Does Safety Netting for Lung Cancer Symptoms Help Patients to Reconsult Appropriately? A Qualitative Study

Safety netting in primary care is considered an important intervention for managing diagnostic uncertainty. This is the rst study to examine how patients understand and interpret safety netting advice around low-risk potential lung cancer symptoms, and how this affects reconsultation behaviours. Methods Qualitative dyadic interview study in UK primary care. Pre-covid-19, ve patients were interviewed face-to-face twice (shortly after a primary care consultation for potential lung cancer symptom(s) and 2–5 months later). The general practitioner (GP) they last saw was interviewed face-to-face once. During the covid-19 pandemic, an additional 15 patients were interviewed once via telephone. Audio-recorded interviews were transcribed verbatim and analysed using a mix of inductive and deductive thematic analysis.


Results
The ndings from our thematic analysis suggest that patients prefer active safety, as part of thorough and logical diagnostic uncertainty management. Passive safety netting may be perceived as dismissive and cause delayed reconsultation. GP safety netting strategies are not always understood, potentially causing patient worry and dissatisfaction. Telephone consultations and the diagnostic overshadowing of COVID-19 on respiratory symptoms impacted GPs' safety netting strategies and patients' appetite for active follow up measures.

Conclusions
Safety netting guidelines do not yet offer solutions that have been proven to promote symptom vigilance and timely reconsultation for low-risk lung cancer symptoms. Patients prefer active safety netting coupled with thorough consultation techniques and a comprehensible diagnostic strategy, and may respond adversely to passive safety netting advice.

Background
Safety netting is considered an important intervention for managing low-risk symptoms of cancer (1,2), particularly in the UK where it forms part of the government guideline for suspected cancer management (3). Safety netting refers to actions taken and advice given to patients by healthcare practitioners about how to monitor and re-seek help for new, recurrent, persistent, or worsening symptoms, which may bene t patients in terms of disease stage at diagnosis, treatment options and survival (4). Safety netting is a high-volume activity, with GPs reporting in a qualitative study that they use some form of safety netting at the end of almost every consultation (5). Indeed, an audit of patient records showed that safety netting is recorded in 44% of all patient contacts where a cancer is eventually diagnosed(6). Safety netting is particularly important in the diagnostic management of lung cancer, where low predictive value symptoms such as cough and tiredness are the most common rst complaints that patients present to primary care with (7,8). Safety netting is transactional (5), and is not effective if the patient does not hear or understand the advice they are given by the primary care healthcare professional (HCP) or if the advice is insu ciently speci c(9, 10). Heyhoe et al. (11) suggest that, as part of effective safety netting, HCPs and patients should work together to develop and agree strategies that encourage sharing of symptom monitoring, re-appraisal and feedback, which will aid diagnosis of cancer at an early stage (12).
Poorly communicated safety netting advice may be worse than none; patients may delay reconsultation for lung symptoms by several months if they perceive that symptoms have been initially attributed to a benign cause, for example chronic obstructive pulmonary disease (13)(14)(15)(16).
Best practice guidelines for safety netting have been developed through consensus (17) and evidence synthesis(10), although HCPs may deviate from this in practice (1,18). There is some evidence to suggest that patients prefer clear directives, for example setting a speci c timescale for reconsultation rather than an open-ended invitation (19); however, there is no evidence about whether patients understand or attend to safety netting advice, nor how this affects reconsultation behaviours.
This study used a qualitative dyadic (paired) design to understand how patients responded to safety netting in primary care for low predictive value symptoms related to lung cancer. We aimed to capture the effect of safety netting on the way patients judge their symptoms and (re)seek help in primary care over time.

Recruitment
Three GPs and 20 patients (Table 1) were recruited purposively to achieve variation in age, gender, geographic spread and ethnic background. The rst ve patient participants were recruited in general practices in deprived areas (lowest 30% SES) by local National Institute for Health Research Clinical Research Network research nurses. Patients meeting inclusion criteria were given a study recruitment pack and encouraged to contact the researcher if interested in participating. Once patients were recruited, the GP they saw most recently was also approached for interview. Participants gave written consent before interview.
Due to COVID-19, the recruitment strategy was amended in March 2020. A specialist recruitment agency, Taylor McKenzie, recruited 15 participants using a database of potential research participants and by approaching patient support groups. A screening questionnaire ensured participants had recently presented to their GP with a low predictive value symptom related to lung cancer and were also from the lowest 30% SES. Verbal informed consent was audio recorded.

Interviews
The rst ve participants were interviewed face-to-face twice in a private room at the practice, once shortly after the consultation and again 2-5 months later. Their GPs were interviewed once face-to-face in their consultation room. The remaining 15 patient participants were interviewed once via telephone.
Before starting the interview, the interviewer made sure that the participant had recently presented to their GP with a potential lung cancer symptom. Flexible interview topic guides (appendix 1) were developed using published literature and feedback from patient representatives. All interviews were conducted by an experienced, female, qualitative social science researcher. Interviews were digitally recorded and transcribed verbatim.

