 14.72 questions, standard deviation (SD) 6.42 questions] than GPs (mean 5.51 questions, SD 4.66 questions), on average, whereas GPs asked more questions eliciting patient concerns or expectations and to obtain medical history than nurses (43% of GPs’ questions were of this nature, compared with only 11% of nurses’ questions). In interpreting these apparently large differences, it is important to note that the nurses, but not GPs, were using computer-aided software when they took a patient history.

Patient safety and health outcomes

A number of papers raise concerns about patient safety in relation to use of the telephone as an alternative to face-to-face contact with GPs, and indeed conclusions with regard to safety have been the subject of disagreement and debate among researchers;63 however, few studies have provided empirical evidence to support these concerns. 64

Studies that reported that telephone consultations appear safe did so only on the basis that patient–doctor communication can be adequate and patient recall of safety-netting instructions is improved. Most recently, although the authors could not rule out differences between groups for measures of safety, the ESTEEM trial found that GP- and nurse-led telephone consultations appeared safe, and were not associated with excess deaths, hospital admissions or attendance at emergency departments. 23

Impact on service utilisation and delivery

Impact on costs

Although significantly increased telephone costs45,49,58,68,71 and concerns around the initial financial outlay30 associated with the implementation of telephone consultation systems were raised in the literature, robust evidence on costs is sparse. An economic evaluation conducted as part of ESTEEM23 found that, although GP telephone consultations were associated with increased contacts, there was no significant difference in average costs of health care over 28 days from a same-day consultation request between patients who received GP telephone consultations and usual care, possibly attributable to the reduction in GP face-to-face contacts.

The impact of telephone consultation on prescribing behaviour was raised as another concern in relation to cost. Some authors argued that a tendency for telephone consultation to foster ‘automatic’ repeats of medication would result in overprescribing and higher costs for patients and the health system. 72 Yet, studies that investigated prescribing patterns31,48 found limited use in telephone consultations or no difference in GP antibiotic prescribing behaviour compared with face-to-face consultations.

The potential for telephone consultations to offer patient cost savings, especially on travel in remote rural areas, has also been noted,73 but remains unexplored in the literature.

Statistical analysis

Two types of analysis were carried out for each of the outcomes. The first was a before-and-after analysis, illustrated by the ‘superposed epoch graphs’ which (see Figures 2–4) in the chapter where the introduction of the system in each practice is set at time zero. A superposed epoch graph is a way of visually representing changes that take place over time in a number of practices when an intervention took place in all of the practices but at different times. The plots show trends over time relative to the time at which the intervention started, which is defined as time zero, enabling visual inspection of changes before and after the intervention. Second, an interrupted time-series regression analysis was performed for each outcome, looking (1) for within-practice step changes at the time when the intervention was introduced and (2) for a within-practice change in the preceding trend (e.g. slowing down of a previous increase). We also model heterogeneity in these changes to examine whether or not the intervention has a different effect in different practices. Full details of the statistical analysis are given in Appendix 2.

A large proportion (30%) of data on length of appointment were missing from the practice data provided by the commercial company, especially for telephone consultations (52% compared with 18% for face-to-face consultations). Ignoring appointments with missing durations would have led to a systematic underestimation of the total time spent consulting. To overcome this, appointment length was imputed for those appointments with missing length and then added to the observed lengths to obtain a better estimate of the total time spent consulting for each day in each practice. Because we were imputing individual consultation lengths rather than total time spent consulting per day in a practice, a single imputation was made using a linear regression model similar to that used in the analysis of individual consultation lengths, but with fixed effects for practice rather than random effects, and stratified by before and after the intervention launch. Single, rather than multiple, imputation was used because of the computational burden. The use of single imputation will not lead to a biasing of estimates but may lead to an underestimate of standard errors (SEs). To combat this, the imputed SEs were multiplied by the square root of the ratio of the proportion of cases requiring no imputation.

Our main analyses were all done on an intention-to-treat basis. This included all practices identified by the commercial company as having used the ‘telephone first’ approach, even when the practices were using a hybrid form of the approach (e.g. by allowing some prebooked appointments) or had since ceased using it altogether. In this and other analyses, a sensitivity analysis (see Appendix 3) was performed, restricting the analysis to practices in which we believed, on the basis of information provided by the commercial company, that the system was being run consistently with the company’s protocols. The companies were asked to classify all practices that had used their ‘telephone first’ approach as ‘running’, ‘hybrid’ (i.e. allowing some additional degree of advance booking of appointments) or ‘reverted’ (i.e. had stopped using the ‘telephone first’ approach). In the per-protocol sensitivity analyses, we included only practices classified by the companies as ‘running’.

Number of appointments

Superposed epoch analysis

The mean number of appointments per 1000 patients per day was 16.5 (SD 6.3) before the intervention started; this increased to 21.8 appointments (SD 8.1 appointments) post intervention (Figure 2). This increase differed by appointment type, with decreases seen in the number of face-to-face appointments, from a mean of 13.0 appointments (SD 4.5 appointments) to a mean of 9.3 appointments (SD 5.5 appointments) (Figure 3), and increases seen in the number of telephone appointments, from a mean of 3.0 appointments (SD 4.0 appointments) to a mean of 12.2 appointments (SD 7.5 appointments) (Figure 4).

FIGURE 2.

Superposed epoch analysis showing the change in the total number of appointments per 1000 patients per day relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 3.

Superposed epoch analysis showing the change in the number of face-to-face appointments per 1000 patients per day relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means. Reproduced from Newbould et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced from Newbould et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

FIGURE 4.

Superposed epoch analysis showing the change in the number of telephone appointments per 1000 patients per day relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means. Reproduced from Newbould et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced from Newbould et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Time waited for an appointment

Superposed epoch analysis

The mean number of days between booking an appointment and having an appointment across all intervention practices was 4.0 (SD 7.0) prior to the intervention and 0.9 (SD 3.9) after the intervention (Figure 5). Similar drops in time between booking and having an appointment are seen when restricting to face-to-face appointments [from a mean of 4.5 days (SD 7.4 days) to a mean of 1.8 days (SD 5.6 days)] and telephone appointments [from a mean of 1.8 days (SD 4.2 days) to a mean of 0.3 days (SD 1.7 days)].

FIGURE 5.

Superposed epoch analysis showing the change in the mean time between booking and having an appointment of any type relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Length of appointment

Superposed epoch analysis

On average, the duration of appointments decreased from 10.5 minutes (SD 6.0 minutes) pre intervention to 8.5 minutes (SD 6.2 minutes) post intervention (Figure 6). The change in duration was less when restricting to face-to-face appointments [from a mean of 10.9 minutes (SD 5.9 minutes) to a mean of 10.2 minutes (SD 6.4 minutes)] or telephone appointments [from a mean of 7.7 minutes (SD 6.0 minutes) to a mean of 6.2 minutes (SD 5.1 minutes)], suggesting that much of the overall reduction in average appointment duration is due to a change in the proportion of appointments that are telephone appointments. It is worth bearing in mind that patients who had a face-to-face appointment may also have had a telephone appointment. For this reason, it is likely that the total consultation duration for each patient increases for patients who had both telephone and face-to-face appointments; however, as we cannot link telephone and face-to-face appointments, we cannot demonstrate that this is actually the case.

FIGURE 6.

Superposed epoch analysis showing the change in appointment duration relative to the intervention launch: face-to-face and telephone. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Total time spent consulting per day by general practitioners

Superposed epoch analysis

The mean time spent consulting per 1000 patients increased slightly from 7.8 hours (SD 2.5 hours) per day pre intervention to 8.5 hours (SD 4.0 hours) post intervention (Figure 7). It should be noted that the number of practices providing data decreased rapidly in the months after the launch. The reason for this is that, although we had a minimum cut-off point for the ‘after’ period of data collection, some practices had longer periods of data available for the period after the change. For this reason, some medium- or long-term effects might not be evident.

FIGURE 7.

Superposed epoch analysis showing the change in the total time spent consulting relative to the intervention launch: hours per 1000 patients per day. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Continuity of care

Superposed epoch analysis

Continuity of care data were available for 50 practices (100,367 data points pre launch and 27,847 data points post launch). The mean continuity index (which takes possible values between 0 and 1) in practices was very similar before and after the launch of the intervention: 0.728 (SD 0.08) and 0.731 (SD 0.79), respectively. This can be seen in the superposed epoch analysis (Figure 8), in which there is some fluctuation over time but values are roughly equal before and after launch. It is also worth noting that the data become sparse at > 5 months post launch.

FIGURE 8.

Superposed epoch analysis showing the continuity index relative to the months to launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Survey development

The survey was designed by the evaluation team, informed by a review of the literature and with reference to a number of existing surveys (including the GP Patient Survey, the General Practice Assessment Questionnaire, the out-of-hours patient questionnaire and the survey used in the ESTEEM trial) from which questions were sourced or adapted. Survey questions were first tested in cognitive interviews with 16 patients in the waiting room of a single practice to determine whether or not they were being interpreted as intended. A revised survey and covering letter were reviewed by PPI representatives on the study advisory committee before full-scale piloting of the survey by post was undertaken (according to the dissemination method outlined in the following sections). Only minor revisions were made following piloting; therefore, pilot data were included in the overall analysis.

Practice selection, sampling and recruitment

The two commercial companies provided a list of practices known to be running the ‘telephone first’ approach. The sampling frame was restricted to practices that had been operating the ‘telephone first’ approach since 24 June 2015 (n = 101) (i.e. those that had been operating the approach for ≥ 6 months at the time of recruitment) in order to ensure that the new appointment system had been given enough time to ‘bed in’. Initially, a stratified random sampling method was used to recruit practices; however, the low participation rate meant that, ultimately, all practices were approached in order to reach our recruitment target. Practices were approached in batches (initially of 10 surgeries and then of five surgeries) by letter; a researcher followed up with a telephone call until either a response was received or researchers had been chasing for 3 months (when attempts were abandoned). This was followed by a mail-out to all of the remaining 61 practices in the final batch. With the final batch for chasing, we randomised the order of the list and worked our way through it, chasing by telephone until we had a sufficient number of practices. The response rate from practices was 20%.

The characteristics of practices that participated in the survey are shown in Table 7. Although there is variation across most of these characteristics, it should be noted that only one rural practice was recruited.

Because of the likelihood that practices that agreed to take part in the survey were more successful in operating the ‘telephone first’ approach than those that declined, we compared the two sets of practices using nationally reported GP Patient Survey data from 2015/16 (weighted for non-response and design). The mean percentage of patients reporting being seen or spoken to on the same or next day was 81.5% (SD 4.2%) for participating practices and 68.6% (SD 14.6%) for those practices that were not recruited. For ‘good’ or ‘very good’ ratings of overall experience, the mean was 86.4% (SD 7.7%) in participating practices and 79.9% (SD 11.4%) in those practices that were not recruited. This confirmed our concern that responses from practices that agreed to take part in this survey were likely to be more positive than other practices using the ‘telephone first’ approach.

Survey distribution

For each of the 20 practices, the survey was distributed by post to a random sample of between 80 and 100 patients who had received a telephone consultation with a GP during the day preceding the survey mail-out. The mail-out day was selected by the practices. Surveys were distributed between 23 November 2015 and 27 October 2016.

