A membership group called COVID-19 Smell and Taste Loss was established on March 24, 2020, by the charity organization AbScent, which is based in England, United Kingdom. Polls were occasionally conducted to determine various aspects of the members, for example, hospitalization for COVID-19. The information on the age, sex, and the countries and cities where the members were located was obtained through the software mechanism called Group Insights. The reports posted were recorded daily as they were in Excel (Microsoft Corporation) spreadsheets with the records of the dates posted and the names, COVID-19 test results if available, a short summary of the symptoms, starting dates of anosmia and ageusia symptoms if available, and recovery dates from these symptoms if available. When the same person posted on a separate day, the posts were recorded under the name of the person with the records of the dates they were posted. Recording of the posts and comments for this study lasted from March 24 to July 1, 2020. Posts and comments without specific content related to their own symptoms were not recorded (eg, expressing their sympathy, excitement, or approval saying “Yes,” “congratulations!” “Thank you,” or “nobody knows”).

The study is approved as exempt (Institutional Review Board protocol number 10082020-1, Exempt Category #3)

Because, in this group, members freely reported their cases, the level of detail for each member was not consistent. We classified the reports by whether the patient had other symptoms than anosmia and ageusia, and whether they recovered their senses, and then selected several reports that described their symptoms in detail.

Fisher exact test was used to statistically evaluate the recovery rate among groups.

Figure 1 shows the time course since the pandemic started. On March 11, 2020, the WHO declared the rapidly spreading SARS-CoV-2 virus a pandemic and warned that it might spread to many countries around the world (Figure 1A). The disease caused by the virus, COVID-19, was first considered to cause fever and coughs as the main symptoms. However, soon it was discovered that there were more symptoms, including the loss of the senses of smell (anosmia) and taste (ageusia). The membership group on social media studied in this project was founded on March 24 (Figure 1B), and the number of members reached about 1000 during the first week (Figure 1). This immediate increase of the members indicates the rapid spread of the disease and the large number of people who suddenly lost their senses of smell and taste at the early stage of the pandemic. The increase in the number of members showed another surge around the time when the second wave in the number of cases started in the autumn of 2020. This may be due to both the increased publicity after the introduction of the group in the media (Figure 1C) and the large number of people with COVID-19–induced anosmia and ageusia. This suggests the existence of a large number of patients who lost their senses during this pandemic.

What were considered the symptoms of COVID-19 have changed from the earlier stage of the pandemic in spring 2020 to summer and fall 2020. Figure 2 shows the outdoor and indoor signs made by the CDC in March, June, and July [20]. In the sign generated in March (Figure 2A), only the symptoms of fever, coughs, and shortness of breath are described, whereas in the June and July signs, the variety of symptoms has increased by more than double, and now it is well known that not every COVID-19 positive patient has fever, coughs, or shortness of breath (Figure 2B and C). In the poll asking if they got the COVID-19 tests immediately after the onset of symptoms, about 70% (n=259) of the 705 people answered that they could not get tested at all or could not get tested immediately and received the antibody tests later (Figure 3A; could not get tested: n=259, 36.74%; antibody test later: n=247, 35.04%). Analyses of the people who added comments (n=54) to the poll revealed that most of the patients with COVID-19 who could get tested immediately after the onset of symptoms was due to their jobs (essential workers, mostly) or those who contracted the virus after June when the criteria to get tested expanded (compare Figure 2A and 2B; Figure 3B bottom; job: n=12, 22.22%; after June: n=26, 48.15%). The comments by the people who replied that they could not get tested immediately and received antibody tests later or could not get any test revealed that they were mostly the people who had an onset of symptoms before June (Figure 3B, top two bars).

Thus during the early stage of the pandemic, there were many people who got sick and self-diagnosed themselves with COVID-19 (suspected COVID-19 patient). This inability to get a COVID-19 test immediately after the onset of symptoms also resulted in the inability to get immediate medical treatments because medicines were not prescribed without an official diagnosis. We asked a question about the way they spent time after the onset of symptoms and found that about 90% (n=204,) of the 221 members who replied either stayed at home with generic medicines and supplements (n=162, 73.73%) or stayed at home without any medicines (n=40, 18.10%; Figure 4). Only 6.79% (n=15) replied that they stayed at home with medicines prescribed by doctors. Although there were some responses showing that there were members who experienced hospitalization in a separate poll (see Figure 3), none of them responded to the poll on the medical treatments, and thus, the option of medical treatment following hospitalization is not shown in Figure 4. Although there is some missing information on the hospitalized people in the results, Figure 4 clearly shows the invisible nature of the patients with mild to moderate symptoms, which also suggests that the number of cases worldwide in the countries where availability of COVID-19 tests was limited or had criteria to get tested is highly likely to be far more than the officially reported numbers.

