Setting up a rapid diagnostic clinic for patients with vague symptoms of cancer: a mixed method process evaluation study

Two focus groups were held with patients and their supporters (usually close family members) where all described their experiences of the RDC in depth. This included the background to their condition, and their care and support journey from initial treatment of symptoms in primary care through to RDC referral. The patients were approached by RDC managers to take part if they so wished. Each focus group was facilitated by the two authors with a Cwm Taf staff member also in attendance and lasted about 90 min.

As GPs are the sole point of referral to RDC, their use and experience of it as an effective clinical service is crucial to its success and it was essential to obtain their views on this. We approached this in two ways:

The online survey comprised multiple-choice questions taking about 10 min to complete and with options to comment further for each - and many did so. Stratification of the interview sample reflected a range of practice sizes, location, and frequency of RDC referrals.

Each face-to-face interview was conducted by both or one of the two male co-authors with experience in conducting qualitative interviews and lasted between 30 and 45 min. Participants were approached via telephone and/or email and the interviews took place in their practices (apart from one that took place in a public space). Detailed notes were taken during the interviews, which were not audio or video recorded, and no one else was present besides the participants and the researchers. This provided an informal atmosphere for discussion without unduly compromising the quality of the information obtained [14]. Interview notes were sent to the interviewees via email to solicit their comments and feedback. Prior to the interviews, neither of the authors was known to the participants and therefore they had no pre-existing relationship with them. Participants were informed about the study both verbally and in writing using a participant information sheet. No characteristics about either of the interviewers were reported. Both the questionnaire and interview guide, which was developed for this study and was pilot tested in the first interview, can be found in the supplementary file. A hybrid approach to thematic analysis was used, whereby themes were identified directly from the data (inductively), and by considering the aims of evaluation (deductively) [15]. Data analysis was carried out by both researchers.

We used several anonymised datasets provided by the Cwm Taf UHB information team. Specifically:

Significant effort was required in data preparation prior to analysis which included: cross-referencing records across the different datasets; accounting for duplicate records (some represented multiple RDC attendances by the same patient); accounting for mis-typed hospital numbers (crucial to linking data items across different activity data sets); referral and treatment dates not correctly connected with appointment dates. We also supported local codification of free-text data fields in the local RDC dataset to enable quantitative analysis.

The linked evaluation dataset was used to conduct descriptive statistical analysis (mean, standard deviation, proportion, median, and interquartile range) and to calculate the following metrics: preliminary cancer diagnosis rate (the number of patient with a cancer diagnosis over the number of patients attending the RDC; sometimes referred to as the cancer conversion rate); the median times and interquartile range (IQR) from GP referral RDC appointment and from GP referral to cancer diagnosis.

Given the documented challenges with conducting cost effectiveness analysis in this setting [16] and as the RDC was already in operation when our study commenced, we adopted a pragmatic design focusing on direct RDC fixed staff costs averaged over first attendances, and marginal treatment costs related to the number of patient first attendances. As we were limited to routinely available data, these figures are illustrative rather than definitive as they do not, for example, include management or estate overhead costs.

We organised and facilitated two patient focus groups attended by seven participants in total; patients (n = 4) accompanied by partners or relatives (n = 3). Both groups were ‘very satisfied’ with their experiences of the RDC process and the sensitive manner with which potentially very difficult results and information were presented. There was sufficient time for clear explanations about the implications of the diagnosis, and direct involvement of patient supporters/ carers throughout. The speed of the referral process was particularly highly valued as it lessened ‘worry time’ for patients and carers. All said they had no hesitation in recommending the RDC to any friends or family with similar symptomatic conditions to proactively seek a referral to the RDC service (and one had actively done so). Specific suggestions for further improvements were offered including:

We undertook two forms of GP engagement: direct interviews with eight GPs on their experience of referring patients to the RDC, and an online survey sent to all GP practices which was completed by 34 /165 GPs (21%), with 11 out of 34 respondents (32%) also providing valuable supplementary commentary. Practices from all UHB localities responded to the survey.