Details of patients who had received a telephone consultation in the day preceding the survey mail-out (or the previous Friday if the mail-out was on a Monday) were extracted into a Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) spreadsheet. These details included:

• patient name

• NHS number

• sex

• age

• ethnicity (if available)

• GP seen

• date and time (a.m. or p.m.) of appointment (if available)

• home address

• usual GP (if available).

Duplicates, based on NHS number, were removed. Children aged between 13 and 17 years were excluded in order to avoid compromising confidentiality when approaching patients in this age group. If the number of patients remaining was < 80, data were extracted on all patients who had spoken to a GP in the 2 days prior to the survey mail-out. If there were > 100 patients (either in a single day or across multiple days needed to get to a sample size of ≥ 80), a random sample of 100 patients was taken. The final sample was reviewed by GPs, who removed any patients for whom they had concerns related to vulnerability.

The survey was posted to patients (or the parents of patients aged < 13 years) with a letter from the practice inviting them to respond. Patients were encouraged to ask for assistance to complete the survey if required. A reminder was posted to non-responders 2 weeks after the original mail-out.

Analysis

Survey responses have been summarised using descriptive statistics. Some patients did not answer all of the survey questions; consequently, the total number of responses presented varies for each question as we present the results based on all those who answered it. In order to account for non-response bias, the results have been weighted based on the age and sex of the patients who were sent a survey. There were insufficient data on ethnicity to include in the weighting. The weighted results represent the views of the patients consulting, which might not be representative of all patients registered at the practice; however, this difference also reflects the fact that some groups of patients are more likely to contact the practice than others.

In order to investigate whether or not certain patient groups had different experiences of the ‘telephone first’ approach, we used logistic regression models. We created six dichotomous outcomes:

1. Was the patient expecting a callback? (Yes compared with no/unsure.)

2. How did the patient rate the length of time to be called back? (Good/excellent compared with acceptable/poor/very poor.)

3. How convenient was it speaking to a GP on the telephone? (More convenient compared with less convenient/no difference.)

4. How difficult was it communicating with a GP on the telephone? (More difficult compared with less difficult/no difference.)

5. Did the patient receive a face-to-face appointment with someone? (Yes compared with no.)

6. Would the patient prefer to return to the old appointment system? (Yes/do not mind compared with no.)

Each outcome was modelled in a separate logistic regression, which included age (categorised into nine age bands), sex (male/female), health status (five categories from poor to excellent), ethnicity (white/non-white), primary language (English/other), ability to take time off work (four categories: not relevant, no, yes with difficulty and yes easily) and a random effect for practice.

Descriptive analysis

Patient experience of using the ‘telephone first’ approach

Of the total sample of respondents, 85.0% (707/831, 82.0% weighted) were calling to discuss their own health (i.e. they were the patient), 9.0% (75/831, 13.0% weighted) were calling on behalf of their child, 4.2% (35/831, 3.5% weighted) were calling on behalf of an adult for whom they cared and 1.7% (14/831, 1.2% weighted) were calling for another reason.

The vast majority of respondents, (91.0%) initially contacted the practice by telephone (753/825, 92.0% weighted), 1.7% initially contacted the practice online (14/825,1.4% weighted) and 5.7% initially contacted the practice in person (47/825, 5.4% weighted). Only around two-thirds (69.0%; 569/825, 67.0% weighted) of respondents expected to receive a callback from the GP.

The length of time that patients waited for a callback from the GP varied considerably: 23% (189/814, 21% weighted) waited for < 20 minutes, 44% (360/814, 44% weighted) waited for 20–60 minutes and 32% (256/814, 34% weighted) waited > 1 hour. In addition, 1.1% of respondents (9/814, 1.1% weighted) reported that they did not receive a callback. The majority of patients who were called back within 20 minutes of the initial contact rated the waiting time as excellent (71%, 75% weighted). Patients called back within 20–60 minutes most commonly rated the callback waiting time as excellent (33%, 38% weighted) or good (36%, 34% weighted). The majority of those who waited for > 1 hour considered this acceptable (43%, 41% weighted) or good (24%, 25% weighted). The waiting times before receiving a callback and how the respondents rated these are presented in Table 8.

TABLE 8 - Rating of callback waiting time

Note

Bold text indicates the category in which the median value falls for each row.

Waiting time for callback	Rating of waiting time, % of patients (n/N)										Total, N (%)
	Very poor		Poor		Acceptable		Good		Excellent
	Absolute	Weighted	Absolute	Weighted	Absolute	Weighted	Absolute	Weighted	Absolute	Weighted
< 20 minutes	2.1 (4/189)	2.1	0.5 (1/189)	0.4	2.7 (5/189)	2.2	24 (45/189)	20	71 (134/189)	75	189 (100)
20–60 minutes	0.3 (1/360)	0.3	1.9 (7/360)	1.4	29 (104/360)	26	36 (129/360)	34	33 (119/360)	38	360 (100)
> 1 hour	6.3 (16/256)	6.5	16 (41/256)	16	43 (110/256)	41	24 (61/256)	25	11 (28/256)	11	256 (100)
No callback	33 (3/9)	31	33 (3/9)	23	22 (2/9)	39	11 (1/9)	6.9	0 (0/9)	0	9 (100)
Total	3.0 (24/814)	3.1	6.4 (52/814)	6.4	27 (221/814)	27	29 (236/814)	28	35 (281/814)	36	814

Overall, 50.0% of calls (421/837, 50.0% weighted) resulted in the patient being asked to come to the practice for a face-to-face consultation: 44.0% (367/837, 44.0% weighted) for a GP appointment and 7.3% (61/837, 6.6% weighted) for a nurse appointment. The outcomes of calls are listed in Table 9 (88 respondents listed more than one outcome). Overall, 89.0% of respondents (695/784, 89.0% weighted) were satisfied with the outcome of their call, whereas 7.9% (62/784, 7.9% weighted) thought that they should have received a face-to-face appointment, 1.2% (9/784, 0.9% weighted) thought that a follow-up call should have been arranged and 2.3% (18/784, 2.2% weighted) chose the ‘other’ option.

TABLE 9 - Outcome of call

Note

Respondents could select multiple outcomes of a call.

Outcome of call	% of participants	n/N, % weighted
Advice only	17	145/837, 17
Prescription	22	183/837, 22
GP face-to-face appointment	44	367/837, 44
Nurse face-to-face appointment	7	61/837, 7
Follow-up call with GP	4	30/837, 4
Follow-up call with nurse	2	17/837, 2
Other	8	67/837, 8

The majority of participants found speaking to a GP by telephone before having a face-to-face appointment more convenient than just attending a face-to-face appointment without being able to talk to the doctor on the telephone first (55%, 56% weighted); 22% of participants (22% weighted) reported that it was less convenient and 23% (22% weighted) reported that there was no difference. Respondents’ weighted ratings of the convenience of talking to a doctor on the telephone before making an appointment are illustrated in Figure 9.

FIGURE 9.

Weighted rating of the convenience of being table to talk to a GP on the telephone first compared with attending a face-to-face appointment without being able to speak to the doctor on the telephone first.

The majority of patients reported finding no difference between their ability to communicate with the GP over the telephone and their ability to communicate with the GP in person (65%, 64% weighted); 23% (24% weighted) reported that it was more difficult and 12% (12% weighted) reported that it was less difficult. Respondents’ perceptions of their ability to communicate with the GP by telephone is illustrated in Figure 10. For those who reported that it was more difficult to communicate with the GP over the telephone than face to face, the reasons for their answers are provided in Table 10.

FIGURE 10.

Weighted rating of the difficulty of communicating with the GP by telephone compared with in person.

TABLE 10 - Reasons why respondents found it more difficult to communicate over the telephone

Note

Respondents could select multiple reasons.

Reason	% of participants	n/N, % weighted
English is not my first language	0	0/0, 0
Doctor could not see my health problem	29	52/182, 28
Telephone line was not clear	8	15/182, 8
Patient had impaired hearing	3	5/182, 2
Difficult to explain the problem	47	85/182, 49
Other	2	4/182, 2

Overall, 31% of respondents (250/816, 32% weighted) reported that they preferred the ‘telephone first’ approach to the previous appointment system used by the practice, 33% (267/816, 30% weighted) would prefer to return to the old system and 37% (299/816, 38% weighted) reported that they did not have a preference.

Regression analysis

The results of the regression analysis exploring whether or not patient demographic and social factors are associated with their reported experience of the ‘telephone first’ approach are presented in Table 11. The survey was not designed to be sufficiently large to reliably look at the experiences of individual patient subgroups, and in most cases the differences we found in this post-hoc analysis were not statistically significant. Given this lack of power, we note that there is a lack of statistical significance even when large coefficients are seen [e.g. odds ratios (ORs) as large as 2]. One statistically significant effect is that women are less likely to report it being more convenient to speak to a GP on the telephone (OR 0.57, 95% CI 0.41 to 0.82; p = 0.002).

There was weak evidence that working status was associated with convenience of speaking to a GP on the telephone (p = 0.07) and ratings of waiting times (p = 0.06). In particular, those who could not take time away from work to receive a callback were less likely to report it to be more convenient to speak to a GP on the ‘telephone first’ (OR 0.46, 95% CI 0.23 to 0.92) and less likely to rate the waiting time as being satisfactory than those who were more easily able to take time away from work (OR 0.45, 95% CI 0.22 to 0.91). There was also weak evidence (p = 0.06) that respondents with less than excellent self-reported health were less likely to report that it was more convenient to speak to a GP by telephone than those who reported being in excellent health (e.g. poor vs. excellent OR 0.34, 95% CI 0.14 to 0.83). There was also weak evidence (p = 0.08) that self-reported health status was associated with the reported ease of communicating with a GP by telephone. Respondents with very good/good/fair self-reported health status were less likely to report that it was easier to communicate by telephone and those in poor health were more likely to report that it was easier to communicate by telephone than those in excellent health (e.g. poor vs. excellent OR 2.05, 95% CI 0.63 to 6.61). Given that there was no correction for multiple testing, these results should be treated with caution.

Space was available at the end of the questionnaire to add free-text comments; these comments showed a wide range of views, which, consistent with the patient interviews (see Chapter 9), included both very positive and very negative views. An illustrative selection of these comments is presented in Box 2.

Survey development

The development of the survey was informed by themes emerging from the qualitative element of the study (based on interviews with GPs and practice staff, and with patients). Survey questions were first tested in a cognitive interview with a practice manager to determine whether or not they were being interpreted as intended, before piloting the revised survey with a small sample of practice managers (n = 3).

Practice selection and sampling

The two commercial companies provided a list of practices known to be running the ‘telephone first’ approach. The sample frame was restricted to practices that had been operating the ‘telephone first’ approach since 24 June 2015 (n = 101 practices).

Survey distribution

A paper-based survey was sent by post and addressed to the practice manager in September 2016 alongside a letter from the study’s principal investigator. When relevant, the letter was tailored to account for whether the practice had already participated or declined to participate in other elements of the study. A reminder survey was sent 4 weeks later, followed by an opportunity to fill in the survey online 6 weeks after the initial mail-out.