The self-isolation situation, staying at home, itself can cause anxiety. With symptoms of suspected COVID-19 and without a COVID-19 test and medical treatment, it is not hard to imagine that this situation causes high anxiety. It is possible during self isolation to obtain information from the internet and, if available, to join a patient group to exchange information. We asked the members the reasons that they joined the group. We found that the top three reasons were to obtain the information about the mechanisms of sense loss (n=401), information on how to recover the senses (n=355), and mental support (n=205), which accounted for 86.65% (n=961) of the 1109 responses (Figure 5). The sum of the responses related to information (5 out of 9 options; n=823, 74.21%) and mental support of themselves or their family (2 out of 9 options; n=218, 19.66%) accounted for over 93% of the reasons to join the group. If we consider that the two other options, networking and checking how others are doing, are relatively similar to mental support, it is possible to say that obtaining information and mental support are the two major reasons for joining the group.

Demographic data for the registered members (n=9227 as of October 12, 2020) of the social media site is shown in Figure 6. The number of women in the group (n=6995, 75.38%) was about three times more than men in the group (n=2272, 24.62%). This number included the members who were administrators of the group or scientists, and not patients, although the estimated percentage of these nonpatient members in the group was minimal. There was a peak in the number of members in the age range between 25-34 years in both men and women (Figure 6). This high number in the rather young ages could be due to the use of social media, although there are multiple other reports on patients with COVID-19 with anosmia and ageusia, which are not using social media, that show the same tendency [6]. On the other hand, in other membership groups of AbScent on social media, the average age is higher, which also suggests that this is specific to COVID-19 and not due to the use of social media. We do not know yet if this is due to the tendency that older-aged people develop severe conditions when they contract COVID-19 and are hospitalized or reach a severe condition such that they are unable to use the internet, or if the symptoms of the senses are specific to patients of younger ages.

Figure 7 shows the top 20 countries of the members as of October 12, 2020. Where the patients were located may have had some influence on the symptoms, considering the mutation of the reported virus [21,22]. There were a total of 99 countries that had members in the group. The fact that it was possible to join the group from anywhere the internet was available, if there was no federal restriction on joining a social media group, was one of the benefits for suspected patients staying at home without medical treatments. The top 2 countries, however, were the United Kingdom and the United States, comprising 75.7% (United Kingdom: n=3744; United States: n=2809; total of the 20 countries: n=8654 members) of the number of members of the top 20 countries. This suggests that it is easy for the people in the countries where English is used to join the social media group. This may explain the names of the third and fourth top countries as well. The Philippines and South Africa are the third and fourth countries, respectively, in the number of members, and English is one of the official languages in these countries. There were rather few members from the countries in Europe even though the pandemic was severe there (eg, Italy: n=68, 0.79%), which could be due to the language barrier.

In December 2019, when the first reports on the outbreak started in Wuhan, China, the major symptoms reported were shortness of breath, cough, fever, and diarrhea. However, it has become clear that many patients who contracted the virus are asymptomatic yet contagious [3-5]. In addition, the symptoms of the symptomatic patients are more varied than first reported (Figure 2). Many patients reported symptoms that suggested encephalitis (headaches, foggy brain, memory loss, hallucinations, pain in the upper end of the nose suggesting inflammation in the olfactory bulb, ear pain, and pain in the back of eyes; Table 1). In addition, some patients also reported a skin rash and tingling in the legs (Table 1), which also has been reported by other recent publications [23].

During the time period that the posts from the members were recorded (March to the end of June 2020), there were 355 members who posted reports of their smell or taste dysfunction and other symptoms they had. Of these 355 members, the percentage of patients without other symptoms comprised 14.93% (n=53), and 85.07% (n=302) of them had other symptoms than the loss of their senses. Almost 10% had recovered their senses by the time this report was summarized (recovered vs not recovered: n=34, 9.58% and n=321, 90.42%). The percentage of patients who recovered their senses was higher in the patients who did not have other symptoms besides anosmia and ageusia (recovered with other symptoms vs without other symptoms: 25/302, 8.28% and 9/53, 16.98%).

