To prevent being stressed-out: Allostatic overload and resilience of general practitioners in the era of COVID-19. A cross-sectional observational study

Abstract Background Responsibility of general practitioners (GPs) in delivering safe and effective care is always high but during the COVID-19 pandemic they face even growing pressure that might result in unbearable stress load (allostatic overload, AO) leading to disease. Objectives We aimed to measure AO of Hungarian GPs during the COVID-19 pandemic and explore their recreational resources to identify potential protective factors against stress load. Methods In a mixed-method design, Fava’s clinimetric approach to AO was applied alongside the Psychosocial Index (PSI); Kellner’s symptom questionnaire (SQ) to measure depression, anxiety, hostility and somatisation and the Public Health Surveillance Well-being Scale (PHS-WB) to determine mental, social, and physical well-being. Recreational resources were mapped. Besides Chi-square and Kruskal-Wallis tests, regression analysis was applied to identify explanatory variables of AO. Results Data of 228 GPs (68% females) were analysed. Work-related changes caused the biggest challenges leading to AO in 60% of the sample. While female sex (OR: 1.99; CI: 1.06; 3.74, p = 0.032) and other life stresses (OR: 1.4; CI: 1.2; 1.6, p < 0.001) associated with increased odds of AO, each additional day with 30 min for recreation purposes associated with 20% decreased odds (OR: 0.838; CI: 0.72; 0.97, p = 0.020). 3–4 days a week when time was ensured for recreation associated with elevated mental and physical well-being, while 5–7 days associated with lower depressive and anxiety symptoms, somatisation, and hostility. Conclusion Under changing circumstances, resilience improvement through increasing time spent on recreation should be emphasised to prevent GPs from the adverse health consequences of stress load.


Introduction
The ongoing pandemic of COVID-19 turned out to be a strong stressor for all medical doctors, causing psychological distress and mental health problems [1]. It demanded hospitals and specialist care to transform into pandemic centres. This has increased the responsibility of family physicians working in primary care to screen and treat serious cases requiring skills specific to other specialities. They were also expected to run their consultations online without physical examination, and were not sufficiently equipped to contact patients when needed [2]. In the last decades, general practitioners' physical and mental health has come into focus [3,4]. Besides extreme workload, moral implications for 'good doctoring' increased their workrelated stress. Major events, but subtle, chronic daily experiences as wellwhich an individual perceives as stressfulactivate regulatory systems (the autonomic, neuroendocrine, metabolic, and immune system) to change a set point and operate at elevated or reduced levels [5,6]. This is called allostasis, the process to achieve stability through change [7]. Increased catecholamine, cytokine and HPA hormone levels are the mediators of this adaptational process resulting in elevated heart rate, blood pressure or inflammation [5]. However, long-term activation of the regulatory systems by repeated stress will lead to overuse and dysregulation of the mediators of allostasis, causing allostatic load, manifesting in anger, fatigue, frustration and feeling out of control ('stressed-out') [8]. When challenges exceed the individual's coping ability, allostatic overload will be the result, a condition with consequent diseases (e.g. hypertension, depression, arthritis, metabolic syndrome or tumorous diseases) [9][10][11]. To understand the role of allostatic overload in the background of ill-health [12], identification of individual stressors, clinical signs and symptoms directly related to stress sources and the individual's response to the stressors give the cue [10,13]. Scientific literature concerning GPs' health focuses primarily on mental ill-health [14]. This is even more essential with the burden of the pandemic on the health care system worldwide. Such an exceptional situation, however, should also lead to exploring sources of resilience beside identifying distress. Increasing well-being will contribute to reaching optimum health through positive affect, personal relationships, and a meaningful and optimistic life [14][15][16]. Besides, cognitive-behavioural stress-management techniques and mindfulness-based education programmes [14], recreation has recently come into focus as a positive coping response to stress [15,17].

Study objectives
We targeted to define the prevalence of allostatic overload among Hungarian general practitioners during the first wave of COVID-19 and define the most important factors associated with it. We postulated that the infection and the related confinements and proceeding rules concerning primary health care resulted in significantly increased stress load of professionals. Additionally, we aimed to measure their wellbeing, regularity, and forms of recreational activity they attain andconsequentlyif these might associate with increased mental and physical health or increased resilience against stress load.