Analysis

Survey responses have been summarised using descriptive statistics. Some practice managers did not answer all of the survey questions; consequently, the total number of responses presented varies for each question as we present the results based on all those who answered it. Free-text responses were read and grouped into overarching themes.

Data source and sampling

The GP Patient Survey is a national postal survey of patients’ experiences of primary care in England. The survey is sent twice a year (July to September and January to March) to ≈2.8 million adult patients per year. A stratified sample of patients aged ≥ 18 years who have been continuously registered with a general practice for ≥ 6 months is drawn from the practice list of each general practice in England, with patients from practices that are known from prior surveys to have low response rates being oversampled. Full details of the sampling strategy have been published. 75

We used data from July 2011 to April 2016. The contents of the survey remained largely consistent over this time period. Response rates have fluctuated over the period: falling from 38% in 2011/12 (1,037,946 responses) to 33% in 2014/15 (858,381 responses) and recovering to 39% in 2015/16 (836,312 responses) following the introduction of a reminder postcard in addition to the two reminder surveys already used. The data period included data from a total of 8323 practices, but not all practices had data for each wave of the survey; the number of practices contributing data varied between 8243 and 7687 in any one wave. This variation is primarily due to practice closures and openings.

Practices using a ‘telephone first’ approach between 2011 and 2016 were identified based on information provided by the two commercial companies providing support for implementing this service. The companies provided details on the practices that were using their system, the date when the ‘telephone first’ approach was introduced and the current status of the system (i.e. whether still running a full ‘telephone first’ system, running a hybrid system or if the practices had abandoned the system). Only practices that launched the ‘telephone first’ approach before 31 December 2015 were included in the analysis to allow sufficient time for the system to have bedded in; those with later launch dates were classified as non-intervention practices in this analysis. In total, 146 intervention practices were identified in the GP Patient Survey data set. This was considerably larger than the expected number of 90 practices used in our original sample size calculations. The number of practices receiving the intervention varied over time (Table 14), and only one practice was using the ‘telephone first’ approach throughout the entire data period. For the intervention practices, we had data from between 0 and 4.5 years pre intervention and 0.5 and 5 years post intervention (nine GP Patient Survey waves in each case). Analysis was restricted to survey respondents who had indicated that they had attempted to see or speak to a GP in the previous 6 months. This restriction allowed us to attribute reported experiences to the time period of the survey. A total of 29,472 surveys were available from intervention practices post launching the ‘telephone first’ approach. To inform selection of control practices, we undertook a survey of 500 practices to establish how widespread approaches similar to the ‘telephone first’ approach were (see Appendix 5). Based on our findings that such approaches were rare, controls were selected at random from all practices classified as non-intervention practices (i.e. any practice in England not on the list provided by the commercial companies or practices that launched the ‘telephone first’ approach after 31 December 2015). For computational reasons, our analysis was restricted to data from all 146 intervention practices and a random 10% sample of control practices, with between 778 and 976 control practices providing data at any one time.

Analysis

We used seven measures of patient experience from the GP Patient Survey in this analysis:

1. Last time you saw or spoke to a GP from your GP surgery, how good was that GP at each of the following?

   1. giving you enough time

   2. listening to you

   3. explaining tests and treatments

   4. involving you in decisions about your care

   5. treating you with care and concern.

2. Generally, how easy is it to get through to someone at your GP surgery on the telephone?

3. Would you recommend your GP surgery to someone who has just moved to your local area?

4. How often do you see or speak to the GP you prefer?

5. How long after initially contacting the surgery did you actually see or speak to them?

6. How convenient was the appointment you were able to get?

7. Overall, how would you describe your experience of making an appointment?

In each case, responses were rescaled between 0 (poor experience) and 100 (good experience). In the case of the first question above on GP communication, a composite variable was created to take the mean of all informative responses as long as three or more informative responses were given. For the fifth question, a binary outcome was created of whether or not the patient was ‘seen or spoken to on the same day’ as the initial contact. The reason for also looking at ‘time until seen or spoken to’ as a binary variable is that the protocol recommended by the commercial companies is that all requests for an appointment with a GP should be dealt with on the same day.

Two types of analysis were carried out for each of the outcomes. The first was a before-and-after analysis, illustrated by the superposed epoch graphs (see Figures 11–18) in which the introduction of the system in each practice is set at time zero. Second, a regression analysis was performed for each outcome, looking (1) for step changes at the time when the intervention was introduced and (2) for a change in the preceding trend (e.g. slowing down of a previous increase). We also model heterogeneity in these changes to examine whether or not the intervention has a different effect in different practices. Adjustment was also made for season and national trends along with adjustment for person-level factors (age, sex, ethnicity and deprivation level). Full details of the statistical analysis are given in Appendix 6.

Our main analysis was conducted on an intention-to-treat basis. It includes all practices identified by the commercial companies even when the companies informed us that the practices were running a hybrid form of the system or were no longer running the system. This was done in order to avoid selection bias, whereby only the successful practices continue with the system in the recommended form. A sensitivity analysis (see Appendix 6) was also performed, restricting the analysis to the 89 practices in which we believed, on the basis of information provided by the commercial companies, that the system was being run consistent with the companies’ protocols.

Before-and-after analysis of GP Patient Survey score (intervention practices only)

Figures 11–18 show the changes in score relative to intervention launch (the superposed epoch analysis) for the seven measures of patient experience. For each graph, the black lines represent the mean score of patients within a single intervention practice relative to the launch time. The green dots represent the mean score of patients in all intervention practices combined relative to the launch time. In all cases, large variation in scores between practices can be seen by the spread in the black lines.

FIGURE 11.

Superposed epoch analysis of GP communication composite mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 12.

Superposed epoch analysis of ease of getting through on the telephone mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 13.

Superposed epoch analysis of mean score on ‘recommending GP surgery to someone who has just moved into the area’ relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 14.

Superposed epoch analysis of ability to see preferred GP mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 15.

Superposed epoch analysis of time until seen or spoken to mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 16.

Superposed epoch analysis of percentage of people seen or spoken to on the same day relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 17.

Superposed epoch analysis of convenience of appointment mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 18.

Superposed epoch analysis of overall experience of making an appointment mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

The most obvious change in score was seen for the patients’ rating of time until seen or spoken to (see Figure 15). The mean score increased from 58.6 points (SD 11.7 points) (on a scale from 0 to 100) before the launch to 74.5 points (SD 10.0 points) immediately after launch (year 0.5 in Figure 15), and increasing to 79.0 points (SD 10.0 points) in the next wave (year 1 in Figure 15). The proportion of patients being seen or spoken to on the same day as making initial contact with the surgery increased from 40.9% (SD 14.1%) before the launch to 67.2% (SD 14.9%) in the year following launch. This change is substantial compared with the background variation between practices.

For other outcomes, changes were far more modest and often not easily discernible from background trends. Scores increased for one of the six other outcomes considered: convenience of appointment [from a mean of 78.3 points (SD 5.7 points) at the time of the launch to a mean of 79.3 points (SD 5.7 points) 1 year post launch; see Figure 17]. Scores decreased for the other five measures considered: (1) GP communication [from a mean of 83.3 points (SD 4.9 points) at the time of the launch to a mean of 82.7 points (SD 5.4 points) 1 year post launch; see Figure 11], (2) ease of getting through on the telephone [from a mean of 62.2 points (SD 12.4 points) at the time of the launch to a mean of 61.5 points (SD 12.9 points) 1 year post launch; see Figure 12], (3) recommend GP surgery [from a mean of 79.4 points (SD 8.9 points) at the time of the launch to a mean of 77.4 points (SD 9.9 points) 1 year post launch; see Figure 13], (4) ability to see preferred doctor [from a mean of 60.1 points (SD 11.5 points) at the time of the launch to a mean of 58.5 points (SD 12.1 points) 1 year post launch; see Figure 14] and (5) rating of overall experience of making an appointment [from a mean of 72.2 points (SD 8.7 points) at the time of the launch to a mean of 72.0 points (SD 9.4 points) 1 year post launch; see Figure 18]. These differences are all very small compared with the mean reduction in time until being seen or spoken to (20.4 points on a 0–100 scale; see Figure 15), the largest being the 1.0-point increase in convenience of appointment time and the 1.6-point reduction in patients who would recommend the practice to someone who has just moved into the area.

Comparison with other practices in England (controlled regression analysis)

The results of the mixed-effects regression analyses for the seven patient experience measures are presented in Table 15. The coefficient for the step change following the intervention can be interpreted as the effect on the intervention immediately following the launch being the difference between the observed scores and that expected had the intervention not happened. Consistent with the graphical analysis, there is a very strong effect seen in the rating of time until being seen or spoken to, which increases by 20.0 points (95% CI 18.2 to 21.9 points; p < 0.001) on the 0–100 scale, and when considered as a binary outcome, we see an increased percentage of patients being seen or spoken to on the same day as making contact, with an OR of 2.70 (95% CI 2.41 to 3.01; p < 0.001). There is also evidence of decreases in the GP communication score (–0.89 mean points, 95% CI –1.40 to –0.38 points; p < 0.001), recommending GP surgery (–2.37 mean points, 95% CI –3.22 to –1.52 points; p < 0.001) and ability to see the preferred GP (–1.25 mean points, 95% CI –2.41 to –0.08 points; p = 0.035). There is no evidence of a change in the other three outcomes considered (p > 0.3).

For five of the seven continuous outcomes and the binary outcome of being seen or spoken to on the same day, there is no evidence of a differential trend between the intervention practices and that which would have been expected (additional yearly change following intervention in Table 15, p > 0.3). For the convenience of making an appointment and the overall experience of making an appointment, there is a small additional positive trend (p = 0.016 and p = 0.002, respectively).

In addition to the average differences, Table 15 also shows that there was evidence of strong heterogeneity in the effect of the intervention, demonstrated by the random effects. For all but one of the outcomes considered, the SD of the random slope for both the step change following intervention and the differential slope following intervention was larger than the main effect, meaning that substantial numbers of practices saw an improvement, whereas others saw declines in scores (see the 95% reference ranges in the step change column in Table 15). For the rating of time until being seen or spoken to, improvement was seen across the board, but with large variations in magnitude: the middle 95% of practices showed increases of between 1.4 points and 38.6 points for the continuous outcome (ORs between 1.15 and 6.31).

Finally, the supplementary analysis found no evidence that the effect of the intervention was different between those in work or not in work (p > 0.1 for all; results not shown). The sensitivity per-protocol analysis (see Appendix 7) produced broadly consistent findings, but with some small differences: there was evidence of an increase in the step change for ease of getting through on the telephone (mean 1.79 points, 95% CI 0.44 to 3.13 points; p = 0.009) and the differential trend in time until being seen or spoken to (mean 1.91 points, 95% CI 0.90 to 2.93 points; p < 0.001 as a continuous outcome and OR of 0.24, 95% CI 0.62 to 1.60; p < 0.001 as a binary outcome).

Before-and-after analysis (intervention practices only)

The superposed epoch analysis of A&E attendances for intervention practices is shown in Figure 19. There is no clear change in the rate of A&E attendances at the time of intervention launch. The mean rate of A&E attendances is 35.9 (SD 10.8) per 1000 patients per month in the year prior to the launch of the ‘telephone first’ approach and 36.7 attendances (SD 11.3) per 1000 patients per month in the year after the launch.