Online survey results indicated that GPs were overwhelmingly positive about the RDC with 88% of respondents either satisfied or very satisfied, and often coupled with strongly supportive additional comments (‘I think this clinic has made a big difference to my practice’).

Aspects of the RDC service (such as clarity of referral criteria; ease of the referral process; speed with which patients were seen; quality of information fed back to GPs) were also highly rated (typically 80–90% of respondents were ‘satisfied’ or ‘very satisfied’). However, certain aspects were less positively viewed: 18% of respondents reporting that there were ‘very often’ problems or significant delays in ordering pre-RDC attendance tests (difficulties in the specimen collection service from practices were mentioned by several respondents, for example, a practice did not have access to a daily transportation service of samples to a laboratory).

An interesting finding concerned levels of ‘RDC awareness’ within primary care. Among GP principals this was ‘high’ or ‘very high’ (94%), dropping to 59% for trainees, and only 10% for locums.

Results from the GP interviews also demonstrated overwhelming support for the RDC and the value it provided for clinicians and patients. Benefits frequently mentioned included:

The last point was particularly notable. GPs were not only concerned about delays in diagnosis and back-and-forth referral and testing for their patients, but also at the ‘loss of professional face’ they sometimes previously experienced when they had referred a patient with vague symptoms only to have them rejected as an inappropriate referral to that particular specialty. With the RDC that potential stigma is removed.

Rapid access to diagnostic testing was seen as particularly helpful. However, issues were mentioned over having pre-appointment tests undertaken in time (by the GP), and the report-back process also came in for some negative comment. However, we understand anecdotally that this can be due to the way in which reports are managed within practices themselves rather than the speed of reporting back to them from the RDC itself.

Specific recommendations about the RDC made by GPs included:

Between July 2017 and 31st March 2019, 681 referrals resulted in 574 (84%) attendances (the introduction of the RDC was staggered across four localities; starting with Cynon in July 2017 and finishing with Taf Ely in February 2018), Table 2. There were 4 hospital cancellations, 6 patient cancellations, 3 patients who did not attend their appointment, and 94 referrals that were either not required to attend (as deemed by the senior RDC clinician) or had been referred, but were yet to receive, an RDC appointment.

Of those attending an RDC appointment, the average age was 68 (68.3 for women; 67.6 for men); 58% were female; and approximately 30% had three or more vague symptoms, Table 3. The mean number of symptoms recorded by the referring GP dropped from 2.83 at the start of the operation of the RDC, to 1.53 symptoms in Quarter 1, 2019. ‘Weight loss’ was by far the most common symptom and there was a strong association between that and ‘lack of appetite’ with 100 patients presenting with both symptoms.

The local RDC database contains outcomes following the patient visit, Table 4. Of those, 59 patients (10.3%) were referred onwards to a specialist cancer pathway (directly to the Multi-Disciplinary Team meeting stage), and an additional 74 patients (12.9%) had a suspected cancer with further investigations deemed necessary. Almost a quarter of patients (128, 22.3%) were given a new, non-cancer diagnosis and referred to specialists in secondary care for further review and treatment. A further 78 patients (13.6%) received a new, non-cancer diagnosis but were referred back to their GP for evaluation and treatment. The remainder received no new diagnosis and were either referred to secondary care (80, 13.9%), or back to their GP (155, 27%).

We also analysed an extract of routinely collected hospital data for cancer patients which is compiled at an all-Wales level (CaNISC; see routinely collected datasets in the Methods section and glossary). As there is a time lag before patient records appear in this database, we did not expect to obtain all confirmed cancer diagnoses following an RDC appointment, but we were able to identify 42 patients with a confirmed cancer diagnosis following their RDC clinic attendance.

Based on the above data, two different preliminary cancer diagnosis rates were calculated: 7.3% (42/574) and 12.3% (71/574), which represent the likely lower and upper bounds of the RDC ‘cancer diagnosis’ rate.