Study design and sample recruitment
We performed a voluntary online survey among Hungarian GPs between 28th August and 16th October 2020. Participants were recruited between 28th and 30th August via institutional sources (1,262 registered email addresses of surgeries or doctors throughout Hungary) and then one reminder was sent between 8th and 10th September. Our invitation letter contained that the Family Medicine Department at Semmelweis University conducted the survey, the time frame for completing the survey (15-20 min) and we defined our aim as to explore the effects of the previous 6 months (the first wave of the pandemic) on them as family physicians and as persons. We did not offer monetary or non-monetary incentives. Personal data was not collected, butto allow possible follow upwe generated an ID code for each participant. We constructed our survey so that all answers had to be given to continue with the survey; therefore, participants answered all questions, and we did not need to exclude anyone due to incomplete questionnaire reply.

Ethics
Online consent was secured by all participants. The study was conducted by the Declaration of Helsinki and was approved by the review board of the Medical Research Council (IV/5657-2/2020/EKU).

Measurements
Sociodemographic and health-related characteristics of the sample. We collected data on participants' age, gender, and place of living (capital, county seat, town or village); on working conditions (actively working during the pandemic; method of working (personal, phone consultations, other), uncertainty about coronavirus in comparison to the first wave (no change, decreased, increased)). We asked if they took an active role in maintaining their health and the number of days they did recreational activities for at least 30 min. We also asked for the number of chronic diseases, any diagnosed psychiatric disease, the number of prescribed and over-the-counter medicines taken daily.
Allostatic overload. We measured COVID-related allostatic overload according to Fava's definition based on the Diagnostic Criteria for Psychosomatic Research-Revised (DCPR-R) and used the Psychosocial Index (PSI) self-rating questionnaire by the same authors to measure each criterion [13,18,19]. The PSI includes 55 items. Sociodemographic and clinical data are measured from 1 to 12, perceived and objective stress by items 13-20 and 22-30 in a YES/NO format with a maximum score of 17, and well-being by items 31-36 with a score ranging from 0 to 6. Psychological distress is measured by items 37-51 addressing symptoms of sleep disturbances, somatisation, anxiety, depression, and irritability on a 0-3 Likert scale with a maximum score of 45. Abnormal illness behaviour contains items 52-54, concerning bodily preoccupations and hypochondriac beliefs on a 0-3 Likert scale with a range from 0 to 9. Quality of life is measured by one direct question (item 55) with 5 possible choices from excellent to awful [19].
We applied these toolsin accordance with previous research [20,21] to measure COVID-related allostatic overload (Table 1). Besides measuring individual stressors, our primary focus was on COVID-related allostatic overload. Therefore, we tailored A2 criterion of DCPR-R to COVID as a particular stressor. According to the instructions provided in the DCPR-R allostatic overload is diagnosed when A1 þ A2 þ B1 or B2 or B3 is present. To measure stress load independent of COVID-19, we applied PSI questions 13-20 and 22-30 [19].

Mental health and somatisation
Mental health was measured with the Kellner Symptom Questionnaire (SQ) and the Public Health Surveillance Well-being Scale (PHS-WB) [22,23]. SQ consists of four scales: depression, anxiety, somatisation, and hostility, each divided into two subscales, one for the symptoms DCPR-R criteria defined allostatic overload with related items from the Psychosocial Index self-rated questionnaire. Text in italics (fulfilling A2 criterion) was formulated to be specific to COVID epidemic as a stressor. PSI does not contain A2 criterion [19,20].
(depression, anxiety, somatisation and hostility) and the other for well-being (contentment, relaxation, physical well-being and friendliness) [22]. The 10-item shortened version of PHS-WB was used to measure physical, mental, and social well-being. The first five items (on scale 0-5) result in a score of mental well-being. The following two items measure social well-being with scales from 0 to 10. The last three items provide the score of physical well-being after their scales being unified. Total wellbeing is then calculated from all converted scores [23].

Qualitative methods
To identify the most burdening challenges Hungarian GPs had dealt with in relation to the pandemic, we included the following question in our survey: 'What was the biggest challenge for you during the epidemic and the quarantine?' Participants gave free-text answers, which ranged from single-word answers to paragraphs. Following standard qualitative analytical procedures, each researcher read all free-text responses systematically, identified blocks of text that reported factors contributing to allostatic overload, and assigned provisional code names. They compared their coding schemas and agreed on a common one. They then examined the codes, identified themes that organised them into higher-level concepts that explained the origins of overload, constantly comparing their interpretation with the original data, and agreeing on a final interpretation (Tables 2 and 3; Figure 1).
To create categories of stress releasing recreational activities, we selected the Mental Health Foundation (UK) 'How to manage and reduce stress' booklet as well as the American Counselling Association's article '100 Ways to Reduce Stress: Making the Balancing Act More Manageable' to base our choices. We offered multiple possibilities for recreation (connection with nature, reading or watching movies, physical exercise, meeting friends and acquaintances, cooking, praying or meditation, creative manual activities and DIY, or beautification and cosmetics) and participants were able to provide their answers on their sources of recreation as well. Their answers were then sorted and counted and presented in Figure 2.

