General Practitioner perspectives and wellbeing during the COVID-19 Pandemic: a mixed method social media analysis

Background: General practitioners (GPs) adapted their work practices rapidly in response to the COVID-19 pandemic. Limited research has explored their perspectives over this time, and factors that may affect their wellbeing. Method: We conducted a social media analysis of NHS GPs practising in the UK during the COVID-19 pandemic to identify issues which may affect their wellbeing. To identify trends, we assessed 91,034 tweets from 185 GPs on Twitter who posted before and during the pandemic, (January 2019 to February 2021). To identify themes related to wellbeing, we analysed qualitatively 7145 tweets posted during the pandemic from 200 GPs. Results: We identified inter-connecting themes that affect GP wellbeing, predominately around resources and support. Lack of personal protective equipment (PPE) and testing led to discussion of safety and risk, as well as increased workload resulting from staff isolating. Expressions of low morale and feeling undervalued were widespread, resulting from the perceived lack of support from the government, media and the general public at a time of staff shortages and high workload. Trends in themes were apparent, with emphasis on PPE, testing and safety March to May 2020 and morale, abuse, 'closed' GP surgeries, testing, flu vaccines and overworked September to October 2020. From December 2020 the COVID-19 vaccine dominated posts. Conclusion: GPs' experiences and perceptions as reflected in their social media posts during the pandemic have changed over time; perceived lack of support and resources, and negative public perceptions have exacerbated their concerns about existing underlying pressures.

Before the COVID-19 pandemic, rising demands on UK NHS general practitioners (GPs) including increasing 3 complexity and intensity of work and difficulties in recruitment and retention led to reports of a service in 'crisis' [1] 4 and a 'fundamental threat to the sustainability of primary care in England' [2]. Forty per cent of NHS doctors 5 reported psychological and emotional conditions before the pandemic [3], with GPs in particular experiencing high 6 levels of burnout [4,5]. 7 8 COVID-19 brought about many adaptations to general practice, including a rapid transition to e-consultations, 9 creation of 'hot' and 'cold' clinics (those examining patients experiencing COVID-19 symptoms and those addressing 10 unrelated needs), managing the manifestations of the pandemic in their practice, managing risk, and responding to 11 rapidly evolving guidelines. Indeed, high levels of stress and burnout amongst GPs during the COVID19 pandemic 12 have been reported internationally, with an expanding literature on this topic (e.g. [6][7][8]. Two UK studies of GP 13 pandemic experiences show greater work stress due to rapid change and uncertainty [9] and the importance of 14 teams in creating a sense of solidarity [10]. These studies are relatively small - Trivedi [9] studied 111 GPs in 15 Leicestershire and Wanat et al [10] is a cross-European qualitative interview study of primary care staff, but includes 16 only seven GPs in England. Research studying the pandemic experiences of a national sample of UK GPs is needed. 17 18 Social media use by health professionals has become increasingly widespread [11,12] and this has expanded during 19 the pandemic; Twitter is one of the most common types of social media platform [13][14][15]. Increasing numbers of GPs 20 are using social media, and increasing frequency of posts provides a useful tool for ascertaining their personal views 21 and experiences. Doctors use Twitter to discuss, rapidly and informally, current issues pertinent to their work and to 22 communicate with colleagues [13]. Unsurprisingly, this has focused recently on issues related to COVID-19 and work 23 practices [16]. Analysis of Twitter posts is used commonly to identify public experiences and opinions, including 24 recently on COVID-19 and its impact [17][18][19][20][21][22][23][24][25][26][27][28][29][30][31].Medical professionals' opinions about COVID-19 have been studied 25 using social media [32], but no studies to date have examined GP wellbeing using social media monitoring, which is 26 the aim of this study. 27 28 Methods 29 30 Firstly, we sought to explore trends pre-and post-COVID using a longitudinal analysis of GPs' tweet content (January 31 2019 to February 2021). Secondly, a more qualitative analysis explored themes emerging from GP tweets during the 32 pandemic (February 2020 to February 2021) with a particular focus on GP wellbeing. This analysis was inductive in 33 nature, and as such we did not seek to confirm or refute an existing hypothesis, but rather to explore emerging 34 themes. This paper conforms to the Standards for Reporting Qualitative Research (SRQR). 35 36 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 Sampling and data 1 2 Twitter profiles of users who tweeted on COVID-19 from 10 March 2020 to October 2020 were shared by Professor 3 Mike Thelwall (see acknowledgments). We limited profiles to those with a self-reported UK location and 'Dr' or 4 'Doctor' in their username (5,512 users and 223 users, respectively) or biographical description (850 users and 3,885 5 users). We then selected GPs manually, excluding non-NHS GPs, retired GPs and practice or organisation accounts. 6 This identified 293 practising UK NHS GPs. We supplemented these by searching for 'NHS GP' as a phrase in 'people' 7 in the Twitter Advanced Search facility, identifying a further 88 Twitter users after removing duplicates. The resulting 8 sample included 381 UK NHS GPs. 9 10 To explore the representativeness of our Twitter sample, we collated available demographic data, such as gender 11 and race (categorised as black, white or Asian), geographical location, and type of GP (such as GP partner or GP 12 trainee is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 common methods for studying information obtained from social media [33] and is appropriate for identifying 1 prevalence [31]. We used an inductive approach as we were not testing an existing theory and had no prior 2 framework. 3 4 To answer our research questions fully, data immersion was essential. Following data familiarisation and immersion, 5 a coding framework and annotation guide was developed by SG, then discussed and refined through multiple 6 iterations with the study team. Tweets were hand-coded as this is still the gold standard form of analysis [34]. To 7 avoid over-interpretation of these short tweets, we coded only what was explicitly stated. We tested for consistency 8 in coding to increase the dependability of the findings by independently double coding 10% of tweets. Level of 9 agreement was high, with 1.2% (11/915) codes changed and three additional codes added to existing codes. 10 Remaining codes were checked by the second reviewer during the coding categorisation process, rather than 11 independently assigned. 12 13