Using the CaNISC data for the 42 confirmed cases, we identified the tumour site for these patients as well as the treatment they received (Table 5). Gastro-oesophageal cancers were the largest group (21%), followed by urological and lung cancers (both 17%), and colorectal and haematological cancers (12% each).

A key RDC objective is to identify non-cancer diagnoses (B1 and B2) as well as cancers in patients referred with vague symptoms. Overall, 73% of patients seen received either a diagnosis (cancer or non-cancer) and/or an onward referral to secondary care, (i.e. all outcomes except C2).

A record of subsequent non-cancer diagnoses is not maintained, but we undertook a classification of the type of condition based on a text field analysis of the diagnosis and specialty of onward referral where known (Fig. 1). Respiratory (n = 46 onward referrals) and GI-related referrals (n = 42) were the most common (whether to secondary care or GP). Several classified as ‘mental health’ (n = 6) relate to drug and alcohol addiction; these were all referred back to the GP. Conversely, cardiac referrals (n = 11) are nearly all to secondary care. Endocrinology (n = 12) includes several patients where their diabetes appeared to be out of control. Several referrals appear to be for (potentially) two different types of diagnosis.

Finally, median time to RDC appointment from GP referral was 12 days (IQR 7, 19) and median time from GP referral to cancer diagnosis was 34 days (IQR 13, 50).

There are two main cost dimensions: fixed (e.g. staff) and marginal (e.g. scans, tests). Fixed costs do not vary in the short to medium term and their unit cost thus decreases as increasing numbers of patients are seen within the available capacity (15 patients per week over two sessions on a Tuesday afternoon and a Thursday morning). Additional capacity would require a step-change increase in fixed costs for more clinic sessions and staff time. Marginal unit costs, however, increase directly with the number of patients seen with every extra patient generating additional expenditure.

Patients attend the clinic for 2–3 h altogether, at least 1–2 h of which is time waiting between the initial consultation and scan for the results and follow-up final consultation and diagnosis. From relevant 2018/19 cost outturn data we derived a staff cost per first attendance of each patient (£321). This includes all direct clinical and non-clinical staff time ascribed to the RDC (which includes consultant radiology and RDC clinicians). We were also provided with marginal cost data associated with undertaking blood tests and radiology examinations in the hospital (£185 per patient first attendance). All patients undergo the same suite of investigations irrespective of clinical presentation. However, we were not able to obtain any RDC clinic overhead costs, so our estimated total cost per first patient attendance – £506 – is an under-estimate.

In this paper, we report on a process evaluation study that explored the initial impact of a new RDC referral pathway on patient and staff experience and on process outcomes using a mixed-method, pragmatic study design. More and earlier stage cancer detection and improved long-term survival rates are key objectives of the RDC model. Although these objectives could not be evaluated per se, our study found that the RDC pathway enables a good patient experience in cancer diagnosis. Patients are seen quickly and the pathway is highly regarded by them, their supporters, and referring GPs.

Using the seven ‘service implementation’ outcomes of feasibility, fidelity, penetration, acceptability, sustainability, uptake and costs described by Proctor et al. [17], the RDC implementation studied here has demonstrated to be a feasible service having been successfully rolled-out across four localities in seven months to include all GP practices. Available clinic slots are all routinely occupied; and we found high awareness and support among GPs about the service. Its take-up is evenly spread across the four UHB localities. Service referral protocols are followed and, where necessary, have been adapted following constructive engagement with primary care. Patients, too, have been supportive in their experience of the clinic as the focus groups showed. Regarding costs, although we were only able to undertake a limited analysis of RDC-specific costs, these were acknowledged to be commensurate with existing local clinic costs and were comparable with findings from the Swansea pilot study [7]. Unfortunately, there were no cost data available to compare with similar RDC-type clinics in England.