Statistical analyses
Chi square tests were used in case of the categorical data, two-tailed t-test for normally and Kruskal-Wallis test for non-normally distributed continuous variables. Dunn's pairwise tests with Bonferroni adjustment for multiple comparisons were carried out for the three pairs of groups. Normality of data was assessed using the Kolmogorov-Smirnov test. In our cross-sectional study, we applied step forward likelihood ratio logistic regression analysis to estimate the role of age, sex, place of living, the number of chronic diseases, the number of stressors and the number of days the respondents applied at least 30-min recreation in the exposure to allostatic overload. We applied 95% confidence intervals (CI). In all cases, a p value < 0.05 was considered statistically significant. We used SPSS-24.0 software (SPSS Inc., Armonk, NY, USA).

Sociodemographic and COVID-related characteristics
After excluding 13 double fill-outs, we analysed the data of 228 GPs, 155 of whom were females. The youngest doctor was 32 years old while the oldest was 88. We did not find any statistically significant differences in health-related and sociodemographic characteristics according to sex (Table 4).
We found that 222 colleagues worked during the first wave of the pandemic. Two-thirds of them (155) worked in person in the surgery during the pandemic. They all used mixedpersonal, phone calls/video calls and onlinepossibilities for consultation.

Sources of stress
According to GPs' answers on the most challenging aspects of the pandemic and the related quarantine, qualitative data showed that work-related conditions and increased workload were the most challenging for the majority of GPs (Figure 1). Electronic prescription and the use of virtual health service space increased dramatically, causing challenge for less frequent users. Structural changes in delivering care (from personal to online and phone consultation) as well as decreased possibility for outpatient specialty care, stood as the most essential points. Additionally, they dealt with increased responsibility of calming and informing patients while also in fear and uncertainty ( Figure 1 and Table 2). As shown in Figure 1, general practitioners mainly reported professional challenges as most burdening but personal difficulties yield important as well ( Table 3).

Prevalence of allostatic overload and factors associated with it
Allostatic overload with somatic symptoms of distress or impaired social and occupational functioning was experienced by 60% (N ¼ 131)    No significant difference was found between the two groups on any of the variables.
CI: 0.72; 0.97, p ¼ 0.020) after adjusting for age, place of living and chronic diseases.

Recreation and well-being
Two hundred and seventeen (95.2%) out of 228 family physicians reported doing something actively for their health in general. They reported spending at least 30 minutes on recreation an average of 4 days a week. We offered multiple possibilities for recreation to choose from besides individual answers. The median number of different recreation types chosen was 4 (IQR: 3, 5). The most popular forms were connection with nature, reading or watching movies and physical exercise ( Figure 2). When grouping family physicians according to the number of days they recreated, we found that being involved in recreation at least 5 days a week associated with lower point scores on symptoms of anxiety, depression, somatisation, and hostility while just 3 days weekly associated with elevated scores on mental and physical well-being (Table 5).

Main findings
We found that 60% of participating Hungarian family physicians suffered from allostatic overload in relation to adverse life events during the first wave of COVID-19 pandemic. Females and those experiencing more stressors in their lives were more vulnerable. Each additional day when time was ensured for 30-min recreation associated with 19% decreased odds of this vulnerability. Elevated mental and physical well-being associated with at least 3 days; lower symptoms of depression, anxiety, somatisation, and hostility, with 5-7 days recreation weekly.