14
Sample demographics 15 Our sample reflects the GP population in terms of broad ethnic group, but over-represents men (Table 1). GPs were 16 located throughout the UK, with a slight over-representation from London. The majority (81%) did not indicate what 17 GP role they held. Of those that did, most were GP Partners or GP Trainees/Registrars. Age was reported by less than 18 5% of GPs. 19  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The number of tweets increased dramatically a few days before the first UK lockdown (20 th March 2020) and stayed 4 high until the end of April 2020 ( Figure 1). The next largest peaks were around the time of the Royal College of 5 General Practitioners (RCGP) Annual Conference (24 th -25 th October 2019) and the US presidential election results 6 (November 2020). Other smaller peaks reflected polling day (12 th December 2019), a GP conference "DGPLondon20" 7 (29 th February 2020), announcement of the second lockdown (31 st October 2020), the first vaccine efficacy results 8 (12 th November 2020), and the COVID-19 vaccine roll-out (January 2021).   Handles (34,931 unique handles used 177,766 times) 2 The most common Twitter handles cited by GPs were organisations (such as @rcgp, @nhsengland, @thebma), 3 politicians (such as @matthancock, @borisjohnston) and fellow GPs. Handles in 2020/2021 were similar to 2019 4 with an increase in mentions of GPs labelled as 'renowned COVID-19 health experts' by Twitter (Supplementary 5  Table 3). Many of the words used were to thank the hard 8 work of colleagues (thank, time, great, work, staff, team, practice). Terms in 2021 reflected the COVID-19 vaccine 9 rollout. 10 Specified Themes (using words and hashtags) 11 During the first wave, tweets related to COVID-19 and interventions to reduce transmission (such as lockdown, social 12 distancing and PPE) were common, as were issues around safety, frontline staff and mortality (Supplementary Table  13 4 and Figure 2). Lack of testing tweets peaked in both the first and second wave. Commentary around remote 14 working increased during the pandemic. References to workload, being closed, flu, GP 'bashing' and low morale 15 peaked around September to October 2020. Issues related to coping, wellbeing, appreciation, sadness and 16 enjoyment appeared throughout the time period. January 2021 saw an increase in vaccine-related tweets. 17 Feb - is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Analysis 2: Qualitative exploration 4 Table 2 summarises the themes which emerged during the pandemic. The 12 most identified themes in descending 5 order of frequency are described. Some of the themes are interconnected. 6 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ; https://doi.org/10. 1101 Resources, lack of (PPE) 521 (7%) 107 (55%) 43%, 61% 52%, 64%, 54% Resources, lack of (testing) 289 (4%) 70 (36%) 34%, 35% 28%, 46%, 36% Resources, lack of (staff) 169 (2%) 64 (33%) 23%, 39% 29%, 40%, 18% Resources, lack of (funding, pay) 21%, 25%, 18% Self-care of GPs in reference to their wellbeing (n=201, 3%) Self-care (positive) 189 (3%) 65 (33%) 37%, 30% 30%, 37%, 36% Self-care (negative) 12 (0.2%) 9 (5%) 3%, 5% 5%, 3%, 9% *Some tweets discussed more than one topic and were coded in more than one category. 1 **Using test of proportions, comparing the % of male tweeting the topic and the % of female -topics of 'negative changes to 2 practice' (p=0.028), 'lack of resources' (p=0.005), 'lack of staff' (p=0.019), 'PPE' (p=0.013), 'adequate resources' (p<0.001) and 3 'positive direction/management' (p=0.019) were posted more heavily by males than females, whereas experience of 'emotions' 4 (p=0.003) were more by females than males. 5 ***Using Fisher's exact test significant differences were found in four categories about 'Risks to BAME GPs' (p<0.001),

