A longitudinal study on symptom duration and 60-day clinical course in non-hospitalised COVID-19 cases in Berlin, Germany, March to May, 2020

Background Detailed information on symptom duration and temporal course of patients with mild COVID-19 was scarce at the beginning of the COVID-19 pandemic. Aim We aimed to determine the longitudinal course of clinical symptoms in non-hospitalised COVID-19 patients in Berlin, Germany. Methods Between March and May 2020, 102 confirmed COVID-19 cases in home isolation notified in Berlin, Germany, were sampled using total population sampling. Data on 25 symptoms were collected during telephone consultations (a maximum of four consultations) with each patient. We collected information on prevalence and duration of symptoms for each day of the first 2 weeks after symptom onset and for day 30 and 60 after symptom onset. Results Median age was 35 years (range 18–74), 57% (58/102) were female, and 37% (38/102) reported having comorbidities. During the first 2 weeks, most common symptoms were malaise (94%, 92/98), headache (71%, 70/98), and rhinitis (69%, 68/98). Malaise was present for a median of 11 days (IQR 7–14 days) with 35% (34/98) of cases still reporting malaise on day 14. Headache and muscle pain mostly occurred during the first week, whereas dysosmia and dysgeusia mostly occurred during the second week. Symptoms persisted in 41% (39/95) and 20% (18/88) of patients on day 30 and 60, respectively. Conclusion Our study shows that a significant proportion of non-hospitalised COVID-19 cases endured symptoms for at least 2 months. Further research is needed to assess the frequency of long-term adverse health effects in non-hospitalised COVID-19 patients.


Background
The coronavirus disease  pandemic caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in Wuhan, China in December 2019 and has spread across the globe. As infections peaked across many countries, more has become known about the epidemiological features of the pandemic, the characteristics of those infected and the risk factors associated with hospital or intensive care unit (ICU) admissions and deaths [1][2][3].
However, detailed investigations on the temporal course of COVID-19 symptoms during the early stages of the pandemic have mostly focused on hospitalised patients or patients admitted to ICU even though the vast majority of infected patients do not require inpatient treatment. The characteristics and duration of symptoms of these patients, who are often referred to as having a mild course of disease and who mostly stay in home isolation or in community health centres during the acute phase of disease, have more recently become the subject of investigation [4][5][6][7][8][9]. These recent findings provide evidence that a subset of patients, including those with a mild course of disease, report persistent symptoms that result in long-term adverse health effects [4][5][6][7][8][9].
The aim of this study was to investigate the prevalence, duration and temporal dynamics of clinical symptoms in non-hospitalised COVID-19 patients to better understand the clinical spectrum and disease duration among this group of COVID-19 patients at the onset of the pandemic.

Study population
The target population for this study was non-hospitalised adults with confirmed SARS-CoV-2 infection who were consecutively notified to the local health authority (LHA) in Mitte, a central district of Berlin, Germany. Additional information on the study area and setting is presented in Supplement S1.
The study period was between 30 March and 9 April 2020 and between 13 April and 20 May 2020; the interruption was due to a public holiday. Inclusion criteria were defined as follows: (i) PCR-confirmed SARS-CoV-2 infection notified to the LHA of Berlin, district Mitte; (ii) case notification to the LHA not later than day 9 after symptom onset (SO) (to reduce recall bias); (iii) patient age at least 18 years; (iv) not hospitalised for COVID-19 infection during the first 14 days after SO; and (v) reachable by telephone (to enable participation). Case verification for eligibility and inclusion of participants occurred prospectively on a daily basis and was based on the official LHA notification registry. In total, 130 cases met the inclusion criteria. Of these, 28 cases had to be excluded for reasons such as not consenting to participate, dementia, or speaking insufficient German or English, leaving 102 cases included in the study. The sampling process is further described in Supplement S2. During the study period, SARS-CoV-2 testing was recommended for individuals either experiencing new symptoms of any kind who had an epidemiological link to a PCR-confirmed SARS-CoV-2 case or experiencing symptoms considered compatible with COVID-19 at the time. Testing of asymptomatic individuals was generally discouraged.

Data collection
A structured questionnaire on the prevalence and temporal variation of 25 symptoms was developed by the authors based on the available literature at the onset of the pandemic (Supplement S3). Data collection was performed by repeated telephone consultation by clinically-experienced physicians, who were also epidemiologists. For each participant, all consultations were scheduled individually based on the participant's symptom onset (SO) date: • The first call was made between day 7 and 10 after SO to assess retrospectively the prevalence of symptoms for each day from day 1 to the day of the call.
• The second call was made between day 14 and 18 to assess retrospectively the prevalence of symptoms for each day after the first call up to days 14 through 18.
• The third call was made between day 28 and 35 to assess the prevalence of symptoms one month after SO (day 30).
• The fourth call was made between day 58 and 70 to assess the prevalence of symptoms 2 months after SO (day 60).
As we assumed that symptoms did not recur once resolved on day 30, follow-up consultations on day 60 after SO (the fourth call) were only made for participants reporting persistent symptoms on day 30 after SO (during the third call).
Definitions and lay explanations of symptoms were agreed upon by the principal investigators (Supplement S4). The perception of disease severity was assessed by collecting data on severity of malaise (mild, moderate, severe). Malaise was defined as a general feeling of illness. On day 60, we asked the cases whether their physical performance capacity had returned to their pre-SARS-CoV-2 levels and whether they had consulted a medical doctor for persisting symptoms since day 30. Additional data on sex, comorbidities, smoking, weight, height and healthcare worker status were collected.