There were 123 members who reported that they had other symptoms and reported the timing of the onset of the loss of senses and other symptoms. Of the 123 members, 112 (91.06%) members reported that other symptoms started first. The average difference of the days between the onset of other symptoms and the onset of the loss of senses was 4.2 days. There were 6 (4.9%) members who reported that the onset of the loss of senses occurred on the day other symptoms started, and 5 (4.07%) members who reported that the loss of senses occurred before other symptoms started. The average difference of the days between the onset of the loss of senses and other symptoms was 1.6 days when the loss of senses preceded the onset of other symptoms.

The duration of time until recovery varied largely, with some people regaining their senses within a week and some people who lost their senses at the beginning of the pandemic and still had not recovered their senses at the end of June 2020. As we found the tendency that patients without other symptoms had a rather higher recovery rate, we examined if these differences correlated with the duration to recover their senses as well. Table 2 shows the 34 patients who recovered their senses in the order of the days until recovery. When the 34 patients were divided into the top 50% and the bottom 50% (fast recovery group and slow recovery group; n=17 for both), the average days to recover was 12.9 days for the fast recovery group and 41.3 days for the slow recovery group. In the fast recovery group (n=17), there were 8 (61.54%) patients without other symptoms, whereas in the slow recovery group (n=17), there was only 1 (7.69%) patient who did not have other symptoms, and there was a statistically significant difference in the duration to recovery (Fisher exact test was used: P=.02; Figure 8). These results indicate that patients without other symptoms than the loss of the senses had the tendency to recover earlier. This, however, did not indicate that patients without other symptoms will always recover their senses faster. Of the 321 patients who reported they had not recovered their senses by the time we summarized this report, there were 44 (13.71%) patients who reported they did not have other symptoms. This suggests that there may be several causations in the loss of senses (ie, one of these could be the severity of the overall COVID-19–related symptoms and some other factors could also be involved, affecting the recovery process).

Patients with anosmia and ageusia occasionally reported distorted smell and taste (parosmia and dysgeusia) as well as the smell and taste without the sources of the smell or taste (phantosmia and phantogeusia). Table 3 summarizes the reports from the patients. Although there are various ways to describe the nature of phantosmia and phantogeusia, a smell and taste of smoke and burnt material was most frequently reported.

Here, we show several examples of reports from the patients. In selecting the examples, we first classified the 355 members who posted reports of their smell or taste dysfunction during March and June 2020 into the patients who recovered (fast recovery group and slow recovery group) and had not recovered, and those with or without other symptoms than anosmia and ageusia.

Case 1 represents a fast recovery group patient who recovered early and did not have other symptoms. The length of time until recovery of smell and taste was 9 days. This patient was not tested for COVID-19. The dates shown in Textbox 1 were added when they were recorded in Excel.

Here is another example from the fast recovery group, but the patient had other symptoms. The patient reported recovery of the senses in 13 days (however, at the time point of summarizing this report, we have heard that this patient started to experience parosmia starting from about 2 months after the recovery of the senses). The dates shown in Textbox 2 were added when they were recorded in Excel.

The next example is a patient of the slow recovery group who recovered after 42 days and had other symptoms. The dates shown in Textbox 3 were added when they were recorded in Excel.

The last example is of a patient who has not recovered their senses yet. The dates shown in Textbox 4 were added when they were recorded in Excel.

The year 2020 has become a tragic year with the pandemic that has over 67 million cases and over 1.54 million deaths worldwide as of December 6, 2020. At the early stage of the pandemic, many suspected COVID-19 patients could not get tested because of the overwhelming number of patients with severe conditions at the hospitals. This suggests that the actual number of patients with COVID-19 could be far more than the officially reported number. Many of the suspected COVID-19 patients with mild to moderate symptoms self-isolated themselves without medical treatment. These suspected COVID-19 patients had to depend on the internet to obtain information, and many of them joined membership groups on social media, which were founded to support patients with COVID-19.