FIGURE 19.

Superposed epoch analysis of A&E attendances relative to the intervention launch. The black lines represent the attendances within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

The superposed epoch analysis of outpatient attendances for intervention practices is shown in Figure 20. As with A&E attendances, there is no clear change in the rate of outpatient attendances at the time of intervention launch. The mean rate of A&E outpatient attendances is 196.1 (SD 49.9) per 1000 patients per month in the year prior to the launch of the ‘telephone first’ approach and 207.8 attendances (SD 55.9) per 1000 patients per month in the year after the launch.

FIGURE 20.

Superposed epoch analysis of outpatient attendances relative to the intervention launch. The black lines represent the attendances within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

The superposed epoch analysis of inpatient admissions for intervention practices is shown in Figure 21. Again, there is no clear change in the rate of admissions around the time of intervention launch. The mean admission rate is 31.3 attendances (SD 6.53) per 1000 patients per month in the year prior to the launch of the ‘telephone first’ approach and 32.2 attendances (SD 7.0) per 1000 patients per month in the year after the launch.

FIGURE 21.

Superposed epoch analysis of all inpatient admissions relative to the intervention launch. The black lines represent the admissions within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

The superposed epoch analyses of inpatient admissions for ACS conditions, elective admissions and emergency admissions are shown in Figures 22–24. As with the other admissions, no clear change is seen that corresponds with the start of the ‘telephone first’ approach. The mean rates of admissions for ACS conditions, elective admissions and emergency admissions are 3.1 (SD 1.2), 17.8 (SD 5.0) and 10.5 (SD 2.7) admissions per 1000 patients per month, respectively, in the year prior to the launch of the ‘telephone first’ approach and 3.2 (SD 1.3), 18.0 (SD 5.1) and 10.9 (SD 2.9) admissions per 1000 patients per month, respectively, in the year after the launch.

FIGURE 22.

Superposed epoch analysis of all inpatient admissions for ACS conditions relative to the intervention launch. The black lines represent the admissions within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

FIGURE 23.

Superposed epoch analysis of elective inpatient admissions relative to the intervention launch. The black lines represent the admissions within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

FIGURE 24.

Superposed epoch analysis of emergency inpatient admissions relative to the intervention launch. The black lines represent the admissions within a single practice relative to the launch time, with each black line representing a different intervention practice. The green dots represent the mean of individual practice means.

Regression analysis (using randomly sampled practices in England as controls)

The results of the regression analysis are shown in Table 16. There is no evidence of a step change in A&E attendance associated with the start of the ‘telephone first’ approach (p = 0.68). There is, however, strong evidence (p = 0.005) that, following the start of the ‘telephone first’ approach, the intervention practices’ rate of increase is slower than that expected given their pre-intervention trends and national trends, although this differential trend is small, at around 2% per year (RR 0.98, 95% CI 0.97 to 0.99) and against a background of A&E attendances that increased by approximately 100% across the study period from April 2008. Although on average there was no change seen in A&E attendances coinciding with the launch of the intervention, there was moderate heterogeneity (variation) in the effect of the intervention. The model suggests that, although on average there was no step change at the transition to the ‘telephone first’ approach, some practices saw increases in A&E attendance of up to 10%, whereas others saw decreases of up to 8% (95% reference range RR 0.92 to 1.10). The results for outpatient attendances are almost identical to those seen for A&E attendances: there was no evidence of a step change at the transition and there was a 2% decrease relative to background trends per year, with slightly greater heterogeneity between practices.

There is evidence that inpatient admissions increase, on average, by about 2% following the transition to the ‘telephone first’ approach (RR 1.02, 95% CI 1.01 to 1.03; p = 0.006); this increase is twice as large in the case of admissions for ACS conditions (RR 1.04, 95% CI 1.01 to 1.03; p = 0.032). There is no evidence that inpatient admissions further change relative to the background trend, although there is an additional increase of around 6% per year for the ACS admissions (RR 1.06, 95% CI 1.02 to 1.11; p = 0.007). When admissions are split into elective and emergency admissions, there is evidence of a step increase in emergency admissions in line with that seen for all admissions (transition RR 1.02, 95% CI 1.00 to 1.04; p = 0.02), and there is no evidence (p = 0.56) of a step change in elective admissions. However, care must be taken to not interpret the lack of evidence of a change in elective admissions as evidence of an absence of a change due to the reduced sample size in these analyses, in which the CIs are wide and substantially overlapping and may be consistent with a common change across the two admission types. There is evidence that elective admissions continue to increase beyond background trends post transition, but no evidence of a similar increase in emergency admissions; again, there is a substantial overlap in CIs. As with A&E and outpatient appointments, heterogeneity in the step change was seen, such that for each outcome considered there were practices that had increased admissions at the transition and others that had decreased admissions at the transition.

Primary care costs

Secondary care costs

Hospital Episode Statistics data were obtained from NHS Digital between April 2008 and March 2016 for intervention and control practices, as described in Chapter 7. The field ‘SUSHRG’ (secondary uses service health resource group) was matched to the relevant Healthcare Resource Group (HRG) 2015/16 reference cost83 for A&E, outpatient contacts and inpatient episodes of care (finished consultant episodes). When there was no valid HRG, a weighted average cost of the HRG stem (e.g. AA23) or HRG root (e.g. AA) was applied. When there was still no valid HRG, an average unit cost for an A&E, outpatient or inpatient cost was applied. 83 To provide context, the mean cost in 2015/16 was £138 for an A&E attendance, £137 for an outpatient procedure and £3750 for an elective inpatient admission. The cost of referrals per 1000 patients in intervention practices was then calculated for the 12 months prior to and the 12 months post the introduction of ‘telephone first’, and the crude change was calculated. The step change attributable to ‘telephone first’ was then calculated by multiplying the costs in the 12 months prior to the introduction by RR – 1, where RR is the proportionate step change reported in Table 16. The slope change attributable is calculated by multiplying the cost in the 12 months prior to the introduction by exp[ln(RR)/2] – 1, where RR is the proportionate slope change reported in Table 17. The total change is the sum of the two.

Table 18 summarises each component of the analysis, the method of analysis and data sources.

Additional cost analyses

At the request of the study PPI group, we included some exploratory analyses of patient out-of-pocket costs. These were limited to travel costs, because a more thorough investigation of, for example, loss of earnings was beyond the scope of this study. Car travel was costed at Her Majesty’s Revenue and Customs (HMRC)-approved mileage rates. 84 Bus travel costs were obtained from Department for Transport statistics. 85

Costs of initiating and operating the ‘telephone first’ approach

Other primary care costs

Secondary care costs

The analysis of HES in Chapter 7 shows that there was, on average, a small (2%) increase in admissions when practices moved to the ‘telephone first’ approach; this increase was mainly concentrated in emergency admissions. Subsequently, the introduction of the new system was associated with a small (2%) reduction in the rate of increase in A&E attendances and a small (2%) increase in the previous rate of increase for elective admissions.

Table 21 shows the changes in costs of secondary care utilisation. Combining the step change at the introduction of the ‘telephone first’ approach with the change in the underlying trend over the subsequent year, there were small non-significant reductions in costs for A&E and outpatient attendance and a significant increase in costs of admissions, leading to an estimated overall increase in secondary care costs of £11,776 per 1000 patients in the first year after changing to the new approach (aggregating A&E and outpatient attendance and admissions coded as emergency or elective).

Patient out-of-pocket costs

An exploratory analysis was performed to predict the impact of the introduction of the ‘telephone first’ approach on patient out-of-pocket costs. We limited these costs to travel costs, which were calculated assuming that travel was by car or bus; we did not attempt to calculate other costs (e.g. loss of earnings). Travel by car was costed at £0.45 per mile, multiplied by the mean distance of the patient’s home address from a GP surgery within each lower-layer super output area (LSOA). 86

The mean distance to the closest GP surgery across all LSOAs was 1.587 km (SE 0.00845 km) (0.986 miles). Multiplied by £0.45 per mile, this results in a round trip cost of £0.89 for each car-based trip to see a GP face to face. The mean cost per bus passenger journey is £1.53, or £3.06 for a return journey, for each face-to-face GP appointment. 85 The mean number of primary care visits per patient per year in 2008/9 was 5.5 (median 5.46 visits, interquartile range 4.85–6.22 visits), equating to a cost of £4.90 (for travel by car) or £16.83 (for travel by bus) per year. This represents the maximum out-of-pocket saving to a patient should all of their consultations be via telephone rather than face to face. For the estimate of a 28% average reduction in face-to-face consultations (see Chapter 3), this would be equivalent to an average annual saving of £1.38 (for travel by car) or £4.71 (for travel by bus).

This estimate assumes that people will choose to go to their nearest GP surgery. As there are many other factors influencing GP choice that could result in a patient choosing a more distant GP, the distance and, hence, cost savings could be higher than estimated. The potential savings would also be higher for those who visit their GP more frequently.

In summary, this exploratory analysis suggests that the potential for out-of-pocket savings for patients as a result of a ‘telephone first’ approach is small. Other factors, such as convenience and waiting time to see a GP, are likely to have a larger influence on patient preferences.

Site selection, sampling and recruitment

Qualitative interviews with patients were undertaken in 12 GP surgeries. The participating practices were located in a number of areas of England: the north-east, the north-west, the Midlands, East Anglia, London, the south-east and the south-west. These practices formed a subset of the practices that were sampled for the patient and carer survey. Surgeries were selected from the sample of 20 to include those with a range of experiences in adopting the ‘telephone first’ approach, including those with positive experiences and those that had experienced or overcome problems. However, it should be noted that the practices that agreed to take part in this element of the study were probably those in which the appointment systems were working better than the appointment systems in those practices that declined to participate (see further details in Chapter 4).

In the first instance, patients who were potential participants indicated their interest in being contacted for an interview by returning a reply slip that accompanied the patient and carer survey (see Chapter 4). Purposive sampling of those who expressed an interest was carried out by the research team, based on survey responses. We sought to gain a range of views and to ensure that the following groups of people were included: older people, people who work outside the home, people with disabilities, people with chronic conditions and people whose first language was not English. Selected interested participants were contacted by a member of the research team by the preferred contact mode (telephone or e-mail) indicated in the reply slip and were invited to take part in an interview.

Data collection

Semistructured interviews were conducted by four researchers (SB, JN, JC and JE), either at the patient’s home (n = 12) or at his/her GP surgery, as requested by the patient. All interviewees gave written consent to be interviewed. A common interview guide informed by the literature was used for each interview, although emphasis was given to allowing participants to talk from their own perspective. The focus of the interview was on patients’ and carers’ views of the advantages and disadvantages of the ‘telephone first’ approach, including its convenience, perceptions of quality of care and impacts on the doctor–patient relationship. Interviews were audio-recorded with the participants’ permission, and transcribed verbatim. Transcripts were anonymised by removing references to identifiable names and places.