Data quality assurance and statistical analysis
A standardised data entry mask including data checks for completeness and limits was configured in EpiData v3.1 [10]. Data quality was assured by duplicate entry and validation by the study team. Anonymity of study participants was ensured by restricting personal

Ethical statement
This study was conducted by the LHA Berlin (district Mitte) within the framework of the German Protection against Infection Act and in response to the COVID-19 We assumed resolved malaise did not recur in the long term. Therefore, patients with resolved malaise on day 30 were assumed to be malaise-free on day 60.
Study population size: n = 98 between day 1 and 14, n = 95 at day 30, and n = 88 at day 60.
pandemic [11]. In Germany, local health authorities are obliged to collect information on clinical symptoms of notifiable diseases for routine surveillance. Thus, institutional review was not sought. No biological materials were collected. Participation was voluntary and verbal informed consent was given before the start of the interviews. Verbal informed consent included informing the participants about the purpose and extent of the study, that participation was voluntary, that their anonymity would be ensured, and that they could terminate their participation or retract collected personal data at any time. Support from the national public health institute -i.e., the Robert Koch Institute (RKI)to the LHA was provided after official request.

Study population
Of the 102 study participants, four cases had an asymptomatic disease course (follow-up stopped 14 days after notification We also found that 51% (50/98) of the cases studied consumed antipyretics during the first 2 weeks for a median of 3 days (IQR: 1-6 days). We observed some differences between sexes: dysgeusia (54% vs 34%, p=0.075) and nausea (39% vs 12%, p = 0.008) were more common among females than males (Supplement S5).
In total, 41% (39/95) still experienced at least one symptom on day 30, with dysosmia and cough being the most common symptoms persisting 1 month after SO. On day 60, 20% (18/88) of the cases were still symptomatic, with dysosmia being the most commonly reported symptom (14%, 12/88) ( Table 2). Moreover, when cases experienced symptoms at day 30 or day 60, nearly all had already experienced the symptoms at least once after SO (Supplement S6).
In addition to the symptoms described above, 8.0% (7/88) of all cases reported having reduced physical performance capacity on day 60 compared to their pre-COVID-19 physical performance capacity. Furthermore, 10% (9/88) of cases reported having consulted medical advice because of persisting or worsening symptoms between day 30 and 60. None of these individuals required hospitalisation during that period.

Temporal variation of malaise and other symptoms
The cases' subjective assessment of feeling ill was reflected by the perception and severity of malaise in the 2 weeks after SO. In the first week of illness, approximately half of the cases reported severe or moderate malaise on a daily basis. After the first week, the number of cases having severe and moderate malaise decreased. However, 52% (50/96) needed more than 11 days for their malaise to subside. Two weeks after SO, 35% (33/95) of the patients still reported some, predominately mild, malaise. Feeling ill lingered in a minority of cases; 7.4% (7/95) and 3.4% (3/88) of them still reported feeling ill on day 30 and 60 after SO, respectively ( Figure 1).
As previously noted, the majority of cases did not experience all the symptoms associated with COVID-19. Moreover, the timing of when symptoms were present varied: rhinitis or cough seemed to occur throughout the 2 weeks after SO. However, the peak of cases experienced rhinitis or cough on day 5 or day 6 after SO, whereas the peak of dysosmia and dysgeusia was between day 7 and day 9. Dyspnoea was not commonly prevalent and occurred mostly between day 2 and day 11. Other symptoms such as chest tightness, chest pain or pleuritic pain occurred rarely in the 2 weeks after SO. Fever and headache, on the other hand, seemed to be present mostly in the first week; however, relatively few individuals experienced fever (Figure 2).

Diversity and duration of symptoms in the first 2 weeks of illness
The clinical presentation of COVID-19 among the study population was diverse; more than half (54%, 53/98) of those symptomatic in the 2 weeks after SO reported at least eight different symptoms. We found that 65% (64/98) reported having symptoms from at least four different symptom groups in the 2 weeks after SO.
COVID-19 symptoms persisted for the entire 2 weeks after SO for the majority of cases; 59% (58/98) experienced at least one symptom every day for 2 weeks. In first 2 weeks after SO, 7% (7/98) had an intermittent clinical course (i.e., a symptom-free period interrupted by a symptomatic period), and 34% (33/98) became and remained symptom-free after an initial symptomatic period. For these cases, only 7 of 33 recovered within one week after SO. The majority of patients (61%, 20/33) needed more than 11 days to become symptom-free.