In this study, we analyzed the information in the social media database of suspected COVID-19 patients, with particular attention to their olfactory and gustatory dysfunction. About 70% of them could not get tested at all or could not get tested immediately and received the antibody tests later. Furthermore, about 90% of them either stayed at home with generic medicines or supplements, or stayed at home without any medicines. These results suggest the limited access of the patients with mild to moderate symptoms to medical services. In the aspect of olfactory and gustatory dysfunction, it was demonstrated that patients who were asymptomatic other than the senses comprised about 15% of patients. Recovery of the senses was higher in the patients who were asymptomatic besides having anosmia and ageusia. Most patients experienced other symptoms first and then lost their senses, on average, 4.2 days later. Patients without other symptoms tended to recover earlier (P=.02). Patients with anosmia and ageusia occasionally reported distorted smell and taste (parosmia and dysgeusia) and experiencing or perceiving the smell and taste without the sources of the smell or taste (phantosmia and phantogeusia). Olfactory dysfunction of a large portion of patients with COVID-19 persisted at least several weeks to months.

Through the analyses of the reports, we have noticed the benefits and limits of using social media (Textbox 5). There are many benefits for patients, which suggest that these membership groups can be a beneficial system for patients to provide information and mental support. For researchers, there are some limits that require improvements to use social media membership groups to enhance understanding of the COVID-19 symptoms and conditions of the patients. These limits are due to the free nature of the involvement in the activities of the group, which causes difficulties in following up with each patient, and control of the conditions. The reports are based on each patient’s subjective experience that the same symptom at similar severity could be experienced as a difficult experience by one person and as an easy experience by another. Some modifications in the way of organizing these membership groups may be able to reduce these limits for the researchers in the future. Demographic data revealed that there were about three times more women than men, and the range of age showed a peak in the age group of 25-34 years. The number of members in each age group decreased in the groups of higher age. These results on the demographic trends may reflect the trends that the symptoms of the younger people are milder compared to older people, and thus, they stayed home without medical treatments. However, there are also possibilities that women used social media more than men and that younger people used social media more than older people to obtain information on COVID-19. There were, in fact, a few cases that wives who joined the group instead of the husbands with suspected COVID-19 and mothers who joined the group instead of their small children who were suspected COVID-19 patients. These are the possible aspects of the social site structures that can skew the information.

Among the symptoms, we especially focused on smell and taste dysfunction, which is now known as one of the major biomarkers of COVID-19. The use of social media allowed members to join the group from around the world. Although there is a limit in the range of information that can be obtained through social media (eg, accurate diagnosis of the patients), the use of it enabled us to obtain the symptoms and the progression of these symptoms from the patients who did not or could not visit the hospitals.

The current analysis demonstrated that about 90% of the 355 patients in the COVID-19 Smell and Taste Loss group that reported their subjective smell or taste disturbance did not recover, at least completely, for months. These results are in sharp contrast to the previously reported findings that 80% of smell and taste dysfunction in patients with COVID-19 is recovered within a few weeks [9,24-26]. This discrepancy may be due to several reasons: the differences in the patient populations (ie, the patients in the previous studies were mostly in-patients whereas over 99% of the patients of this report had never been hospitalized and most of them did not or could not have access to medical treatments at a hospital), this is a self-selected group who joined because of their smell loss, and one limit of social media, which is the inability to confirm the recovery unless the patients post their recovery to the group.

The pathophysiology of human olfactory dysfunction is divided into three major categories. One is an airflow problem, where the airflow toward the olfactory cleft is blocked by mucosal swelling or mucus hypersecretion [27-29]. The second is a sensorineural problem, where the perception of odorants is disturbed due to damage to the olfactory mucosa, such as olfactory neural degeneration, downregulation of the odorant molecules on the olfactory neurons, or impaired transport of the odorant molecules to the olfactory receptors due to secretory dysfunction of Bowman glands [27-29]. The third is a central nervous problem, in which the olfactory pathway from the olfactory bulb to the olfactory cortex is damaged [27-29]. Airflow problems could be resolved in a relatively short time as mucosal swelling by viral infection decreases in several days to weeks, and usually the smell problem could fluctuate in its severity during the time course, depending on the nasal condition. In contrast, sensorineural and central nervous problems are usually severe, do not fluctuate, and take a longer time (months to years) for recovery. Postviral olfactory disorder, an olfactory problem developing after an upper respiratory viral infection, is usually considered as a combination of a sensorineural and central nervous system problem [27-29]. In contrast to such current concepts, previous studies reported that a large portion of patients with COVID-19 recovered from their olfactory dysfunction within a few weeks [9,24-26], suggesting that the problem was due to an airflow problem. In fact, a case report of olfactory dysfunction in a patient with COVID-19 showed mucosal swelling in a restricted area of the olfactory cleft on magnetic resonance images [30]. However, our analysis, as well as some more recent studies [31], demonstrated that a larger number of those with olfactory dysfunction were not resolved in such short time frames, suggesting that a neural problem could be also involved in COVID-19–induced olfactory dysfunction. Previous reports suggest that the virus may migrate along the peripheral and central nervous system. Studies using animal models have shown that human coronaviruses can make axonal transports enabling neuron-to-neuron propagation [32]. Development of parosmia, a sign of neural olfactory dysfunction, supports this hypothesis.