Analysis
Transcripts were imported into NVivo 11, qualitative data analysis software, and coded by two authors using an inductive approach. Author 1 and 2 (initials removed to allow blind review) initially read all transcripts, and author 1 coded all transcripts with a particular focus on GP-patient dyads. To ensure the coding represented the data it was discussed with author 2. After analysis of the ve dyads, the remaining dataset of 15 single patient transcripts was coded. Initial codes were then grouped into potential themes, and a thematic map was produced. The themes were then discussed and nalised by author 1 and 2, who agreed that the nal themes were supported by the data. The nal analysis and a draft of this manuscript were checked by all co-authors.

Theme 1: Patients prefer active safety netting strategies
Patients preferred safety netting that included advice and actions that actively promoted re-consultation or involved pro-active follow-up.
[I appreciated the] information in the initial conversation with the receptionist, and the prompt contact from the doctor. And the advice, and him wanting to know, in a few days how I was feeling, was anything getting any worse, any more symptoms. (Patient 20, female, 57, telephone interview).
Active safety netting as part of a thorough and logical approach to managing their lung cancer-relevant symptoms allowed patients to understand the diagnostic strategy, in turn making the safety netting advice easier to understand. For example, in the example presented in Box 1 both the patient and GP describe the 'logical' steps the GP took to reach the diagnosis over several visits. In his rst interview the patient described how the GP had addressed each of his concerns through a series of logical steps, in his second interview he re ected on how this approach had been reassuring. Several patients commented that their telephone consultation was with a GP they had not met before, who had not read their notes and was unaware of their history. They were dissatis ed with the level of thoroughness in the consultation, were not sure that all possible diagnoses had been considered, and felt 'unsafe' as a result.
If I get on the phone and talk to a GP, and they work through my symptomology, I discuss some of my background, what's been happening with me, get some historical data, and to then come to a conclusion in terms of a prognosis. Had that been done, in that kind of environment, I suppose, I would have thought "You know what, they've taken the time to research what's going on with, happen with me historically".
(Patient 14, male, 50, telephone interview) Patients who had attended a telephone consultation during the COVID-19 pandemic, where the focus was perhaps mostly on acute symptoms and/or ruling out COVID-19, found it di cult to remember everything they wanted to discuss with the GP without a thorough discussion of wider issues.
GPs reported that they used active safety netting with certain patients, but not others, particularly when their "level of worry is a bit higher" (GP 3). Two GPs were concerned that active safety netting could worry patients unnecessarily.
A lot of what we do in general practice is actually reassure the worried well and a lot of discussions about cancer safety netting is actually just doing the exact opposite. (GP 1, male). GP 3 believed that explicitly naming red ag symptoms could raise anxiety among patients at very low risk of cancer.
Once you explain the list of things to look out for that people start getting them more often. This suggests that GPs have particular concerns about speci c safety netting devices, including highlighting red ag symptoms, for fear of generating anxiety, although this was not reported in our patient interviews. Patients engaged with diagnostic uncertainty and management as part of a thorough and attentive consultation, especially when active steps were taken to promote re-consultation. Some patients experienced feelings of shame following passive safety netting, and felt dismissed particularly when access to a blood test was restricted. These patients did not receive speci c advice on how to deal with COVID-related disruptions to additional tests they may need.
And so, there is nothing they suggest 'Oh, call me back at such and such at a later time and we'll check it.
Obviously, the problems is right now I can't go and get a blood test. So, I guess that's made it problematic. Or a follow-up appointment.
[…] Yeah, it did feel rushed. This is what you're getting, and see you later. (Patient 10, female, 40, telephone interview) GPs justi ed passive safety netting in consideration/management of cancer risk, particularly in face to face consultations where they felt more con dent about their appraisal of the patient's likelihood of cancer.
Theme 3: Patients don't always know about diagnostic strategy, either because they don't understand it or where GP hasn't told them everything GPs described diagnostic strategies predominantly in relation to managing cancer risk, including some strategies not communicated to patients, such as heuristics about patient's characteristics (e.g. age), and future plans in the event that symptoms persisted. When patients were unaware or had not understood the GPs 'logic' or diagnostic strategy, there was often a feeling of concern or lack of resolution. Box 2 presents an example of misalignment between patient and GP, where the patient was unaware that the GP was ruling out anaemia [to exclude cancer] and did not feel reassured by blood tests and an x-ray as a diagnostic management strategy. She was, however, reassured following an MRI scan, which she felt was the only way to properly rule out cancer. I did, so when I requested the bloods, I requested a bone pro le as well and full blood count to make sure there was no anaemia. So yes, I did. But I didn't feel, based on her history, that was likely to be cancer. I did the tests to make sure we weren't missing it, but because she was otherwise well it didn't t with myeloma or anything like that. I still did all the investigations.
[…] Yes, I think the fact that she was running a business and she looks after the grandchildren and she's standing a lot all made me think well this is most likely getting to be mechanical back pain rather than something more sinister. So, I guess from the history what I was gathering was pointing me toward benign Similarly, Patient 2 was not aware of the diagnostic management strategy that GP 1 was using to resolve her cough (Box 3). She was concerned that the GP did not appear aware of her medical history of whooping cough and silent re ux, and was not aware of the GP's strategy in using a 'trial-of'-treatment' approach (silent re ux) as well as waiting for potential parapertussis symptoms to resolve. The GP, on the other hand, thought that he had clari ed this, and that they were working to an 8-week timeframe for reconsultation. In her second interview the patient re ected on her experience and explained that she felt that the GP had most likely reached the correct diagnosis, but that the diagnostic approach and next steps had been unclear to her throughout. These ndings indicate that safety netting is dependent on patient understanding of the diagnostic process in addition to comprehension of speci c signs, symptoms and timeframes that should trigger later actions.