Data analysis

Data analysis proceeded in parallel with data collection and informed the iterative development of the interview topic guide and observation protocol. Thematic analysis of the data was conducted based on the principles outlined by Boyatzis. 87 Transcripts were read and re-read, and codes were applied to meaningful sections of text. Coding was conducted by Sarah Ball, Jennie Corbett, Josephine Exley, Jennifer Newbould and Emma Pitchforth. As analysis progressed, codes were organised into overarching or organising themes using NVivo version 10 (QSR International, Warrington, UK). Data within themes were scrutinised for confirming and disconfirming views across the range of participants. The emerging findings were shared and discussed regularly among the study team.

Does the ‘telephone first’ approach work for patients?

The nature of the responses of patients to the ‘telephone first’ approach was extremely varied: although some patients reported being highly satisfied with the approach, others found it to be unacceptable or that it did not meet their needs. Many participants provided strong positive endorsements:

I really don’t find anything difficult [about the system] – I really don’t. I think that it works so well.

105_1040

In contrast, a small number of patients reported being so dissatisfied with the approach that they were considering moving to a different practice. Others were ambivalent, considering the new approach to be the least bad option:

I’m prepared to stick with it because, I mean, going back to the old system, no. That’s even worse.

106_1077

In describing their experiences of the approach, patients outlined a broad range of advantages and disadvantages across five distinct categories: (1) access to care, (2) quality of consultation, (3) continuity of care, (4) patient safety and (5) confidentiality. Within each of these categories, there were contrasting views with respect to whether the introduction of a ‘telephone first’ approach was perceived to represent an improvement or a hindrance.

What factors influence whether or not the ‘telephone first’ approach works for patients?

As outlined in the previous sections, there were a range of patient views regarding the nature of the advantages and disadvantages of the ‘telephone first’ approach; however, it was also apparent that the weight or value given to particular advantages and disadvantages varied significantly. For example, a disadvantage that represents a mild annoyance for one patient could represent a ‘deal-breaker’ for another, rendering the approach completely unacceptable. With reference to the descriptions given by patients regarding how the ‘telephone first’ approach worked in the practice they attended (see Table 22), it is clear that there was considerable variation between practices in how the approach was implemented and the nature of the challenges experienced by patients using it.

Site selection, sampling and recruitment

Qualitative interviews were undertaken with staff at 12 GP surgeries from which patients were recruited. The methods for practice selection are described in Chapter 4. Up to five members of staff were recruited from each practice. Practices selected staff to participate based on who they deemed most suitable to talk about the ‘telephone first’ approach within their practice. In general, two GPs, a practice manager and a receptionist/administrative staff member were interviewed.

Data collection

Semistructured face-to-face interviews were conducted by four researchers (SB, JN, JC and JE). Two interviews, both with GPs, were conducted by telephone (by JN and EP): in one case, the GP was not at work because of sickness and asked to be interviewed over the telephone, and, in the other case, the interview was held after the practice visit at a more convenient time for the GP. All participants gave written consent to be interviewed. A common interview guide informed by the literature was used for each interview. The interview explored the reasons for switching approach, the setting-up process, perceptions of quality of care and safety as well as impacts on the doctor–patient and intrapractice staff relationships. The advantages and disadvantages of the ‘telephone first’ approach were also discussed.

With participants’ permission, interviews were audio-recorded and transcribed verbatim. For one practice, audio-recordings were unavailable because of technical problems, and detailed notes were taken. All transcripts and notes were anonymised. Data analysis followed the same methods as those described for the patient interviews (see Chapter 9).

Respondents interviewed

A total of 49 members of staff from the 12 practices were interviewed. The number of staff members interviewed in a single practice ranged from three to five, with four interviews being completed in the majority of practices. Twenty-one (43.8%) of the interviews were conducted with a GP, 14 (29.2%) with reception/administrative staff, 10 (20.8%) with practice managers, two (4.2%) with a nurse and one (2.1%) was a joint interview with the practice manager and an administrator.

The 12 practices were located in a variety of geographical locations across England and were from seven Clinical Research Network areas. Seven practices had paid for the support of commercial companies to adopt a ‘telephone first’ approach; for five of these, fees were paid by the CCG. Practices varied regarding the level of deprivation in the areas they served, with two practices noting particular deprivation among their practice populations, with high rates of unemployment (practice 103) and high numbers of refugees and immigrants (practice 108). Seven practices had been supported by Doctor First and five were supported by GP Access. Some additional characteristics, such as practice size, of the practices from which staff were from are given in Table 22.

This chapter explores staff views on the adoption of a ‘telephone first’ approach in primary care. The chapter first examines why practices chose to implement a ‘telephone first’ approach and their experiences of adopting the system. We then explore the implications for staff and the practice and staff perceptions of the impact on patients. The chapter concludes with the enablers of and barriers to the adoption of a ‘telephone first’ approach in general practice.

Why adopt a ‘telephone first’ approach?

In all 12 practices, participants identified problems meeting demand as a key reason for changing to a ‘telephone first’ approach, including that, under their previous systems, patients had often had to wait a long time to see a GP:

So, you know, it was getting up to, sort of, 3 or 4 weeks, you know, before people would get a routine appointment and we had a duty doctor system, where the duty doctor was here and booked on the day. We were finding that was getting incredibly onerous and stressful for the duty doctor because he might get, oh I don’t know, 70, 80 sometimes even 100 calls in a day.

Practice manager 5001, practice 105

A number of respondents, from different practices, described how, over time, the practice had struggled with demand for appointments and had added extra face-to-face appointments or telephone appointments to manage demand. GPs in particular often mentioned that struggling with their workload was a motivator for trying a different approach:

I think the thing which worked for everyone was obviously altruistic motives like ‘yes we want to see patients who need to be seen’. But also managing our own workload and knowing what comes through the doors was the most important selling point if I can remember all the years back, what worked for others.

GP 5001, practice 108

. . . our system was ‘ring a week ahead’, so ‘ring on Monday for the following Monday’, but the phone lines opened at 8 and by 5 past all the appointments were gone. And we would get a huge queue; people would start queuing before 7 o’clock, so you’d get this awful queue at the door and you’d finish your morning surgery and there’d be like 30 extras waiting to be seen.

GP 5003, practice 112

Often, the decision to change approach related to a number of factors, but some participants described something that made the situation come to a head, for example a member of staff leaving. Two GPs noted particular cases in which patients had waited an unacceptably long time to see a GP as being factors in their decision to change to the ‘telephone first’ approach. In one case, a patient had died, possibly as a result of delays in being seen, and, in another, the patient had waited for an appointment for a condition that required urgent attention.

A concern for fairness and equity in the system was also often mentioned by staff. Many felt that a traditional booking system favoured those who were quickest to call the surgery rather than those in the most clinical need:

Just a feeling of unsustainable demand and also a real feeling of sort of unfairness in our access to the surgery, we had patients who shouted loudest at the desk got seen for things that weren’t urgent and things that potentially should have been seen quickly waited 2 or 3 weeks.

GP 5004, practice 101

Other reasons for changing to the ‘telephone first’ approach included a large number of appointments that patients sometimes do not attend, and CCGs offering funding to cover input from the commercial companies.

During interviews, a number of practice staff mentioned that one person at the practice had led the introduction of a ‘telephone first’ approach; this was usually a GP, but sometimes a practice manager. As well as researching the concept and liaising with the commercial companies, this individual often provided enthusiasm and leadership for the change, which enabled other staff to learn about the approach and raise queries or concerns.

Participants described the process by which practices had made the decision to switch to a ‘telephone first’ approach. For some practices, this involved doing research and visiting other practices using the approach. Many participants stressed the importance of all practice staff being brought into the discussion as an enabler of successful adoption. In a number of practices, particular individuals, usually GPs, had been reluctant to change to the ‘telephone first’ approach.

Implementing the ‘telephone first’ approach

Practices varied in the extent to which they were using telephone consulting prior to the adoption of the ‘telephone first’ approach. Some had a conventional booking system, in which all patients were pre-booked into face-to-face appointments. Others were already using some form of telephone triage for appointments, for example emergency appointments or when face-to-face slots were fully booked.

All participating surgeries had used a commercial company (Doctor First or GP Access) to support the transition to the ‘telephone first’ approach. Over time, most practices had modified the ‘telephone first’ approach they had adopted, although all practices had largely maintained the key component that face-to-face appointments with a GP had to be preceded by a telephone conversation with a GP. Exceptions to this were usually very limited, for example 6-week baby checks being pre-booked. However, there was some variation between practices, with some GPs booking some patients into slots in advance.

Practices varied in the extent to which they would make modifications to the way the ‘telephone first’ approach was outlined by the commercial companies. A number of practices had made a number of modifications to the ‘telephone first’ approach, and one surgery was continuing to make modifications to the approach. Although deviation from the ‘prescribed’ approach was sometimes described as being discouraged by the commercial companies offering support, some practices felt that modifications helped them to overcome challenges and tailor the approach for their population and staff.

It was evident from the interviews that the ‘telephone first’ approach was operating more smoothly in some practices than in others. In nine practices (100, 101, 103, 104, 105, 106, 108, 112 and 114), staff, overall, felt that the ‘telephone first’ approach was working well, although they had experienced a number of challenges. In one practice (110), all staff identified problems with the approach; this was also evident in interviews with patients (see Chapter 9). In two practices (102 and 117), there was a varied response to how the ‘telephone first’ approach was working. Even in practices in which it was felt that the approach was working, there was variation between staff members: although one GP might feel that the ‘telephone first’ approach was working well, another GP in the same surgery might have found the new system more challenging. At the end of the interview, all participants were asked if, given the choice, they would prefer to retain the ‘telephone first’ approach or revert to a conventional appointment system. Of the 46 participants who answered this question, 41 replied that they would prefer to stay with the ‘telephone first’ approach, one was unsure and four preferred to revert to a traditional appointment system, three of whom were from the same practice. However, it should be noted that the practices that agreed to take part in this element of the study were probably those in which the appointment systems were working better than those that declined to participate (see further details in Chapter 4). However, even those participants who indicated that they would like to keep the ‘telephone first’ approach noted ongoing challenges:

Whenever we’ve talked about it we’ve said that we’d stick and I think that’s because of this . . . it’s worry about if we didn’t speak to them today we’d be back to this 3-week thing [wait to see the GP].

GP 5004, practice 105

Yes, I mean it is fraught with difficulties and problems but there is no way we would go back, no way.

GP 5002, practice 112

Patient safety

Given the literature on concerns about safety regarding telephone triage approaches, all respondents were asked about this issue during the interviews. The majority of GPs felt that a ‘telephone first’ approach was safer than a traditional booking system because all patients who want an appointment with a GP would, at a minimum, speak to a GP on the same day:

. . . our health system is changing, there isn’t enough time, so you’ll either change with it. Or you let it crumble and the doctors who are saying, ‘Well, we don’t think it’s safe’, well, they’ve got 4-week waits to be seen. And, well, you’ve got all those patients who you’re holding back onto [an] iceberg; you don’t know what’s wrong with them, so how’s it safe to have them waiting 4 weeks? So it doesn’t make sense to have that argument.

GP 5002, practice 104

Several GPs spoke about how they dealt with safety when speaking to the patient by telephone and the importance of ‘safety netting’; for example, many GPs stated they had a low threshold for bringing people in for a face-to-face consultation and would call a patient in if anything concerned them:

I mean, I feel if there’s any, sort of, if I’m speaking to someone and I’m not sure what’s going on and I just don’t quite trust what I’m hearing I would always then get them in.

