The data were collected in the Free State of Saxony, Germany, between May 2021 and July 2021. GPs were selected by availability of their e-mail addresses, which were obtained from the Association of Statutory Health Insurance Physicians Saxony (Kassenärztliche Vereinigung Sachsen). In 2020, a total of 1632 GPs with a mean age of 54 years were practicing in Saxony, of which 62.4% were female (36). GPs were invited by e-mail to voluntarily participate in this online survey. In May 2021, the first invitation e-mail was sent to 1444 GPs. The email contained information about the purpose of the survey, pseudonymous data handling, and a link to the online survey. The access to the survey was provided by an individual access code (token) to ensure that each GP participated only once. In June 2021, the first reminder containing the same information was sent to non-responders of the first e-mail [following the recommendations of (37)]. In July 2021, a second reminder was sent to non-responders of the second e-mail (see recruitment process in Figure 1).

The questionnaire was self-developed in the Department of General Practice of the Leipzig University by an interdisciplinary research team (medical scientists, biochemist, and GPs) in a multi-stage revision process. The development process was complemented with extensive literature research aimed at identifying relevant factors for the treatment of long COVID. The final version can be found in Supplementary material S1. For the web-based evaluation, the software LimeSurvey (http://www.limesurvey.org/) was used, hosted on a secure server of the Leipzig University Computer Center. The completion of the online survey took ~15 min.

Participating GPs were asked to click the “I agree” button of the online informed consent form. Then, the survey started and comprised of the following topics: (1) demographics (e.g., age and sex), (2) current number of patients with acute and long COVID and their symptoms, (3) current treatment of long COVID symptoms. The response formats were choice answers, 4- and 5-point Likert scales, and free text entries. Likert scales for assessing the frequency of observed symptoms ranged from mostly, often, rarely, very rarely to never (5-points). Likert scales for assessing the current capabilities for diagnosis and therapy of acute COVID-19 and long COVID ranged from very good, rather good, rather poor to very poor (4-points).

Prior to implementation, the questionnaire underwent a think-aloud pre-testing (38) aimed at identifying problems or misunderstandings related to each item and to further develop the questionnaire. The provisional questionnaire was filled out by five GPs, who were instructed to think aloud while answering each item and report every spontaneous thought. After completing the questionnaire, the GPs were briefly interviewed about general issues with the questionnaire (e.g., length, structure, and general comprehensibility). After pre-testing, the provisional questionnaire was adjusted and further developed.

Participating GPs were asked to indicate the following, if needed, in free text fields: additional symptoms observed in acute and long COVID patients as well as medical and non-medical treatment and rehabilitation plans for 13 common long COVID symptoms. Free text entries were re-coded in major and subcategories by two authors of this study (AS, AB). Differences in coding were discussed until agreement was reached. Additional observed symptoms were in some cases aggregated if synonymous terms appeared in one field. Further, drugs were reported as generics, trade names, drug classes, or abbreviations. All drug entries were then grouped in the main category “drug class” and trade names or generics were coded as subcategories. As dosages or length of prescriptions were rarely stated, these statements were not analyzed.

The study was carried out in accordance with the Declaration of Helsinki and the study protocol was approved by the research ethics committee of the Leipzig University (reference number 157/21-ek). Online informed consent was obtained from all participants. They did not receive an incentive for their participation. No personal data besides age, sex, and education level were assessed.

All statistical analyses were carried out using IBM SPSS Statistics 27 (Armonk, NY, USA). For descriptive statistics, missing values in single variables were considered by presenting frequencies as % (n/n_valid). Continuous variables were presented as mean ± standard deviation (SD).

Of the 186 GPs who participated in this study (13% total response rate), 43 GPs left the survey incomplete after the first page (demographic information). In total, 143 GPs were included in the analyses (mean age = 50.2 years, 61.1% female). GPs in our sample were slightly younger, but did not differ in gender distribution compared to the total population of GPs in Saxony.

