Family medicine practitioners’ stress during the COVID-19 pandemic: a cross-sectional survey

We conducted an online cross-sectional survey of every GP registered in the Auvergne-Rhône-Alpes region, a 7,948,287 inhabitant’s administrative area where the first clusters of cases were reported [10]. The eligibility criteria was to be a GP registered in the Auvergne-Rhône-Alpes area, agreeing for data collection and analysis. Exclusion criteria was to be a medical student or a GP working exclusively in hospitals. To reach the GPs, we released our survey through the official mailing list of medical professionals’ network for the area (union régionale des professionnels de santé, URPS). This mailing list included 5344 GPs. They were invited to fill an online self-questionnaire, through the software of a private society (GIDE). This cross-sectional survey was conducted from the perspective of a prospective cohort to monitor the self-perceived stress of GPs of the Auvergne-Rhône-Alpes area over the first wave of the pandemic. Each GP participating in the survey were asked to be enrolled within the cohort. Herein we are reporting the results of this baseline assessment, composed of answers collected from 8th April to 10th May 2020. Throughout the study period, ~ 10% of French COVID-19 cases and deaths occurred in this area, and the national lockdown was in place.

The self-questionnaire was regrouping five main parts: 1) socio-demographic characteristics of GPs, 2) practice and workflow before and during the pandemic, 3) their self-perception of stress according to the PSS-10 validated in French, 4) their agreement with 31 positive assertions to identify potential origin of stress, and 5) an open-ended question (Supplementary Table 1). It was devised by the scientific study committee, involving: 2 residents in family medicine, 2 infectious diseases physicians, 3 specialists in epidemiological studies including prevention of stress at work, and 1 methodologist. It was tested on 12 GPs in a pilot phase to ensure the comprehension and calibration of the questions.

The Perceived Stress Score with 10 items (PSS-10) was used to assess the level of self-perceived stress [11]. It can be self-administered, and is validated for French with a Cronbach’ alpha coefficient above 0.70 [12]. The PSS-10 allocates GPs under three categories: “not stressed” (PSS-10 ≤ 20), “borderline” (21 ≤ PSS-10 ≤ 26), and “stressed” (PSS-10 ≥ 27) [11]. We also identified and analyzed a “very stressed” (PSS-10 ≥ 30) category [11].

The 31 positive assertions were developed by the scientific study committee to explore the root-causes of stress. They were regrouped under 8 themes: 1) workload (3 assertions); 2) emotional requirements (5 assertions); 3) conflict of values (5 assertions); 4) economic insecurity (2 assertions); 5) working relationships and social reports (4 assertions); 6) personal protection equipment (5 assertions); 7) access to information (4 assertions); and 8) others (3 assertions) [13]. For each affirmation, GPs had to express their agreement using a 10-point numeric scale, ranging from one (“I do not agree at all”) to ten (“I totally agree”). An agreement ≥6/10 was considered as “agree”, while an agreement < 6/10 was considered as “not agree”.

In complete cases, we explored factors associated with a self-perceived stress defined as a PSS-10 ≥ 27, using a multivariate logistic regression adjusted for gender, age as continuous value, and practice location. No imputation of missing data was performed. The variable selection was made using a backward elimination based on the Akaike criterion. The associations are represented using odds ratio (OR) with their 95% confidence interval (95CI). The level of significance was set at 5% bilateral. Statistical analysis was performed using the R software version.4.0.2. (the R foundation, Vienna, Austria).

Over the study period, we collected 1050 GPs’ responses, of whom 5 worked exclusively in the hospital, 6 were not actually in the targeted administrative area, and 141 (13%) have not validated their questionnaire. Therefore, the final set used for analysis was constituted of 898 GPs’ responses (17% of the 5344 emails sent, and 86% of the initial responses), of which 879 were complete.

Table 1 describes GPs’ characteristics. The majority of respondents were women (60.5%) and the average age was 47.7 years. Women were younger than men (44.9 vs. 51.9 years, p < 0.001). The sample was representative of the GP population in the area, with a small gap in gender (61% of women vs. 51 to 54% in the control data from URPS and the French Ministry of Health) and age (47 on average vs. 48 to 52 in control data).

Overall, the average PSS-10 score was 26.4 (±6.4): 169 GPs (19%) were “not stressed” (PSS-10 ≤ 20), 292 (33%) “borderline” (21 ≤ PSS-10 ≤ 26), and 437 (49%) “stressed” (PSS-10 ≥ 27). A total of 283 GPs (32%) were “very stressed” (PSS-10 ≥ 30).

They were 880 (98%) GPs to adapt their practice and facilities to implement barrier measures. Issues in obtaining protective equipment were reported by 531 (59%) of them. Most of GPs (741, 83%) considered having less work over the study period (Table 2).

Two-third of GPs had someone to talk to (556, 62%), they were mostly supported by their family and friends (617, 69%), and very few had the willingness to withdraw from practice (46, 7%).

Table 3 details the results of the multivariate analysis of complete cases. Over 879 GPs, women were twice as stressed as men (OR = 1.88, 95%CI [1.31–2.72], p < 0.001). In addition to gender, other factors were found being a source of self-perceived stress among the 31 positive assertions: making difficult decisions, being affected by patient anxiety, being overwhelmed by information, having a heavy workload, having the feeling of being alone, and the feeling that work time impacted on personal life. Factors associated with lesser stress were: being in line with the job, having the feeling of being useful, having trust in the future and having the ability to forget work when reaching home.

We collected 294 answers through the last open-ended question, of which 202 were suitable for verbatims analysis (Supplementary Table 2). A total of 153 verbatims were related to the 8 themes developed for the 31 positive assertions, and 173 verbatims were about 5 new themes. The three main new themes were: 1) The place of GPs in the organization of the response to the health crisis (n = 72); 2) The feeling of GPs during health crisis (n = 67); and 3) New concepts about information and guidelines (n = 36). From these three new themes we could describe three networks for communication around GPs during the health crisis. The first one was between health authorities and GPs. In this first network, the communication was a one-way flow from health authorities to GPs. GPs considered that information they received was too vague, unsuitable to their practice, and sometimes contradictory (n = 58). They described a difficulty to assimilate knowledge as guidelines were frequently modified (n = 29). Few noted that they had the same type and level of information as their patients (n = 8). The second network was a two-way flow between hospitals and GPs, which was sometimes reported as being non-existent (n = 13). This lack of cooperation has led to feelings of frustration and sometimes even guilt (n = 7). The last network was a two-way flow between GPs themselves, and was reported. It has been in many places as being a source of mutual help (n = 29). Finally, GPs had the feeling of being neglected, not taken seriously and not fully considered by health authorities (n = 49), while they were standing alone in the hospital-based response to the pandemic (n = 43).