GP 5003, practice 105

A few GPs who were interviewed spoke of individual attitude to risk as being particularly important when using a ‘telephone first’ approach, and also in general practice more widely. Some GPs were said to be more comfortable with risk than others. This might be influenced by how long the doctor had been a GP, how well they knew the patient, how much telephone consulting they had done previously and their personality:

So the people who have done a lot more telephone triage are probably . . . it’s all about you managing risk. So it’s about how much risk you’re willing . . . I suppose that’s being a GP is we manage risk.

GP 5004, practice 106

General practitioners’ experiences of the ‘telephone first’ approach

The GPs were asked how the ‘telephone first‘ approach had affected their working day, compared with working with a traditional appointment system.

All GPs were speaking to more patients than under the traditional system, but they were seeing fewer patients face to face. For some GPs, this was a better way of working:

. . . it’s still hard work, and sometimes it’s hard in different ways because you’re making more decisions because you’re dealing with more patients, so it actually can be quite exhausting even though you aren’t seeing as many patients . . . It’s really nice, so at the end of the day if it’s gone to plan we’ve spoken to everybody that wants to be spoken to that day . . . we’re sort of on top of our demand – that psychologically is a real plus to think that’s it, we’re done, we’re sorted.

GP 5004, practice 101

Some mentioned that face-to-face appointments, under the ‘telephone first’ approach, tended to be more intense than under a traditional appointment system:

So one of the big problems with it is the appointments we do book are much more intensive. So, you know, under our previous system I would pretty, much every surgery, have a couple of catch-up appointments. You know, I’d have somebody just needed their sick note renewing for whatever reason and sometimes it could take a long time but quite often that would be very quick, or a pill check . . . Whereas, all that now is gone because it’s done over the phone. So actually the only people who make it through the door are the people . . . none of those really short, dead easy, ‘I’m going to catch back a bit of time’. They’ve all gone because it’s all done on the phone.

GP 5004, practice 117

There was considerable variation in individual views about changes in workload. For GPs at one surgery, the ‘telephone first’ approach meant less stress, less workload and shorter working days:

We definitely go home earlier, definitely. At the time [before the ‘telephone first’ approach], 3 years ago, I used to work a Monday evening and, in theory, we should finish at 6.30 [p.m.], I was still here at 8 o’clock most Monday evenings. Now, our Monday evening team, by a quarter to seven, they are gone, and all the patients have been managed and seen.

Administrator/receptionist 5001, practice 114

However, in some other practices, GPs were concerned about increases in demand:

So my main worry about this is that demand has increased and continues to increase because we are so accessible and there is no barrier there.

GP 5004, practice 117

Despite implementation of the ‘telephone first’ approach, a few practices noted that they were overwhelmed by demand. They had experienced days when there were too many requests for telephone calls and not enough slots for GPs to call them all back, even if GPs were willing to work late. In such instances, practices often had an approach of turning the telephone lines to ‘emergency calls only’. The time of day when this took place varied: for some it was 5 p.m., for others it was 3 p.m. and for one surgery, which was struggling with the approach, it was quite early in the morning:

. . . when telephone triage first started the latest was 2 p.m., now the latest is probably 11:30 a.m. and that’s an exception.

That’s when you go through to emergencies only?

Yeah, I mean this morning it was, oh my god I’m not sure what time it was this morning, I wasn’t . . . I didn’t really keep an eye on it. Quite often it’s before 9 a.m.

Practice manager 5001, practice 110

A few GPs reported that patients had started to telephone them about things because they were available, for example to tell them about having a poor night’s sleep or to ask for advice when their child had diarrhoea for half an hour:

I think there is evidence that we take calls about things which are basically because we’re so easy to access. In years gone by, you know, I’m not sure I would have had people ringing up and just checking about the ingredients of their shampoo or their medication. Just because they couldn’t have done.

GP 5004, practice 117

In contrast, one GP did not feel that demand had increased and, in fact, noted a decline in demand as a result of the telephone first approach:

And I think we’re now, I think 2 years in, I think we’re seeing a reduction in demand, and there is some evidence that this does happen, I think we’re seeing a reduction in demand for things that we don’t need to see.

GP 5004, practice 101

Some GPs liked the flexibility that the ‘telephone first’ approach afforded them to arrange their workload and felt that it gave them more control over their day as it was them who called patients and allocated them, if appropriate, to face-to-face appointment slots. Others liked, on occasion, being able to book longer face-to-face slots for complex cases. Conversely, other GPs reported that they found the system harder in terms of balancing the callbacks with other tasks such as supervising students, home visits and paperwork. Practices and individuals had varied approaches to address such problems, for example by not having any face-to-face appointments until later in the day.

Many GPs mentioned the unpredictable nature of the day. Unlike a traditional appointment system, in which a GP comes in to a diary full of appointments each day, under the ‘telephone first’ approach, the diary started empty of appointments each morning. Telephone consultations then began and face-to-face appointment slots would begin to be booked in. Some respondents liked the unpredictable nature of the day, whereas for others it brought some anxiety about the day ahead. Some GPs felt that the fact that calls came in continually throughout the day was a strain:

And things like meetings; you know you have a meeting and you can see that people are sitting there getting twitchy, you know, looking at their watch thinking I’ve been in this meeting an hour, how many more telephone calls are going to be on my screen now than there were an hour ago and how much time have I got to deal with them, am I actually going to be able to get through before 7 o’clock or whatever.

GP 5002, practice 102

A number of GPs felt that they were working in a more isolated way under the ‘telephone first’ approach, as they spent more time on their own in their rooms making telephone calls rather than seeing patients face to face. A number of surgeries had introduced measures to try to mitigate this, for example joint coffee and lunch breaks so that GPs could talk to each other or shared call lists to increase the feeling of teamwork.

The experiences of reception staff

One component of the ‘telephone first’ approach, as advocated by the two commercial providers, is for reception staff to take a brief note of the patient’s problem. This is to assist GPs with deciding in which order to call back patients, with patients with more serious complaints being called back first. From interviews at different surgeries, it became clear that the role of the receptionist in the ‘telephone first’ approach varied. In one surgery, the receptionists took no details of the patient’s complaint; in some surgeries, they took minimal details about the condition. In other surgeries, reception staff had been trained to take a brief reason for the call and to signpost patients to alternative routes of care if appropriate. Some practices had modified the system so that the GP could flag patients they wished the receptionist to book for a face-to-face appointment based on the description taken by receptionists. When reception staff had more of an active role, it appeared to alleviate some of the pressures from the GP and was seen as an enabler of the successful adoption of a ‘telephone first’ approach.

Reception staff spoke about the change in their role and the increased responsibility now that they were taking details of the condition from the patient. Most reception staff felt empowered by this enhanced role, although a few felt associated anxiety about potentially serious cases:

Chest pain or symptoms of like a stroke but again it’s really hard but if we are just worried then we can let the doctor know because we’ve got instant message as well, we can always just do, or ‘Not sure about this call. You might need to take it’ or we could just simply pop them on hold, give the doctor a ring and then the doctor can say ‘Well I’ll call them back or I’ll take it’.

Administrator/receptionist 5003, practice 108

Some reception staff felt that their role was less pressured as they no longer had to spend the day telling patients that there were no longer any appointments available, as had happened under the previous system. Several staff who were interviewed spoke of how the ‘telephone first’ approach had led to a happier reception team:

. . . the admin staff like it in the fact that they don’t have patients shouting at them now, like, ‘What do you mean you haven’t got any appointments? It’s only, you know, 8:45 a.m. and how can you have run out of appointments already?’ That sort of thing.

Practice manager 5001, practice 102

Although the nature of the working day had changed for reception staff, there was a consensus among surgery staff that the overall workload of reception staff had remained the same; in particular, the telephone lines often continued to be busiest at the very start of the day.

Changes in practice culture

Some respondents commented on how the ‘telephone first’ approach had altered the culture of the practice, sometimes for better and sometimes for worse. Some felt that the staff had become more cohesive. For example, in one practice, clinicians and administrative staff felt that there was greater appreciation of each other’s roles because the duty GP now took calls in the reception office. Others felt that it had brought staff together in a shared commitment to try to make ‘telephone first’ approach work:

The doctors now, sounds horrible if you say they look out for each other now, not that they didn’t before but they are far more supportive of each other than they were before.

Practice manager 5003, practice 106

Some participants spoke of the importance of the practice agreeing rules of how the system would work so that patients negotiated the system in the same way each time, and they saw this as an enabler of the successful adoption of the approach:

. . . someone who has got a sore throat will ring up the GP instead of going to the pharmacy. I think we circumvent that, partly . . . we do limit the call numbers, partly ‘cause we are reasonably singing from the same hymn sheet; so, if you do ring up with a sore throat you know you’re going to get told to go to the pharmacy, so you know people don’t tend to do that.

GP 5003, practice 112

There were a number of reported conflicts in situations in which not all GPs used the ‘telephone first’ approach as intended. In one practice, one particular GP opted out of the ‘telephone first’ approach and had turned all his appointments into face-to-face appointments; that GP had since left the practice. In another practice, one GP refused to use the ‘telephone first’ approach, and so saw patients only face to face, booked in through telephone consultations by other GPs, and this had caused some confusion for the patients.

In some practices, respondents described how the ‘telephone first’ approach had led to greater divisions between administrative and health-care staff, with GPs rarely leaving their rooms. This contrasts with the practice reported above, in which staff had become more cohesive; this emphasises the large variation between practices in how the ‘telephone first’ approach had worked for them.

Staff perceptions of implications for patients

We asked staff about the effect of the ‘telephone first’ approach on groups of patients who might be adversely affected by the approach, including patients for whom English was not their first language, older people, deaf or hearing-impaired people and patients without telephones. Respondents often commented on the nature of the population that the surgery served and the impacts this had on the implementation of the ‘telephone first’ approach. For some practices, the population they served acted as an enabler of the approach, for example if most patients spoke English as their first language or had their own transport.

Two practices in deprived areas identified elements of their populations that were challenges to the ‘telephone first’ approach:

The population which it really doesn’t work with is our, kind of, immigrant population, our asylum seekers and refugees. Sometimes there’s language problems and problem with expectations; we have a low threshold for calling them in. The only [way] it helps is that we can arrange [an] interpreter for them, so rather than them booking an appointment and turning up without an interpreter.

GP 5001, practice 108

For other practices, challenges were identified to be high numbers of people receiving state benefits, unemployed people or people with learning difficulties. Perhaps surprisingly, these practices had successfully adopted the ‘telephone first’ approach and noted that they had made adaptations to the system for these groups. In fact, for those with a chaotic lifestyle, the immediacy of having the appointment on the same day was said to have been advantageous.

All practices had systems in place for deaf or hearing-impaired people. These included the ability to book straight into face-to-face consultations, being able to text the practice to request an appointment or a carer or relative being able to speak to the GP on a patient’s behalf.