First, we assessed the current number of patients with long COVID treated in GP practices. Almost all participating GPs had previously treated patients with persistent symptoms after SARS-CoV-2 infection. On average, each GP practice was treating 12 patients with symptoms lasting between 4 and 12 weeks after infection with the coronavirus between May and July 2021. During this survey period, approximately six patients with long COVID symptoms lasting longer than 12 weeks were treated by each participating GP. Earlier in the pandemic, 53.8% of French GPs treated at least one patient with long COVID in their practices, particularly in the most strongly affected areas (39). With the progression of the pandemic, it is conceivable that the long COVID syndrome will play a lasting role in GP practices.

We also assessed the most frequent symptoms in patients with acute COVID-19 and long COVID who presented in participating practices. In line with previous research (2–4), GPs most frequently observed cough, fatigue, fever, and loss of smell and taste in patients with acute COVID-19, indicating that the patient population seeking treatment by their GP and with a potentially milder courses of the disease exhibited a comparable spectrum of symptoms than, for example, clinical patient populations. Free text analyses revealed recurring observations of pain symptoms, especially aching head and limbs. Around one in five of the patients with COVID-19 presenting in GP practices in our study were on sick leave for more than 6 weeks due to the severity of their symptoms, which is consistent with previous studies (40).

In patients who reported symptoms lasting between 4 and 12 weeks, GPs frequently observed cognitive and psychosomatic dysfunctions, such as fatigue, reduced performance, and lack of concentration, persistent respiratory dysfunctions, such as cough and dyspnea, as well as loss of smell and taste, muscle and joint pain, and headaches. Our results are in line with patient reports on the most pervasive symptoms (3, 6, 7, 14, 21, 41–43). Additionally, GPs mainly observed cognitive and psychosomatic dysfunctions, such as fatigue, reduced performance, and lack of concentration, in patients with persistent symptoms lasting longer than 12 weeks, which is in accordance with previous reports (6, 42). Although depressive symptoms or anxiety have been found to be common in patients with long COVID (44, 45), participating GPs rarely mentioned these symptoms in our study. Potential explanations could be that GPs treat to a great extent patients with a milder course of the disease and therefore with a lower prevalence of depressive symptoms. However, while some studies found fewer depressive symptoms and anxiety in non-hospitalized compared to hospitalized patients (14) and in ward patients compared to ICU patients (46), other studies found no relationship between initial disease severity and depression (45, 47), indicating that further research is urgently needed. It is further conceivable, that the frequent occurrence of fatigue and lack of concentration reported by GPs is associated with underlying depressive symptoms, as has been shown in previous studies (48, 49). Finally, the diagnosis of depression, especially in less severe cases, has been found to be low among GPs (50–52). Besides a lack of knowledge, several reasons have been discussed: GPs might want to exclude somatic diseases first and observe the symptoms and the circumstances over a longer period of time before diagnosing depression and initiating a respective treatment (50, 53). Further, patients might underreport mental symptoms due to associated stigma (50, 53, 54). Lastly, patients without prior depressive episodes have been found to be less likely diagnosed as depressed by their GPs (50, 55), indicating that newly emerging depressive symptoms following infection with SARS-CoV-2 might be less likely recognized.

Our results suggest that the long COVID syndrome is widely established in German GP practices and that addressing cognitive, psychosomatic, and respiratory impairments plays an especially prominent role in the treatment of these patients.

In addition, we investigated GPs' strategies to treat patients with long COVID symptoms. At the time of the survey, a national guideline for long COVID and causal therapies were lacking, suggesting that GPs made their treatment decisions under uncertainty (56–58) and may have focused on symptom-oriented treatment. In the meantime, several management recommendations for the outpatient setting have been proposed (26, 31, 34, 59–61) and a national guideline was established. In our study, most GPs rated the current treatment options for long COVID as poor or rather poor. Symptoms such as loss of smell and taste, fatigue, lack of concentration, loss of hair, or skin diseases might be particularly difficult to manage for some GPs, indicating that treatment options for these symptoms are limited or symptoms are difficult to define. These perceptions are in line with a recent review showing a discrepancy in the literature between the current knowledge about long COVID characteristics and rehabilitation recommendations (60). In contrast, GPs felt more confident in treating symptoms such as thrombosis, cough, headaches, or dyspnea, which might be due to more available medication, treatment, or referral options.