Discussion
Summary This is the rst study to our knowledge to examine how patients understand and interpret safety netting advice around potential low-risk lung cancer symptoms. Our results suggest that patients strongly prefer active safety netting, as part of a thorough and logical approach to managing diagnostic uncertainty.
Conversely, passive safety netting may be perceived as dismissive and cause patients to delay reconsultation. In contrast, the GPs in our study worried that active strategies, particularly mentioning red ag symptoms, may cause unnecessary concern. Our analysis also suggests that GPs do not always make their safety netting strategy understood, potentially resulting in a misalignment where the GP thinks they have made an active safety netting plan while the patient feels worried or dissatis ed by what they perceive to be a passive safety netting approach. Telephone consultations and the diagnostic overshadowing of COVID-19 on respiratory symptoms were likely to affect GPs' safety netting strategies and patients' appetite for active follow up measures.

Strengths and limitations
The strength of this study is the focus on patient interpretation of real safety netting experiences, and pairing these with their GPs' aims and perspectives. We interviewed patients from a wide range of geographic areas, giving us a varied picture of patients at risk of lung cancer. Recruitment of GP-patient dyads was limited by the pandemic. The 15 patient participants recruited during the pandemic described signi cantly different, COVID-related, experiences from patients interviewed pre-pandemic. As much of the focus was on COVID-19, it was challenging to disentangle which actions were 'routine' safety netting and which were pandemic-related. However, given the likely persistence of remote consultation in primary care, our ndings will have relevance as new guidance and local practices emerge.

Comparison with existing literature
Our study builds on previous suggestions from a study of hypothetical safety netting preferences, which reported that patients need active reassurance around reconsultation (11). Furthermore, our study demonstrates that patients are more likely to feel a sense of subjective 'safety' when safety netting is part of a robust and logical consultation, and re-accessing care is assured. This is in line with previous studies reporting patient perceptions of under-support and over-reassurance following all-clear diagnoses(16), and studies showing that patients' perceptions of safety are associated with holistic and individualized care, and challenged by system barriers to healthcare access (20).
Our study extends recent research looking at the mismatch between safety netting in guidance and practice(1), by collecting empirical data about real consultations rather than hypothetical or consensusdriven designs. Our ndings mirror previous interview studies with GPs highlighting variability and uncertainty in safety netting approaches, and worry about managing cancer risk amidst busy workloads(5, 18).
Implications for research and/or practice Future safety netting research should measure patient understanding and reconsultation behaviour, developing strategies that improve these outcomes without raising unnecessary anxiety. Future studies should conceptualise safety netting as a complex intervention for patient safety and diagnostic management, with the aim to achieve alignment between patient and GP about the presenting problem's signi cance and next steps. Taking a Health Literacy Universal Precautions approach (21,22), will improve the likelihood that patients understand advice and create an aligned diagnostic strategy (23).
Practices should improve clinical safety netting practice by measuring processes such as delays to reconsultation, missed tests, unful lled prescriptions and other measures of concordance. Practices should move away from measuring safety netting quality in terms of missed or delayed cancer diagnosis, which is a relative rarity and a poor 'calibration' measure (24).

Conclusion
Diagnostic management of patients with low-risk lung cancer symptoms in primary care is a crucial mediator in promoting early diagnosis. Safety netting guidelines do not yet offer solutions that have been proven to promote symptom vigilance and timely reconsultation. Patients prefer active safety netting coupled with thorough consultation techniques and a comprehensible diagnostic strategy, and may respond adversely to passive safety netting advice.

Declarations
Ethics approval and consent to participate The study was approved by the NRES Ethics Committee London Central (REC Ref: 18/LO/1550). We con rm that all participants declared their informed consent to participate in a written form and that all personal identi ers have been removed or disguised so that the persons described are not identi able and cannot be identi ed through the details of their stories.
All methods were carried out in accordance with relevant guidelines and regulations.

Consent for publication
Consent for publication was obtained from all participants.