Respondents spoke of services that were available for individuals who may find it challenging to navigate the ‘telephone first’ approach in the traditional way:

. . . some of them we have a flag on [the clinical system] saying ‘if this person rings up for a consultation just book them in’ because sometimes, particularly when we have say hard-of-hearing, deaf patients, vulnerable, learning difficulties, we just book those in, they don’t have a call anyway. And there are some patients that don’t fall into these groups that still it doesn’t work on the phone. But they are few and far between really and most people . . . we’re all used to using the phone now.

GP 5004, practice 101

In addition to provision for patients with certain conditions, staff often spoke about individual patients for whom an alternative system had been set up, for example for a homeless patient and for a patient with memory problems. The default was often that such groups would get booked directly in for a face-to-face consultation.

It has been suggested that elderly patients may be disadvantaged by increased use of telephone consulting, but the majority of staff who were interviewed felt that this was not the case and that most elderly people liked the system once they had experienced it. A few GPs did note that some elderly patients missed the contact that a face-to-face consultation afforded. Some participants noted making special arrangements for elderly patients who relied on family and friends or public transport to visit the surgery, as they were less able to come to the surgery immediately if a face-to-face consultation with the GP was thought necessary.

Owing to the nature of the ‘telephone first’ approach, which relies on the patient being available to take a telephone call from a GP, other groups of people for whom the system may be a challenge were identified. A number of respondents mentioned the need to help patients who worked in roles in which they were unable to take telephone calls during the day:

. . . then you have doctors, teachers, call-centre workers, taxi drivers, HGV [heavy goods vehicle] drivers, policemen, people who work in large open-plan offices who are very unhappy.

GP 5004, practice 105

Some surgeries, in such situations, asked patients to indicate a time during the day during when they could be contacted, for example during their lunchbreak or once they had finished work.

A number of participants reflected on the provision of a service for hard-to-reach groups that took away the telephone call element of the approach. Although necessary for some patients, there was concern that it provided an inequitable service for a group of patients who automatically would be given a face-to-face consultation:

Yeah and sometimes, I mean, we’ve discussed that whether it’s . . . that it’s an unfair advantage to them because there’s language problems, they get appointments.

GP 5001, practice 108

In the case of some groups, such as patients who claimed to not have a telephone, there was sometimes concern that patients were using this excuse as a way to directly obtain a face-to-face appointment without the need for a telephone call with a GP. There was also a feeling, among some participants, that the system would not work if too many patients were exempt from the telephone call element of the approach.

Staff perceptions of patients’ opinions of the system

Many staff felt that patients were pleased with the ‘telephone first’ approach, and they reported positive feedback from patients. Respondents also pointed to positive feedback from NHS Friends and Family Tests and NHS Choices indicating patients’ satisfaction with the ‘telephone first’ approach. One practice had noticed an increase in the number of patients registering with them who had previously been registered with another local practice. Again, in interpreting these data, it should be noted that the practices that agreed to take part in this element of the study were probably those in which the appointment systems were working better than in those that declined to participate.

Some respondents felt that patients preferred the ‘telephone first’ approach, but that it sometimes took time for them to get used to it:

I think the hardest part at first was obviously make sure the patients understood how it worked. Obviously information had gone out to them prior, etc., but obviously there’s always a little bit of confusion . . . So I think it was probably a little bit more time-consuming at the beginning and obviously to have a few patients who weren’t very happy. But I think as time went on and obviously people became aware of the system and realised that actually they get to speak to a doctor quite quickly. If they need to be seen they get seen more soon, you know, I think that resistance tapered away and people came to realise that actually it’s a better system.

Administrator/receptionist 5003, practice 101

One respondent felt that the approach took a while to settle in, and that it took 2 years to see benefits. Practices varied in their approach to determining if a face-to-face consultation was required. One practice had a policy of ending all telephone consultations by checking that the patient was happy to not be seen face to face. In others, the decision was solely that of the GP.

In one practice that was struggling with the approach, respondents believed that the number of patients was declining as a result of the ‘telephone first’ approach, because patients had disenrolled to register at other local surgeries. Patient feedback had been very negative in this practice:

. . . we’ve had an awful lot of complaints but, I mean we seem to have a very complaining population anyway. We were getting very demoralised with NHS Choices, all the sort of comments on there, I think we ourselves sympathise with some of the patients’ comments.

Practice manager 5001, practice 110

Impacts of the ‘telephone first’ approach on the nature of consultations

Some GPs did not like communicating on the telephone and preferred face-to-face consultations, although a number of GPs felt that their telephone consultations had improved as they had become more experienced. A few GPs reported that the ‘telephone first’ approach had enabled them to practise medicine in a different way to that enabled by a traditional appointment system. For example, they could give a patient advice in the morning and then call them back in the afternoon to see if there was any improvement or if additional treatment was required. Some GPs also felt that they had more control over which patients were seen, rather than the reception team acting as a gatekeeper as is the case in a traditional appointment system.

The opportunity to educate patients was also mentioned, with more frequent contact meaning that there was the opportunity to teach patients about conditions or when they needed to contact a GP. One practice reported that demand had gone down, which a GP attributed to better patient education:

So the sort of education about the role of antibiotics in ear infections in children, you know very limited evidence, . . . if you’re otherwise happy with the child, they’re not unwell, we’re probably not going to do anything different today, call us back in a couple of days. So that message I think has got through . . . Certainly I feel I’m speaking to less people that really I think why did they phone about that, you know, I think those calls do . . . I think patients learn that they don’t need to call.

GP 5004, practice 101

Some GPs noted that the ‘telephone first’ approach also enabled them to appropriately book patients into appointments with other practitioners, such as practice nurses, for issues for which it was not necessary for the patient to see a GP:

We’ve realised; some of them can be directed to other people. Previously they just ended up with appointments with a GP when they actually could have gone and seen the practice nurse or the HCA [health-care assistant]. So we found that takes out a workload, it’s a better management of other clinicians’ appointments.

GP 5002, practice 104

One GP felt that the ‘telephone first’ approach led him to do more medical tests than perhaps were necessary:

. . . the ultimate is face-to-face consultations, but if you can’t do face to face maybe telephone is the best option in a way. But I think you need to do a lot more training. But then if you have 40 phone calls and you want to spend 10 minutes on a phone call to try and make sure you’ve not missed anything or do you go the other way and say well actually she’s complaining of this I’d better do all the blood tests in the world and then think well clinically why have I done those, well it’s cleared me as a conscience?

GP 5004, practice 110

Continuity of care

The GP respondents were asked about the impacts of the ‘telephone first’ approach on continuity of care. A number of respondents felt that there was no real difference in continuity of care under a ‘telephone first’ approach compared with a traditional booking system:

I suppose I don’t know. I don’t think that’s much different from our old system to the new one to be honest, with our particular group of patients anyway.

GP 5003, practice 105

Some GPs noted that continuity of care was becoming harder to deliver, but felt that this was related more to issues such as more part-time working than to factors associated with the ‘telephone first’ approach. When interviewed, several reception staff talked about trying to ensure that patients saw their preferred doctor if possible:

So like when they register the doctor’s names are on there and they can ask what days their doctor works, so most of the patients know what days their doctors work, so they’ll probably only phone on those days that they want a specific doctor, not if their doctor is not working then usually they don’t phone.

Administrator/receptionist 5002, practice 105

It was evident that practices varied in their policies; for example, some practice staff would ask the patient which doctor they would prefer to see or speak to, whereas others would give an appointment with a named GP only if it was specifically requested by a patient.

Patient education and knowledge of the ‘telephone first’ approach

Patient education and knowledge of the system was seen by respondents as an enabler of the successful adoption of the ‘telephone first’ approach. Prior to launching the ‘telephone first’ approach, practices had communicated the change to patients in a variety of ways, often using material provided by the commercial companies. There was variation in the extent to which this was done: some practices had written to every patient registered with them, whereas others had put notices up in the practice. Respondents also varied in the importance they placed on patient education:

So it was planning and patient education – patient education I think was the most important part I think. I know one of our neighbouring practices tried it and didn’t get on and I think that’s where they fell down was the patients didn’t understand what they were trying to do.

Administrator/receptionist 5003, practice 101

Some respondents spoke of explaining to patients the importance of telling the receptionist the nature of their problem in order for them to be triaged appropriately. At times, patients struggled to understand elements of the system, such as the inability to pre-book for appointments on different days:

We do try and keep to the appointments that we like, the calls that we take today, are for today, not trying to overflow onto the next day. Because otherwise we get to the same point where we’re full again and it’s just, you know, it just goes on and on and on. But they [patients] do struggle with the ‘oh well why can’t you just put me down for tomorrow?’ And then we have to say sorry you have to call back from half past eight in the morning. They really struggle with that.

Administrator/receptionist 5002, practice 103

Practices that had attempted to adopt a ‘telephone first’ approach but had ceased using it

In our qualitative analysis, we were aware that the majority of staff from GP surgeries that were involved in the study had felt that the adoption of the ‘telephone first’ approach had been broadly successful. There was also evidence, from our quantitative data, that practices that took part in that element of the study had a somewhat more positive patient experience than those that declined to participate. As a result, we sought to conduct interviews with staff from two general practices that had adopted a ‘telephone first’ approach but had then stopped using it.

Using information from the commercial companies on practices that they knew had stopped using the approach, and practices identified from the practice manager survey as no longer using a ‘telephone first’ approach, practice managers were phoned and invited to participate in the study. As in the main study, recruiting to this element of the project was challenging, but two practices agreed to participate. Interviews were conducted as in the main study, with the topic guide modified accordingly. Interviews were transcribed verbatim and analysed following the same approach (see Chapter 10, Methods, for more detail).

Four interviews were conducted with five members of staff from two GP surgeries. Interviews were conducted with GPs (n = 3) and one joint interview was conducted with an administrative manager and member of the administrative team. Three interviews were conducted face to face and one was conducted by telephone.

As in the main study, staff from both practices described being motivated to adopt a ‘telephone first’ approach because of problems with demand and concern that patients were waiting a long time for appointments with a GP. However, all staff reported how the ‘telephone first’ approach had been challenging from the start as a result of the high number of telephone calls it generated. All of those interviewed spoke about the pressure on GPs, with GPs reporting not enjoying their work as much, feeling stressed and feeling more isolated:

. . . it also felt sometimes there was so much work and you just saw a growing list of patients and you were just trying to get through it and you were trying not to bring people down [to the surgery].

GP 5001, practice 201

Participants at both practices mentioned challenges with staff retention as a factor that had an impact on use of the ‘telephone first’ approach. In one practice, two partners and four salaried GPs left in a 2-year period. The other practice lost two partners in quick succession. Respondents noted that reasons for GPs leaving were not solely attributable to the ‘telephone first’ approach, but they often commented that it had not helped. Matching capacity to demand is key to the ‘telephone first’ approach as outlined by the commercial companies; this is something that both practices had struggled with because of staff shortages.

Both practices also attributed the challenging patient populations they served as a reason why the ‘telephone first’ approach had not worked:

We broke a lot of doctors by employing that particular service [telephone first] in the context of our patient population.