9
Changes to practice 10 Most posts regarding changes to practice were neutral or negative. Neutral posts simply stated changes being made, 11 most commonly remote working (IT systems and software, working from home and triaging patients in their own 12 homes) with some referring to safe practices for face-to-face consultations (cleaning between patients, PPE, scrubs, 13 and 'hot hubs'). Most of the negative comments about changing work practices related to remote working. GPs 14 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ; https://doi.org/10.1101/2021.10.19.21265194 doi: medRxiv preprint expressed concern around missed diagnoses, widening health inequalities and increased time and fatigue associated 1 with remote consultations. GPs felt job satisfaction, personal care, emotional support and patient satisfaction were 2 all challenged by remote working. Problems contacting patients were common because of patient availability, phone 3 networks, internet providers, or IT systems. Working from home brought about additional challenges, especially for 4 those with children. 5 6 GPs emphasized the need to remain accessible to patients through face-to-face consultations, but safety measures 7 (e.g. PPE, cleaning) increased time pressures. Safely visiting care homes was a major challenge, with concerns around 8 transmission to vulnerable patients. Challenges around patient non-attendance for potential cancer, stroke and 9 heart attacks also caused anxiety for GPs, and they described an increasing number of patients with mental health 10 problems. 11 There was concern about GPs delivering the COVID vaccination programme; some expressed the view that this could 12 not be done without deprioritising other services, that reimbursement was low and that other health care 13 professionals may be more appropriate as GP staffing levels were already critical. 14 15 There were a small number of positive posts praising colleagues or software or stating the benefits of remote 16 consultations in terms of efficiency and patient care, alongside some speculation on whether some changes could be 17 permanent by taking the 'best of the changes into the future'. A few mentioned how retired GPs could be deployed 18 by remote working. Lack of Resources 26

COVID-19 Testing 27
GPs expressed anger about the lack of testing in the first phase of the pandemic, resulting in risk to patients and 28 unnecessary self-isolation creating added staffing pressures. There was frustration that GPs were perceived to be 29 lower priority than high-profile public figures, while 'we risk our lives'. Furthermore, when comparing themselves to 30 hospital staff, GPs were confused and angry as to why they were not a priority group, given they had more contact 31 with patients. 32 33 GPs reported that the testing system again caused problems in September 2020 with long waiting times for results, 34 or long distances to testing centres. GPs also complained about 'rationing' patient tests. Problems with the supply 35 and accuracy of lateral flow tests was emphasised in January 2021. 36 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Reported concerns related to a perceived shortage of doctors and nurses, with many referencing a decline in GP 3 numbers over recent years, expressing the view that primary care staffing levels were critical before the pandemic 4 and COVID-19 self-isolation exacerbated these problems. GPs reported attrition throughout the pandemic, due to 5 factors such as workload, underfunding, low morale and lack of appreciation. 6 Personal Protective Equipment (PPE) 7 Lack of PPE appeared to be a common problem leading some GPs to report purchasing their own, reusing 8 equipment, improvising, accepting donations or posting pleas on social media for supplies. GPs felt they were lower 9 priority than secondary care staff and 'even supermarket employees', and initial public panic buying meant low 10 supplies of hand sanitiser and cleaning products. 11 The quality of PPE was described as 'substandard' or even 'hopeless' with flimsy paper masks, thin plastic aprons and 12 masks four years out of date. Pleas were made for World Health Organization guidance on PPE to be followed 13 including FFP3 masks. Questions on the effectiveness of surgical masks increased as more emphasis was placed on 14 aerosol transmission. 15 There were concerns expressed about being silenced, with social media appeals taken down, Clinical Commissioning 16 Groups asking GPs not to speak out and media reports of 'whistleblowers threatened with job loss for speaking out 17 on PPE'. GPs expressed anger at the UK Health Secretary's comment that PPE should be treated as a 'precious 18 resource' as this appeared to blame staff for not using equipment carefully. GPs questioned how many lives were 19 lost because of inadequate PPE supplies. 20