Our analysis also demonstrated that the smell problem of patients with COVID-19 tends to recover sooner when the patients only have smell dysfunction, not multiple COVID-19 symptoms. The reason for this finding is unclear, but one possible explanation is that, for the outpatients without access to medical treatments, recovery from olfactory dysfunction is faster if the patients have a lower viral load and therefore fewer symptoms. In other words, olfactory dysfunction may be one of the most sensitive indicators of SARS-CoV-2 infection. In fact, the percentage of patients showing anosmia or ageusia was especially high in the asymptomatic to mild level patients [9]. This indicates that anosmia and ageusia are several of the important symptoms to diagnose COVID-19 in the early stages, and to detect patients who are asymptomatic or have mild symptoms [12,15]. The official classification of the levels of illness is based on the ability to breathe, chest pain, and ability to talk or move [1], although often the ability to breathe becomes the major criteria. The level called mild refers to the lack of these signs of severity, although often the pains and severity of symptoms in patients classified to have mild or moderate symptoms are not mild or moderate (Table 1). Some patients with mild symptoms have fatigue and pain in various parts of the body for a prolonged length of time [33]. This indicates that patients with COVID-19 need to be diagnosed as soon as possible to enable immediate treatment not only to avoid the symptoms turning into a severe condition but also to avoid the prolonged time of these symptoms.

As described in this study, social media can become an important platform to collect data from outpatients, especially in the era of infectious disease pandemics, when hospitals are overwhelmed with patients that have severe conditions and patients with mild or moderate symptoms are often overlooked. Following the outbreak of COVID-19, virtual visits to doctors have started to become more popular, which can provide medical service without in-person visits. The benefits of social media are that patients can post their symptoms and questions whenever they are in need without appointments and can receive responses from the public, either patients, nurses, or doctors, through the internet, which enables real-time monitoring of the infected condition, patient symptoms, and the clinical course. In the United States, there is a project called DETECT, which uses a smartphone and a wristband to sense changes in the heart rate, which can be an indicator of changes in health conditions [34]. This device also allows the movement of people to be tracked, providing epidemiologically useful data. However, DETECT cannot distinguish between changes due to COVID-19 or other causes. The use of social media in addition to these other media (ie, virtual doctor visits and portable tracking devices) can become a strong monitoring method in the era of infectious disease pandemics. What is necessary to further strengthen the merits of social media could be the establishment of a system that serves as a route from the social media to, for example, virtual doctors or clinics nearby. At the current stage, obtaining enough information to provide diagnosis and treatments through social media sites contains challenging aspects for medical professionals. A system that guides patients from social media to the adequate medical support they need, to counselling, to other types of support systems, and the establishment of the rules in detail to organize such systems will help social media to become one of the new tools for the medical system in the near future.

Our analysis of a social media database of COVID-19 patients’ posts demonstrated that olfactory dysfunction of a large portion of patients with COVID-19 persisted at least several weeks to months. This information is important in understanding the pathophysiology of COVID-19–induced anosmia. Although accurate diagnosis of patients is not always obtained with social media–based analysis, it may be a useful tool to collect a large amount of data on symptoms and the clinical course of worldwide, rapidly growing, infectious diseases.

Our study explores the benefits and limits of using social media membership groups in studying the symptoms of COVID-19. The patients are suspected COVID-19 patients and not all of them have been tested for COVID-19, but all have experienced loss of the senses of smell and taste.

This is the first study to determine the benefits and limits of using social media in studying the symptoms of COVID-19.

Social media membership groups can function as a source of information and mental support for patients with COVID-19. Although some improvements in the system are necessary for its use for research purposes, it is a rare source of a large amount of data for researchers as well.