GP 5003, practice 202

Although they were in different geographical locations, both practices reported that their patient population was in an area with a high level of deprivation, with a high prevalence of drug and alcohol problems and many patients experiencing mental health challenges. Nevertheless, it should be noted that the main study included practices with similar challenging patient demographics, and they had successfully adopted the approach. Two GPs, from different practices, commented that when they had abandoned the approach they saw patients who had not contacted them under the ‘telephone first’ approach:

It was interesting when we changed back to a different system, there are a whole load of really vulnerable people who we were suddenly seeing again that we hadn’t been seeing.

GP 5001, practice 201

All staff who were interviewed, particularly GPs, had found admitting that the ‘telephone first’ approach had not worked distressing:

I think it was hard to admit it because you feel like it should work . . . you know you do not want to be the one who has failed at something. I guess that’s part of being a doctor is not it? That we are quite competitive . . . you know saying you cannot do something is a difficult thing to say.

GP 5001, practice 202

I did not want to say anything because I felt like everybody else was probably fine and it was just me and then I had one of those unintended conversations with one of the other partners . . . so I said to her [that I was not coping with ‘telephone first’] and I just saw this kind of massive sense of relief and she said, ‘do you know I hate it and I think I’m going to have to leave if it carries on’.

GP 5001, practice 201

A decision to abandon the ‘telephone first’ approach posed challenges for all staff at the practices, and the period when practices decided how they would reconfigure the appointment system was described as difficult. Morale was often low because the ‘telephone first’ approach had not succeeded. Both practices had stopped using the ‘telephone first’ approach but had decided not to revert to a traditional booking system using all face-to-face consultations, but instead decided to retain some elements of telephone consultations. Both had adopted a hybrid approach, with telephone consulting used for on-the-day appointments and face-to-face appointments being available in advance with pre-bookable slots.

Although participants were upbeat about the future and the new approaches they had implemented, it was evident that for staff in practices that had unsuccessfully tried to implement the ‘telephone first’ approach, there could be considerable negative impacts on individual staff members and the practice as a whole.

Summary of quantitative findings

In our quantitative analyses, we principally report an intention-to-treat analysis, including the 89 practices that were operating the system as recommended by the companies alongside those practices that had modified the ‘telephone first’ approach and were using a hybrid form of telephone consulting and a small number of practices that had ceased running the system altogether. We also included a sensitivity analysis that was restricted to practices that we believed were running the system as recommended by the companies throughout the data collection period. In general, these two analyses produced similar findings.

Data from the computer systems of practices using one of the commercial companies showed that there had been a major effect on patterns of consultation as a result of the introduction of the ‘telephone first’ approach, with a 12-fold increase in telephone consultations and a 38% reduction in face-to-face consultations. There was an overall increase in the total number of consultations (face to face and telephone) of 28%. There were equally dramatic changes in time from booking to appointment time – from an average of 4 days for face-to-face appointments before the change to an average of 0.9 days after the change. The length of consultations was similar before and after the change (10.9 minutes before and 10.2 minutes after for face-to-face consultations and 7.7 minutes before and 6.2 minutes after for telephone appointments), with an overall increase in length of surgery consulting time of around 8% following the change (face-to-face and telephone consultations combined). Although these results were broadly similar in the sensitivity analyses that were restricted to practices operating the new system throughout the period of data collection, there was wide variation in patterns of consulting in individual practices, with some experiencing a marked decrease in overall consulting time and others experiencing a substantial increase. Practices also varied greatly in the proportion of telephone consultations that were followed by a face-to-face consultation.

In the controlled analysis of A&E attendances with use of HES, the numbers of A&E attendances before and after the launch for patients registered in practices using a ‘telephone first’ approach were compared. There was no evidence that use of a ‘telephone first’ approach led to a change in the average rate of A&E attendances. There was evidence of a slower increase in attendances for intervention practices than for control practices, but the trend was very small. We also noted small increases in emergency admissions following the introduction of the ‘telephone first’ approach, which were associated with small net increases in secondary care costs. Although, again, there was wide variation between individual practices, we found no evidence to support the claims of widespread cost savings from the new approach.

An analysis of patient experience from the national GP Patient Survey compared ‘telephone first’ practices with control practices in England. The biggest difference by far was a 20-percentage-point improvement in intervention practices in responses to the GP Patient Survey question on length of time to see or speak to a doctor or nurse. This was equivalent to an average improvement of nearly one full category in the four-point scale used in this national survey (the question was ‘How long after initially contacting the surgery did you actually see or speak to them?’ and the answers were ‘on the same day’/‘next working day’/‘a few days later’/‘a week or more later’). Other differences were small in comparison, and included an increase in the convenience of the appointment (0.38 percentage points more positive), small but statistically significant reductions in overall experience of making an appointment (0.44 percentage points more negative), reported doctor–patient communication (0.89 percentage points more negative) and ability to see the patient’s preferred doctor (1.25 percentage points more negative). Patients were also 2.4 percentage points less likely to recommend their GP surgery to others after the intervention.

In the patient survey that we conducted of people who had recently had a telephone consultation in ‘telephone first’ practices, 21.6% reported that the telephone consultation was less convenient, 23% reported that there was no difference, and over half (55.4%) said that it was more convenient; however, we have evidence that the practices that agreed to take part in this survey had more positive experiences than those that declined to take part.

One factor that is likely to influence patients’ reported experiences is whether or not they were expecting to have to speak to a GP before making an appointment. Of patients surveyed, 70% of those who phoned expected to get a callback, but 25% did not. Those who were least likely to expect a call were those who visited the surgery to request an appointment.

As in the patient survey, patients expressed a wide range of views in the interviews we conducted. One of the most striking features of these interviews was the extremes of responses, as exemplified in the following patient quotations: ‘I have rang the GP’s surgery at 8.30 a.m. and redialled 152 times before getting through to surgery then waited till 5.30 p.m. for callback from a doctor’ and ‘This new system has proved truly excellent and my doctor has seen me on the same day if she has thought it necessary’. The ‘telephone first’ approach clearly suited some people by reducing the need for them to come into the surgery, but it was problematic for others (e.g. when it was difficult for someone working in an open-plan office to take a callback). However, a substantial proportion of negative comments were about the operation of the scheme itself rather than the principles behind it. So, for example, patients may not have minded being phoned back by a GP, but they complained about the difficulty of getting through on the telephone, restricted times when the telephone lines were open, lack of flexibility in scheduling the callback from the GP and lengthy delays before the callback was made.

Summary of qualitative findings

Patients also described the characteristics that made it easier for them to use the ‘telephone first’ approach. These included being articulate and having the confidence to press for the outcome they wanted, having flexibility in their daily schedule and having an existing trusting relationship with the GP who was deciding whether or not a face-to-face consultation was necessary. Some practices had clearly organised themselves so that it was easier for the patient to see a doctor of their choice; in others it was harder. Not having English as a first language was identified as a problem by patients but, contrary to our expectations and concerns expressed in the literature,30 older people expressed generally positive views about the new system. Indeed, none of the nine patients aged > 75 years who we interviewed said that they would prefer to return to the old system. Patients who were at work found it less easy to manage the new system (e.g. not knowing when the GP would call back). A few practices were able to give patients a specific time when the doctor would call, but others were unable to do this.

A wide range of opinions were also expressed in staff interviews. These included examples of practices in which the approach appeared to be working well and others where it was working less well. Within practices, individual staff members also expressed strong views about liking or disliking the new approach. For some GPs, the ‘telephone first’ approach had meant that they were able to go home earlier, whereas others felt overwhelmed by increased demand. For most GPs, the new scheme involved a major change in the way they worked. It also involved redistribution of work, especially in practices in which partners took the telephone calls but salaried or locum doctors saw a higher proportion of patients who came for a face-to-face consultation. Practices were generally noted to be quieter during the day, with fewer patients attending for face-to-face appointments.

Receptionists had a very important role in implementing the new system and most of those interviewed welcomed the increased responsibility that they were given and the increased ability to offer patients appointments (albeit not face-to-face appointments). Practice managers were key to the effective implementation of the ‘telephone first’ approach and described the importance of understanding patterns of demand before deciding how to introduce the new approach.

Concerns were sometimes expressed about the safety of telephone consulting. The doctors we interviewed did not see this as a major problem. Indeed, they thought that the new system was safer because patients were able to have contact with a doctor much more quickly (by telephone) – they could then be given a face-to-face appointment rapidly if needed. Patients also commented that the new system avoided delays in speaking to someone about a problem that could be serious.

Members of staff who were interviewed also described changes in the culture of practices, including greater cohesiveness, with staff being more supportive of each other. However, again, there was a wide range of views including descriptions of tension within the team, especially when one GP was opposed to the new system. For staff in practices that had unsuccessfully tried to implement the ‘telephone first’ approach, there could be considerable negative impacts on individual staff members and the practice as a whole.

There were parallels between the staff and patient interviews in that when the ‘telephone first’ approach was successful it often worked for both staff and patients. Yet when the system was strained, for example when there were not enough slots on the day for telephone calls or difficulties with patients getting through on the telephone, both staff and patients struggled with the approach. Overall, the majority of patients and staff interviewed would choose to keep the ‘telephone first’ approach over a ‘traditional’ appointment system.

Summary of the cost–consequences analysis

The cost–consequences analysis reported a median cost of £10,810 paid by practices to commercial providers for support with adopting the ‘telephone first’ approach, but other set-up costs for practices were generally small, and few made staffing changes when they introduced the new system. We found no evidence to support claims of major reductions in secondary care costs, although, as with other aspects of the study, there was wide variation between individual practices with, for example, practice patients making substantially more, or less, use of A&E after the ‘telephone first’ system was introduced.

Superposed epoch analysis

FIGURE 25.

Superposed epoch analysis showing the change in total number of appointments per 1000 patients per day relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 26.

Superposed epoch analysis showing the change in the number of face-to-face appointments per 1000 patients per day relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 27.

Superposed epoch analysis showing the change in the number of telephone appointments per 1000 patients per day relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Regression analysis

Superposed epoch analysis

FIGURE 28.

Superposed epoch analysis showing the change in the mean time between booking and having an appointment relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Regression analysis

Superposed epoch analysis

FIGURE 29.

Superposed epoch analysis showing the change in appointment duration relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Regression analysis

Superposed epoch analysis

FIGURE 30.

Superposed epoch analysis showing the change in total time spent consulting relative to the intervention launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Regression analysis

Superposed epoch analysis

FIGURE 31.

Superposed epoch analysis showing the continuity index relative to the months to launch: practices with status ‘running’. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Regression analysis

Before-and-after analysis of GP Patient Survey scores (intervention practices only)

FIGURE 35.

Superposed epoch analysis of GP communication composite mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 36.

Superposed epoch analysis of ease of getting through on the telephone mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 37.

Superposed epoch analysis of recommending GP surgery mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 38.

Superposed epoch analysis of ability to see preferred GP mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 39.

Superposed epoch analysis of time until seen or spoken to mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 40.

Superposed epoch analysis of percentage of people seen or spoken to on the same day relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 41.

Superposed epoch analysis of GP convenience composite mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

FIGURE 42.

Superposed epoch analysis of overall experience of making an appointment mean score relative to the intervention launch. The black lines represent the mean within a single practice relative to the launch time, with each black line representing a single intervention practice. The green dots represent the mean of individual practice means.

Comparison with other practices in England (controlled regression analysis)