Funding and Pay 21
Many GPs posted about funding cuts over the past 10 years. The frustration surrounding perceived underfunding of 22 primary care grew over time as increasing amounts were spent on privately provided interventions including 'NHS 23 Test and Trace', which many GPs felt would have been better managed by the NHS. 24 There was reported concern over what was viewed as a real terms pay cut, and frustration when comparisons were 25 made to MPs' salaries or management consultants funded by government. 26

Adequate Resources 27
These posts were far fewer in number (see Table 2) and tended to refer to having PPE or testing available. Some of 28 these stated that 'at last we have plenty of PPE', thus referring to a time when supplies were inadequate. 29

Direction and management 30
Approximately a tenth of posts relating to direction and management were positive, and these were mostly directed 31 at organisations such as the RCGP, King's Fund, or Public Health England with a few directed at government actions 32 such as the NHS workers' visa extension and abolishing NHS fees for overseas staff. 33 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ; https://doi.org/10.1101/2021.10. 19.21265194 doi: medRxiv preprint The vast majority of posts, though, were negative. A few GPs were negative about organisations such as the RCGP, 1 NHS England and the BMA, expressing a need for more support and action. Most criticism was focused at the 2 government, particularly in England. Issues were related to other themes in our analysis such as underfunding, 3 declining numbers of GPs, lack of PPE and testing, inconsistent or poor guidance for GPs and GPs being used as a 4 scapegoat. 5 Shielding lists were seen by some as poorly managed centrally, with GPs rectifying errors. The paperwork required 6 for returning GPs was criticised, and some thought the scheme was putting older people at risk. Poor management 7 of care homes during the pandemic and risks to staff and patients were also discussed; there were calls for the 8 government to be accountable for the deaths of NHS staff. 9 GPs felt overlooked by government with regard to testing and mandatory public face coverings. There was a general 10 sentiment that the focus of government and media was on hospital patients and staff. Anger was expressed about 11 the behaviour of political figures who 'break the rules'. 12 Some GPs did not feel properly supported to carry out the COVID-19 vaccination programme and held concerns 13 around supply chain issues and fielding patient enquiries. 14

Misinformation about or received by GPs 15
Some GPs complained of confusing information and guidance received. For example, in March 2020, concerns were 16 raised about the mask policy in primary care, when GPs should isolate, suspension of routine work, and whether to 17 move to remote consultations. As the pandemic progressed other issues were raised such as the 'shambolic' 18 correcting of shielding, advice from NHS 111 and government for the public to contact their GP inappropriately, and 19 misinformation on dexamethasone use. 20 Misinformation about GP surgeries being 'closed' persisted through the pandemic, with peak frustration expressed 21 by our sample of GPs in April and September 2020. The government was seen as perpetuating this misinformation. 22 Many GPs were particularly affronted by a letter from NHS England in September 2020 to remind GP practices of 23 their duty to provide face-to-face appointments, warning that failure to do so could constitute a breach of contract. 24 GPs declared that they were not 'tucked away safe' or 'twiddling our thumbs' but 'working harder than ever'. Pleas 25 to the public and reassurances that practices were open continued into 2021. 26