Strengths and limitations
There is insufficient literature mapping general practitioners' mental health, but even those few concentrate mostly on negative aspects of it [14]. It is a rarity to find studies on resources to promote well-being which also support the ability to cope and perform under extreme stress circumstances. The strength of our research is to explore distress symptoms (depression, anxiety, hostility, and somatisation) as well as wellbeing (mental, physical, and social) under an acute stressor (COVID-19) amongst GPs. We defined the association of regular recreation with lower distress levels alongside with higher level of mental and physical well-being. Our online survey reached an 18% response rate. Since response rates of 70% or higher are considered good, our response rate is low. Compared to other web-based GP surveys [24], however, our response rate did not seem inferior to others with similar constructions. One shortcoming of our data collection was that we could not separate non-respondents who did not receive the invitation (invalid email addresses) from those who did not provide a fill-in. Approximately one-third of the email addresses belonged to the surgery and not the doctor. High workload and administrative workload are main sources of GPs' nonresponse to surveys. Our results show that the COVID pandemic put extraordinary burden on GPs (increased workload was the fourth most important source of stress). This might increase the possibility of nonresponse, especially when the request arrived at surgery-related email address. Online surveys are less preferred than paper-based among family physicians. Computer illiteracy might be one cause for that. The finding supports that participant GPs found the changes in consultations most challenging. We found that our respondents were slightly younger than the average age of Hungarian general practitioners (57 years in the sample vs 64 years in the total population) and consisted of more female general practitioners (68% vs 53%). Similarly, French and Swiss GP respondents of a web-based survey were younger and contained fewer males than the community-based GP population [24]. COVID-19 related changes in Table 5. Mental health parameters of general practitioners according to the number of days they spent at least 30 min for recreation during the week (N ¼ 228). 30  professional and personal life or emotional or psychological discomfort related to this topic could also influence participation. Recreational sources are individual sets of interests, relations, values, and goals developing throughout life, and practising them is advised by experts to prevent 'corona phobia' [25]. Although we could predict their role in lowering the odds of allostatic overload, defining a true causal relationship will be achievable by longitudinal research.
Allostatic overload and the most important factors associated with it in GPs during the first wave of COVID-19 The first wave of the ongoing pandemic shed light on the psychosocial burden health workers faced [26,27]. Job strain, social isolation, fears of stigmatisation and uncertainty about the future added to stress, exhaustion, and depressive mood nurses and doctors had experienced [27]. While most studies focus on those in close contact with COVID-19 patients [20,26,27], quantitative data about the types and levels of COVID-19 related stress among family physicians are scarce, even though they are first contact to most patients. Recent research in a hospital environment has confirmed that job strain and uncertainty about the future were the most common causes of higher levels of stress and depressive mood healthcare workers experienced [28]. Our results are in line with these findings, showing that in primary health care settings changes in working conditions, uncertainty and emotional issues multiplied the burden of the pandemic on them. Females and those who experienced additional stressors simultaneously to the pandemic were at higher risk. Exploring mental ill-health and constituents of GPs' well-being enhance the knowledge in the field.

Mental health and well-being of GPs and regularity and forms of recreational activity
According to literature, general practitioners are more depressed than white-collar workers [29] and experience higher patient-related stress than other medical specialists while their self-estimated health and workability is lower [30]. However, the well-being of British general practitioners was comparable to the local population, and GPs above 55 years showed higher hope and optimism than their younger counterparts [31]. Our sample showed comparable levels of mental and social well-being during the COVID-19 pandemic to a community sample [21]; however, anxiety and hostility scored higher, probably referring to the high level of additional professional stress load. Most of the GPs ensured time regularly for recreation. According to our results, higher frequency of weekly recreation associated with higher mental and physical well-being and lower distress symptoms. A recent review article on interventions highlights that besides psychotherapeutic programmes [14], increasing awareness on thoughts, beliefs, self-care, personal health and selfcare boundaries improved mental health. Our results strengthen these findings because individually chosen types of recreation were equally able to improve mental health. This is even more important during the burdening time of the pandemic, when besides psychosocial support and a better infrastructure adjustment, leisure time is the second biggest resource following interpersonal connectedness [28].

Implications for practice
Besides providing eminent care for patients, it is of utmost importance to take conscious care of ourselves. Recreational activity can be easily achieved and is provenly effective in maintaining better mental and physical health and significantly reducing distress symptoms. Actively applying 30 minutes of recreation 5-7 days a week might dramatically improve our ability to succeed.

Conclusion
Our study demonstrates that Hungarian general practitioners were burdened by the first wave of COVID-19, with 60% of the participating physicians presenting allostatic overload. Professional challenges were most demanding, and females and those experiencing additional life stresses were more vulnerable. Regular recreation associated with elevated mental and physical well-being, lower distress symptoms and lowered odds of AO. Longitudinal research is needed to support our results further.