We analysed data from our three GP offices in rural Germany between the onset of Covid-19 in our country on 27.01.2020 until 20.04.2020. The mean age of our patients was 47.03 years, which is quite young considering that mortality seems to increase in COVID-19 patients beyond 65 years; patients less than 65 with little predisposing factors may be at a low risk of severe disease(5).
Only five out of 80 tested patients were positive for SARS-CoV-2 (positives). Due to this low number and a potential reporting bias of symptoms, we refrained from using mean comparison tests and only depicted absolute values. However, we noticed that almost all patients in the positives suffered from a sore throat (4/5; 80%), while only 12% in the negatively tested group (negatives) showed this symptom (9/75). Furthermore, rhinitis was more prominent in the positives (60% vs. 26.67%). Due to the limited case number generalisability of these findings is not possible.
Known comorbidities, especially pre-existing lung and cardiovascular disease, in positives and negatives were quite low (see table 3). The most common cardiovascular risk factor was arterial hypertension, which has already been published(5). SARS-CoV-2 uses ACE-2 as a cellular entry point(23) and has raised concern about continuation of RAAS inhibitor intake in patients with chronic heart failure(24). However, recent data has shown that there is no evidence of increased disease severity or mortality in hospitalized patients on RAAS blockers(25-27). Additionally, pharmacological data suggests that ACE-2 expression is not increased in patients on RAAS blockers(28). Thus, current recommendations support continuation of RAAS blockers in patients with arterial hypertension and chronic heart failure(24). In our study two out of five positively tested patients were on RAAS blockers and did not display more severe symptoms than the others. Also in the negatives RAAS blockers were the most commonly prescribed antihypertensive drugs (15 out of 23 patients received RAAS blockers), showing no difference in clinical severity.
We applied the first algorithm (figure 1) provided by the RKI(22) to decide which patient needed a smear. 6.25% of the tested patients were positive. This algorithm was introduced at the beginning of the crisis, at a time where travel restrictions had not been implemented across Europe. There had been limited knowledge on the diversity of symptoms associated with SARS-CoV-2: Therefore unspecific respiratory symptoms alone did not qualify to send a patient to a smear center. Traveling to a risk area and contact to positively tested patients were important criteria in the first RKI definition.
Schmithausen et al. found that early temporary loss of taste and smell occurred in 68% of oligosymptomatic patients tested positive for SARS-CoV-2 after a cluster outbreak at a carnival celebration in North Rhine Westphalia, Germany, less than 100km away from our offices(29). Similary to our study (80%), the most common symptom was a dry, non-productive cough (73%). As a consequence the RKI adapted their recommendation for SARS-CoV-2 diagnostics (after the end of our study) and implemented loss of taste and smell as a paramount factor into their algorithm (figure 2). It has to be noted that acute respiratory symptoms of any severity now qualify to take a smear. As a consequence the number of smears will increase. In our population this adapted RKI algorithm would have detected all five positive patients, but would have produced a much larger number of tested individuals (n=369) and significantly higher costs for local health authorities. 1.36% of tested patients would have been positive, as opposed to 6.25% following the first RKI recommendations. Additionally, due to a reliable recall system from our offices, the clinic with the smear centre, and the health department, we were able to confirm that none of the patients tested negative for SARS-CoV-2 was tested positive at a later stage during the study period. This may suggest that the more conservative first RKI algorithm would have been sufficient in our low-prevalence population and the adapted, more liberal algorithm may have led to over-testing and more costs. Our five patients tested positive were kept in quarantine for two weeks and symptoms had subsided by then. Therefore, further transmission after release is highly unlikely.
During our study initial national efforts were put into measures to make testing available to everyone, especially in rural areas, such as ours. During the study period we had to send all patients to the local smear center, which changed in May 2020, where we performed smears directly in our offices.
A more focused approach of smear-taking has been used since the advent of clustered outbreaks(29) with a large amount of positives alongside a very low number of positives, such as in our area. The debate on the idealsmear strategy is ongoing. Our data contributes essential input to this debate, because we can show that intensifying the smear strategy, as it was done by the German RKI, does not yield a better detection of positives in low-prevalence populations. Maybe algorithms have to be adapted to the federal or even local prevalence as well as to regional cluster outbreaks. A very liberal strategy also binds human resources in the GP offices, which can create new problems.
Another issue, which has not been entirely solved, is sequential testing: A median incubation period of five days was estimated(30), which contains the danger that a negative test during early infection may pretend a false safety. Sequential testing may circumvent this, but was not applicable in our study.
In Germany we have a health system based on solidarity, in which most people have health insurance and thus have easy access to health care. GP’s, as the first medical contacts, have to filter many patients directly in the office. The Center for Disease Control (CDC) has issued similar recommendations for the public as the RKI in Germany, namely (1) to cover mouth and nose with a cloth, (2) call the GP’s office first instead of showing up directly, (3) not to get in close physical contact to others and (4) engage in regular disinfection of hands and surfaces, (5) and self-monitor symptoms(31). The paramount aim of medical personnel in this crisis is to maintain optimal medical care and personal health in a high-risk environment. Thus, stringent algorithms for GP’s and other medical specialties need to be introduced to achieve this goal. By segregating patients suggestive of respiratory infection and treating them under high standards of hygiene and protection, we believe that we have achieved this goal in a practicable and efficient manner. None of our employees called in sick during the study period, which may suggest that our stringent selection process prior to entering the office is a success. Our GP experience could now help colleagues in other countries with a later onset of COVID- 19 than Germany to organise their offices with the available resources. Transparency of strategies from different countries on how to deal with COVID-19 in inpatient and ambulatory settings is of paramount importance to optimize further patient care and improve educational measures(32).
Our data shows that it is not possible to accurately differentiate between oligosymptomatic COVID-19 patients and ordinary respiratory infection by analysing symptoms alone. Recently, Arons et al. published the spread of COVID- 19 in a US nursing home, in which more than half of positively tested patients were asymptomatic(33, 34). This clearly demonstrates that strategies focusing only on symptoms fail to prevent further transmission.
There is a conceivable limitation to our study. Although we have investigated a great amount of patients, we live in a low-prevalence area, which excludes inductive statistics and allows only a description of data. The validity of both suggested algorithms from the RKI has to be analysed in a larger and more diverse cohort. Large, prospective, randomised multicenter trials will be necessary to answer the question on the most suitable test strategy.
To sum up, our study illustrates that application of the modified and more liberal RKI algorithm (figure 2) to filter patients with minor respiratory tract infections in the GP office in a low prevalence area would lead to more (negative) testing than the first and more conservative version (figure 1). Different results may be gained in high risk areas. We did not observe a conceivable difference in symptoms of COVID-19 patients and patients with ordinary respiratory tract infection. Our study also adds valuable information to the literature on how to practically manage patient care in a GP office during the peak of the COVID crisis in Germany in 2020. Additionally, we would suggest integrating working in a high risk profession (such as hospitals or nursing homes) as a factor for future risk assessment scores for COVID-19 infections. Furthermore, lung ultrasound, as an easily available tool in outpatient care, could help to better detect COVID-19, since it has been shown that B-lines, a typical sonographic sign of pneumonia, are present at an early stage of the disease(35).