Information use and sharing among GPs 27
Some GPs shared advice on working practices, whilst others asked questions of their colleagues. Sharing petitions on 28 issues such as testing or PPE and work surveys was commonplace, as was sharing factual information on issues such 29 as GP deaths, risks for BAME GPs, and doctors with long COVID. 30 Information to support GP wellbeing 31 These posts often consisted of links to webinars, events or resources to support GP wellbeing. Other posts simply 32 suggested ways to help such as 'being kind to yourself', 'taking breaks', 'taking down time' and 'keeping active'. 33 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 (Lack of) appreciation of GPs 1 Many GPs expressed feelings of being unappreciated by media, government and the public. In the first lockdown 2 there were a few reports of stealing from GP practices such as toilet rolls and hand sanitizer as well as vandalism, 3 graffiti and abuse from patients. Public perceptions were viewed as being exacerbated by GPs being accused of being 4 'closed', needing to 'reopen' and the insinuation that GPs are 'sitting around doing nothing', 'lazy', 'selfish' and not 5 doing the job they 'signed up to'. GPs expressed frustration about this and felt they experienced a lack of respect 6 and appreciation. There were references to 'GP bashing' and GPs being blamed for the 'failures of government'. GPs 7 complained of feeling like 'public enemy number one', particularly from September 2020. 8 The 'clap for carers' was met with a positive emotional response by some GPs. Some said that 'it bought a lump to 9 my throat' or 'a tear to my eye'. As time went on, however, expressed views of this became less positive and by 10 January 2021, discussion around a return of 'clap for carers' was met with calls for the public instead to observe the 11 rules to protect the NHS. 12 There were fewer posts relating to positive appreciation of GPs and these tended to reference appreciation shown in 13 the form of gifts from businesses or the local community, allocated shopping times, donations of PPE, and clapping 14 in the first lockdown. GPs recognised patient gratitude, particularly during the vaccination programme. 15

Appreciation of others 16
GPs expressed gratitude to those supporting them in the pandemic, including organisations, volunteers, the public, 17 neighbours, and local businesses. There was also admiration for other professions such as scientists and teachers. 18

NHS work colleagues 19
There were expressions of gratitude to NHS staff, particularly primary care colleagues, secondary and community 20 care staff (particularly care homes), domestic staff and administrative teams in hospitals. There were comments 21 about altruism and dedication with staff 'going above and beyond the call of duty', 'showing courage' and being 22 'amazing', 'world class' and 'heroes'. There were also views expressed about how 'fantastic' the NHS itself is by 23 providing free care, rapid adaptations to change and an incredible response to the crisis. 24 There was, however, some criticism of other parts of the NHS treating GPs 'like commodities instead of human 25 beings', 'bullying', 'too much bureaucracy', and a management team 'devoid of reality'. 26

Colleague health and wellbeing 27
Issues were raised regarding burnout, stress, anxiety and even suicide resulting from the 'extreme pressure' and 28 'overwhelming workload'. There were concerns about GPs leaving the profession at 'an alarming rate', and calls for 29 support for GP wellbeing. There was resistance to 'resilience training', seen by some as 'blaming colleagues' and thus 30 'insulting'; others welcomed 'support groups', 'retreats' and being kind to colleagues. 31 GPs expressed worries for their colleagues' safety, particularly given lack of PPE and testing, likening GPs to 'soldiers 32 fighting without armour'. GPs felt their colleagues were 'putting their lives on the line', particularly returning retired 33 GPs and BAME GPs. In response to the risks, GPs reported that colleagues were 'preparing their wills', looking into 34 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ; death in service benefits or seeking guardianship of their children in preparation for the worst. Other posts 1 announced colleagues hospitalised with or dying from COVID-19, and numbers of GPs dying. 2 Workload 3 Throughout the time-period under consideration, GPs expressed anxiety over their workload. This resulted in GPs 4 reporting working long shifts, frequently working over 50 hour weeks, working on days off, not taking annual leave 5 or bank holidays. Before COVID-19, primary care was described as at 'breaking point'. During the pandemic, 6 workload was described as having 'gone through the roof' creating 'immense pressure' with GPs 'pushed to the 7 limit'. The situation was described as unsustainable. 8 Additional pressures reported included the increase in remote consultations (described as taking longer), NHS 111 9 referrals, hospitals reducing non-COVID-19 services, keeping up to date with COVID-19 evidence and safe working 10 practice guidelines, dealing with patient shielding lists, donning and doffing PPE, sanitising between patients, 11 requests for mask exemption letters, shielding notes, isolation notes, sick notes, and early ordering of prescriptions. 12 In addition, patient demand was perceived to have increased due to a rise in mental health issues, an expanded flu 13 vaccination programme and the COVID-19 vaccination rollout. 14 A very small minority of posts related to a reduction in workload, mostly during April 2020 (see Table 2). 15 Discussion regarding ways in which workload could be managed centred around improving mental health services, 16 community volunteers, self-referral services, increased capacity, and managing patient contact with practices for 17 unnecessary reasons (e.g. vaccination dates). Some GPs reported working part-time as a means of coping with the 18 workload. 19