A number of similar initiatives have been launched in the UK, Table 6. In London, Guy’s and St Thomas’s NHS Trust (GSTT) was an early adopter of the RDC in the UK. Together with Oxford, this is a Cwm Taf comparator site (discussed below). In England, Cancer Research UK and Macmillan established the Accelerate, Co-ordinate, Evaluate (ACE) Programme of 50 sites trialling an RDC approach focused on cancer detection and diagnosis [18]. A second phase – ACE 2 (or ACE-MDC) – ran in five sites (including Oxford) which included non-cancer diagnoses [19].

Initial findings from ACE-MDC [20] indicate similar achievements to Cwm Taf, and similar difficulties in obtaining cancer staging data (noting many cancers only first presenting at an advanced stage, and it is too early to assess if there has been any significant increase in earlier stage diagnoses). In Oxford, for example, only 51% of diagnoses had cancer staging data available, and 61% of patients first presented at a late stage.

Parallels with Cwm Taf are also seen with the Rapid Access Diagnostic Clinic (RADC) clinic at GSTT. This clinic has a larger throughput (30–40 per week), and also accepts direct A&E and GI specialty referrals as well as from GPs. They report (unpublished internal report) 51% of patients had attended their GP more than three times before referral; 48% of patients reported experiencing symptoms for 2–6 months and 38% more than 6 months.

Patient satisfaction was high both in Oxford (70%) and GSTT (91%) as was that of GPs (Oxford 90%; GSTT 84%).

An important caveat is that although the metrics presented are similarly described, there was insufficient information to ascertain exactly how data were recorded (e.g. definition of eligible RDC population), or how some measures were calculated locally (e.g. recording time period).

In comparison to site-specific cancer pathways, our estimate of cancer diagnosis rate is in line with estimates in national studies that reported median rates between 8 and 17.0% across a number of urgent suspected cancer pathways.

The other Wales RDC pilot project at Abertawe Bro Morgannwg UHB began a few months earlier than Cwm Taf and was evaluated after a similar length of time of operation. A very different evaluation approach was adopted here using discrete-event simulation to model a cohort of 1000 patients from referral to radiological diagnosis based on routine RDC and hospital data [7]. These were compared with a control cohort of patients who had been referred to an Urgent Suspected Cancer pathway but subsequently downgraded. Their conclusion was that the RDC for patients presenting with vague or non-specific symptoms suspicious of cancer in primary care reduces the time to diagnosis and provides excellent value for money if run at ≥80% capacity with at least five patients per clinic (£646 per patient). There are currently no comparable cost data available from the ACE-MDC sites in England.

There are several limitations associated with this study. As our data collection and analysis was situated in a single organisation, we do not know whether our findings and conclusions can be generalised in different settings. However, given that many of our findings are in line with studies using different methodologies in different settings, we are confident that the lessons learnt here have a wider applicability. The lack of a control or comparator group of patients limits the conclusions that can be drawn from our study which is a similar issue faced by all other studies that have evaluated comparable new clinic designs. The nature of the study design also limits our capacity to analyse the outcomes of those patients who were diagnosed with cancer, as does the incomplete data on cancer staging (and which we thus we did not report on). Although we made an effort to interview GPs from practices with a range of referral activity and locations (and solicited the opinion of all GPs through the survey), participant bias may be present in the sample of GPs who responded and in the patients who participated in the two focus groups.

In line with previous studies, it was a challenge to identify and use a control group for comparison purposes. Future research and evaluation methods can contribute to the evidence base by overcoming this challenge, for example by using a ‘difference-in-differences’ study design [21]. Long-term patient outcomes are unknown and future studies should make an effort to address including long-term cancer survival, and types and outcomes of patients receiving non-cancer diagnoses. Further research is also needed to estimate the costs and benefit implications of extending the RDC model to additional cancer diagnostic pathways.

A number of practical recommendations for the RDC service were made during the course of the study arising from observations by the research team as well as from recommendations made by GPs (interviews and survey) and patients (focus groups). These can be grouped under three sub-headings:

Suggestions from patients and supporters

Suggestions from GPs

Suggestions from the evaluation team