Further, our results show the use of a variety of medications for the treatment of long COVID symptoms. Compared to previous studies reporting an increased prescription of triptans in case of symptoms such as headaches or migraine (5), GPs in our study rarely utilized triptans and instead mainly prescribed acidic and non-acidic non-opioid analgesics (e.g., paracetamol, ibuprofen) to patients with long COVID presenting with headaches. We also did not find an excessive prescription of antibiotics, which has been reported elsewhere (62). However, the use of corticoids or bronchodilators for respiratory symptoms were common, which has also been found in other studies (5, 62). Notably, none of the drugs have been approved for the treatment of long COVID so far and their safety and effectiveness needs to be further evaluated (63). Further, some GPs also recommended a variety of supplements, such as vitamin D and vitamin B-complex. Whereas, a higher vitamin D level might be protective against a worse outcome after infection with SARS-CoV-2 (64), the therapeutic effects of vitamin D supplementation have not yet been confirmed (65). To our knowledge, the evidence for therapeutic effects of B vitamins is to date non-existent.

In addition to drug therapies, GPs in our study often referred patients to physical therapies and moderate exercises to improve symptoms and performance, which have also been recommended by current research (32, 60, 66). However, physical exercises can also trigger symptom relapses, especially in patients experiencing dyspnea, muscle and joint pain, or chest pain (7, 34). In addition, reports from patients with long COVID highlight the need for an individually tailored physical activity plan establishing thresholds to avoid relapses (28, 67). Particularly, unadjusted physical activity plans and a subsequent worsening of symptoms might discourage patients from further training (67). Further, in accordance with a recent review (60), the reported use of non-drug therapies for patients with long COVID in our study focused primarily on exercise and physical rehabilitation, whereas the use of educational, behavioral, or psychological therapies was low. Given the potential complex sequalae following an infection with SARS-CoV-2, a more comprehensive and holistic rehabilitation is recommended.

Our results also show that specialist referrals for long COVID symptoms were common, which might increase coordination difficulties for GPs, especially in patients with more complex symptoms. The urgent need for multi-disciplinary rehabilitation teams addressing persistent symptoms and a low threshold for referrals has been discussed previously (26, 32, 60), however, access to specialists as well as their perceived insufficient communication with GPs might hamper effective therapy of patients experiencing long COVID symptoms (29, 56).

The treatment of patients with long COVID was assessed via individual symptoms rather than by often occurring groups of symptoms, potentially leaving some more holistic treatment strategies unreported. Further, as our data were collected in mid-2021, our results might not be applicable to long COVID manifestations caused by other SARS-CoV-2 variants occurring after the study period (e.g., Omicron). We also did not assess potential differences in long COVID symptoms in hospitalized and non-hospitalized patients in the acute phase of the SARS-CoV-2 infection. Another limitation is the absence of patient data or case reports for each therapy and treatment strategy reported by GPs in our study. This also includes potentially differing treatment strategies applied by GPs in patients with pre-existing illnesses or co-morbidities to avoid, for instance, drug-drug-interactions (68) or a worsening of the underlying disease (69). Further, we did not ask GPs about additional diagnostic methods, such as X-rays or blood tests, potentially used to guide their decision. Given the nature of our study, a selection bias might have occurred. Lastly, we did not conduct a power analysis, wherefore the data are to be considered exploratory and cannot be generalized.

Our study contributes information on the current state of treatment and rehabilitation plans for patients with long COVID symptoms presenting in German GP practices. As evidence-based management recommendations and approved drugs were absent at the time of the survey, our results indicate that GPs relied on their experience with related conditions when treating patients with persistent symptoms after infection with SARS-CoV-2. Further, as research is constantly advancing the knowledge about long COVID management, GPs should be informed regularly about new or adjusted options and newly emerging treatment programs, e.g., through workshops. However, our results also emphasize a gap between the current knowledge of the long COVID manifestation and knowledge about effective rehabilitation. The current German national guideline on post-/long COVID published shortly after our survey was conducted provides broad recommendations but lacks specific evidence-based guidance on the treatment of some symptoms. Research and health authorities should especially focus on the development of holistic and multi-professional treatment plans for common, but more ambiguous long COVID symptoms, that are still difficult to manage for GPs, such as loss of smell and taste, fatigue, or lack of concentration.