Emotions and stress 20
GPs reported anxiety about their own safety, the safety of their families, the 'tsunami' in workload, and lack of 21 resources. Many stated that they were 'fatigued' or 'exhausted', 'fearful' about the level of care for patients, and 22 'heartbroken' by patients suffering or dying alone. GPs talked about being 'fearful about the uncertainties', and 23 'dreading the future'. 24 GPs also referred to the pressures of providing care in a pandemic, using phrases such as 'unbearable pressure', 25 'completely overwhelmed', and 'never felt so exhausted'. Some perceived an impact on their mental health, with 26 comments that they felt 'mentally drained ', 'broken', 'wiped out', 'worn down', 'teary' and 'burnt out'. 27 GPs reported feeling like 'nobodies' or 'expendable' and viewed this as resulting from government and media 28 actions. Low morale was exacerbated by 'false rumours' and a 'constant attack on GPs' by media, government and 29 the public. Many GPs commented on how they were feeling frustrated or 'insulted' by public behaviour such as 30 noncompliance with lockdowns or isolation, not wearing masks, and vaccine uptake. Others mentioned feeling guilty 31 because they were shielding or taking annual leave, or even feeling guilty for catching COVID-19 as if it was a 32 'lifestyle choice'. Some felt 'not listened to' and powerless in the face of adversity. 33 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 From late November there were more positive posts with GPs reporting that they were at last 'feeling hopeful'. 1 Reports of getting the COVID-19 vaccine were met with comments like 'Fantastic Day!', 'feeling privileged' and 'so 2 delighted'. Those reporting on their involvement with the vaccine roll-out described feeling 'emotional' and 'proud'. 3 Experience of  In the earlier posts, before testing commenced, many felt confident that they had COVID-19 stating that they 'had 5 textbook symptoms' or 'had seen enough cases to know'. Some had confirmation later in the year via an antibody 6 test. Many reported receiving their COVID-19 vaccination. 7 8 Risk 9 Some GPs resigned themselves to 'inevitably' catching COVID-19. GPs commented that they were more vulnerable 10 than other professions such as shop workers and those in secondary care, stating that 90% of patient contact is with 11 GPs and contact is at closer proximity. Concerns around risk to BAME GPs centred around the disproportionately 12 higher death rate in BAME GPs and calls for 'appropriate measures' to be put in place. These views were more 13 commonplace among BAME GPs. 14 Once the vaccine was available, some GPs were frustrated by delays to their own vaccination. 15 In terms of concern for family members, some GPs talked about 'living in fear of unknowingly passing it on to my 16 family and loved ones', particularly more vulnerable family members. 17 Communication 18 There was praise for primary care teams 'pulling together' and a clear sense of 'solidarity', alongside comments 19 about how well community teams and volunteer/good neighbour schemes worked with GPs. While closer working 20 relationships between primary and secondary care were referenced in some posts, there was a realisation that this 21 was much needed. A 'Berlin Wall' and a 'them and us mentality' was described between primary and secondary care. 22 Criticisms of hospital communications included hospitals 'bouncing back' GP referrals, and delays in patient test 23 results. There were calls for better IT systems and for secondary care staff to spend time in primary care. Some 24 praised technology and 'online channels' that enabled improved communication and made the situation more 25 'bearable'. 26 Pleas were also made for better communication between the NHS and government, particularly as GPs had no 27 warning of policy announcements such as shielding changes, and flu and COVID-19 vaccine roll-outs. 28