Discussion
This study focused on duration and temporal variation of symptoms in non-hospitalised COVID-19 cases in Berlin, Germany. The most common symptoms were malaise, headache, and rhinitis. One month after SO 41% of the cases reported symptoms and 2 months after SO 20% of the cases reported symptoms.
The detailed analysis of the duration of symptoms in the 2 weeks after SO showed that malaise, dysosmia, dysgeusia and cough were present the longest (median duration 8-11 days). However, because the daily symptom assessment during the first two-week period was right-censored at day 14, the symptom duration might be underestimated.
As with other investigators [15,16], we observed a wide clinical variety of symptoms: 53% of cases complained of more than eight different symptoms in the 2 weeks after SO. As a consequence, patients often presented with an unspecific clinical picture where gastrointestinal, respiratory and neurological symptoms overlapped. These findings underline the difficulty in formulating a sufficiently sensitive but workable clinical case definition. Furthermore, the diverse clinical presentation strengthens the importance of laboratory confirmation to establish the diagnosis of SARS-CoV-2 infection, both for clinical management as well as for public health action intended to contain further spread of disease [17,18].
From a public health point of view, prolonged symptoms among COVID-19 cases might create uncertainty about when to conclude home isolation [19]. This might lead to an increased workload for public health authorities as well as frustration and possibly negative psychological effects for infected individuals. Our findings emphasise that having mild COVID-19 can have relevant health implications for at least 2 months after disease onset in younger adults, who have a low prevalence of pre-existing conditions. Although longitudinal studies with longer follow-up are under way to investigate the long-term health effects following COVID-19, our findings can serve as early guidance for physicians treating COVID-19 outside a hospital setting.
We describe in detail the duration of specific symptoms including a 30-and 60-day follow-up in nonhospitalised COVID-19 patients in the early phase of the COVID-19 pandemic. Symptoms most commonly reported after 2 months were malaise, dysosmia, dysgeusia, cough and dyspnoea, most of which were also more frequent from the onset. Gastrointestinal symptoms did not appear to persist over a longer period. As this study did not include questions on cognitive functions, possible detrimental COVID-19 effects on cognitive health cannot be excluded and have recently been reported by other studies [20,21]. The findings on symptom persistence 2 months after SO, including reduced physical performance capacity for a substantial subset of patients, are concerning as they add to a growing body of evidence that acute COVID-19 can result in prolonged illness even among individuals with mild disease, which has been confirmed over time [22][23][24][25]. Our results underscore the need for studies with

Figure 3
Duration of specific symptoms in the 2 weeks after symptom onset in non-hospitalised COVID- extended follow-up to assess the long-term effects of COVID-19.
Current investigations mostly assess the prevalence and duration of symptoms of COVID-19 based on populations sampled in hospitals or care facilities. In comparison, our study population was sampled in direct collaboration with the LHA of the study area and based on the LHA's official notification registry. As the German Protection against Infection Act requires that both direct care providers and diagnostic laboratories notify the LHA of the residence of every suspected and PCR-confirmed case of COVID-19 within 24 hours [11], the total target population should be represented by the study population. Second, the study was performed in an urban setting where free testing had been offered since the early phase of the pandemic, and several testing sites were actively promoted (e.g., hospitals, outpatient departments, drive-by testing centres, and general practitioners' offices) [26]. Therefore, we assume a low level of, if any, physical or financial barriers towards healthcare seeking and diagnosis for residents in the district.
Our study participants were generally younger and healthier than cohorts from other studies. As for older age, comorbidities such as hypertension, pre-existing cardiac conditions, chronic obstructive pulmonary disease, and diabetes are confirmed risk factors for a more severe disease course and hospitalisation [12,13].
Our study has some limitations. Despite low physical and financial barriers towards SARS-CoV-2 testing in the study district, we cannot entirely exclude that psychological or cultural barriers might have influenced the acceptance of free diagnostics and therefore the representativeness of the study population. Furthermore, the composition of our study population might be influenced by the recommended testing strategy for SARS-CoV-2 infection. With the exemption of outbreak investigations and without an epidemiological link, testing of asymptomatic patients with atypical COVID-19 symptoms was not recommended in Germany at the time. Therefore, we assume that individuals with very mild or asymptomatic disease are underrepresented. Moreover, because our participants were asked about their symptoms retrospectively by trained physicians, we cannot exclude the possibility that recall and attention bias influenced our findings -e.g., participants might remember their symptoms better because they are being interviewed. However, we believe that our results resemble those of a thorough and systematic history taken by a physician in routine clinical settings.

Conclusion
Our findings illustrate that non-hospitalised cases with a mild course of COVID-19 experience a multitude of symptoms over a prolonged period. These patients might not be able to fully engage in working activities for a long period and/or might need assistance and care from their relatives, possibly resulting in negative economic effects to the individual as well as to society as a whole. Further research is needed to assess the frequency and prevention of long-term adverse health effects in these patients. Finally, the clinical presentation of non-hospitalised COVID-19 cases from our study partly differs from symptoms used for case-finding activities in the early phase of the pandemic. With the attempt to capture as many cases as possible to guide decision-making, symptoms in case definitions might need to be adapted to better cover COVID-19 outpatients.