Self-care 29
GPs were very aware of the potential impact of the pandemic on their mental health; some reported looking after 30 themselves, mostly through exercise and eating well, as well as some 'self-care' activities. The importance of taking 31 annual leave and having days off 'even in the middle of a pandemic' was also emphasized. 32 Others, though, disliked the 'self-care mantra' and felt resilience planning was insufficient to 'reverse the 33 unprecedented levels of stress faced by primary care doctors today'. 34 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The engagement of UK GPs' with Twitter made it possible to conduct a mixed-methods social media analysis to 3 explore large volumes of tweets relating to their perspectives and wellbeing during the pandemic and to compare 4 this with pre-pandemic. The analysis reveals trends in the social commentaries made by GPs during the pandemic, 5 including issues pertinent to GPs that may have affected their wellbeing. In our quantitative analysis, amongst a 6 number of interesting patterns observed over this period, those of particular note relate to the strength of feeling 7 around protection, risk and testing during the first wave of the pandemic; communication issues with and lack of 8 appreciation by government, secondary care and the public, and peaks in commentary during the initial COVID-19 9 vaccine rollout. Similarly, our qualitative thematic analysis revealed key issues around perceived lack of resources 10 and support, which had implications for GPs' safety, workload and wellbeing. Perceived lack of support from 11 government, media and the public affected their morale. 12

13
Our analysis identified comments about wellbeing which in 2019 were more predominantly related to patients, 14 whereas 2020 saw more focus on GP wellbeing. Posts related to coping commonly reflected work pressures and 15 fluctuated throughout the time period studied, suggesting that such views were prevalent well before the pandemic. 16 Our qualitative findings highlight the perceived sources of increased workload and stress during the pandemic, 17 including rapid moves to remote working (with remote consultations described as taking longer), GP self-isolation or 18 shielding increasing pressures on colleagues, poor or confusing dissemination of policy guidance, increased patients 19 with mental health problems, time taken cleaning and donning/doffing PPE. Meanwhile, frustrations were raised 20 following media and even government references to GP practices being 'closed'. 21

22
One other study has explored health care professionals' wellbeing using social media, finding issues of lack of PPE 23 and testing and changes in practice due to telemedicine predominate amongst US doctors [16]. Top phrases by 24 physicians were 'help us' and 'need PPE'. This concern was also voiced by UK GPs in our study. The US study also 25 found discourse regarding unemployment (including furlough and pay cuts) was high among US physicians [16], 26 which we did not identify. This may reflect differences between US and UK primary care. 27 It is already known that health professionals use social media to create virtual communities [15]. This is also evident 28 from our study by how many of the GPs in our sample follow and send messages (often of support) to other GPs,29 including to others in our sample, demonstrating the degree of connectivity and support between GPs on Twitter. The use of social media to explore GP wellbeing is novel and this paper is the first focusing on UK GPs. Our analysis 33 ended in February 2021 but debate has continued, for example tensions arose in May 2021 due to a further call from 34 NHS England and Improvement for GP practices to be 'open'. GP workload resulting from public misconceptions 35 about lateral flow testing for symptomatic patients was just beginning to be discussed in our data. 36 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 1 As with all social media research, we are limited by its content and by the sample. Online personas may be different 2 to offline personas and GPs may be strategic in how and what they post. GPs have in the past been sued for 3 discussing patients in such forums. Carville (2020) suggests they may also be cautious in discussing workplace issues 4 for fear of disciplinary action [37]. GPs in our sample appeared to share their general views and opinions openly, but 5 there was a tendency for them to refer to experiences and concerns around the wellbeing of colleagues, public 6 access to GP services or public mental health rather than discussing their own personal experiences. 7 8 Twitter may not fully represent the demographics of the GP population. In general, social media users tend to be 9 younger [38] and have a higher level of education [38][39][40]. In other respects, such as gender, race and ethnicity they 10 tend to reflect the population [39,40]. Although we identified GPs from different ethnic groups and regions, our 11 sample was somewhat over-representative of GPs who are white, male, and living in London. Insights demonstrate the impact of the pandemic on existing pressures that may provide some potential areas for 20 improvement in GP wellbeing: communication with secondary care, investment in staff and other resources, media 21 relations and a need to re-establish teams and wider networks that were disrupted during the pandemic. 22 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 22, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021