2.1 Study design and population

Two hundred eighty-six patients with previously confirmed SARS-CoV-2 infection by polymerase chain reaction(PCR) or antigen testing who were diagnosed with long COVID at our clinic between May 2021 and December 2021 were potentially eligible for inclusion in this study.

Clinical data, such as age, sex, BMI, and COVID-19 onset, PCR-positive date, consultation day, and laboratory data, including aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels, of the selected patients, were collected and summarized in our database designed for this study. In addition, medical histories of diabetes mellitus, hypertension, hyperlipidemia, migraine, bronchial asthma, sleep apnea syndrome, and other conditions such as any cancer, Graves' disease, atrial fibrillation, cerebral infarction, depression, epilepsy, and so forth, as well as a history of smoking were similarly recorded based on self-reported information. For the evaluation of therapeutic efficacy, 29 censored patients were excluded from the analysis. A summary of the data collected from the study patients is presented in Table 1. The study protocol was approved by the ethics committee of UnMed Clinic Motomachi, Kanagawa, Japan (authorization number: UM22-01). The review board approved and waived the need for written informed consent from the participants due to the retrospective, noninterventional nature of this study. We have read the Declaration of Helsinki and have followed the recommended guidelines in this study.

2.2 Therapeutic interventions for long COVID-related symptoms

Because herbaceous medications are generally agents that easily ameliorate a variety of systemic symptoms, including dysautonomia, they are actively administered to most patients, depending on the manifestations.

Therapeutic options for long COVID-related symptoms in our clinic are listed in Table 2. As previously reported, hochuekkito (TJ-41, 2.5 g three times daily), juzentaihoto (TJ-48, 2.5 g three times daily), and ninjinyoeito (TJ-108, 3.0 g three times daily) were administered for fatigue, and kakkonto (TJ-1, 2.5 g three times daily) and/or acetaminophen (500 mg orally as needed; not to exceed 3000 mg/day) were administered for fever, headache/joint pain, along with some other herbal drugs.¹⁵ For smell disorder, tokishakuyakusan (TJ-23, 2.5 g three times daily) was prescribed according to the clinical and basic evidence of its effectiveness for olfactory dysfunction.^{16, 17} In addition, rikkunshito (TJ-43, 2.5 g three times daily), bukuryoingohangekobokuto (TJ-116, 2.5 g three times daily), and hangeshashinto (TJ-14, 2.5 g three times daily) with a proton pump inhibitor (esomeprazole 10 mg or vonoprazan 10 mg), as an acid suppressant, and/or a dopamine-2 receptor antagonist (metoclopramide 100 mg orally as needed) were appropriately administered for digestive manifestations, such as nausea, stomachache, abdominal fullness, diarrhea, constipation.^{18, 19} On the basis of good results of a recent clinical trial in the UK, inhaled budesonide was actively used for pulmonary symptoms including cough, sore throat, sputum, dyspnea, chest pain with kikyoto (TJ-138, 2.5 g three times daily), carbocysteine (500 mg three times daily), and ambroxol hydrochloride (15 mg three times daily).²⁰ For sleep disturbance, eszopiclone (1, 2, or 3 mg daily) and/or lemborexant (2.5, 5, or 10 mg daily), which are independent sleep aids, were palliatively introduced as needed. As a Japanese study previously successfully managed alopecia using saikokaryukotsuboreito (TJ-12, 2.5 g three times daily), this herbal drug was also selected to treat alopecia in the present study.²¹ Last, betahistine mesilate (12 mg orally as needed) and ryokeijyutsukanto (TJ-39, 2.5 g three times daily) were used to relieve dizziness symptoms. We were unable to provide appropriate care for brain fog or dysgeusia.

2.3 Outcomes

Primary outcomes were the clinical course with a futuristic overview, evaluation of the effectiveness of treatment for each manifestation, and risk factors affecting the recovery period of long COVID. Fatigue, smell disturbance, fever, headache/arthralgia, respiratory manifestations, digestive symptoms, circulatory disturbance, sleep disorder, hair loss, brain fog, and dizziness were separately assessed as distinct manifestations of long COVID in the analysis. For the validation of the recovery period, the clinical courses were divided into four groups: good (within a month or less), moderate (within 2 months), slow (within 3 months), and resistance (3 months or more). Clinical outcomes were measured at more than 3 months of follow-up, depending on the condition. The secondary outcome was the relationship between serum transaminase levels and fatigue determined by measuring blood samples from 164 selected patients.

2.4 Statistical analysis

Demographic and clinical data were extracted from electronic medical records. All statistical analyses were performed using SAS version 9.4 (SAS Institute). The χ² test was used to compare distributions of categorical variables in Tables 3–6. Significant levels were set at p < 0.05. In the comparison of improvement of complaints after treatment, Bonferroni adjustment was conducted for each test separately to consider the number of significant tests undertaken.

3.1 Patient characteristics

Two hundred eighty-six participants with long COVID completed the survey. Twenty-nine patients were excluded from the analysis of therapeutic effects because they were unable to fully follow their clinical courses. The demographic and clinical characteristics of the participants are summarized in Table 1. The median age of the enrolled patients was 35.8 (14–64) years, with 137 (47.9%) men and 149 (52.1%) women. The proportions of different BMI statuses among patients were 12.2% underweight, 66.4% normal weight, and 15.7% overweight. Sixty-four (22.4%) patients were current or former smokers, and 41 (14.3%) participants had any comorbidities, including 16 (5.6%) with respiratory diseases and 12 (4.2%) with metabolic diseases. Overall, 249 (87.1%) were outpatients with a mild illness, and 37 (12.9%) had a moderate (33 patients, 11.5%) or severe (4 patients, 1.4%) disease requiring hospitalization. The median duration from the onset of COVID-19 to initiation of treatment for long COVID was 54.4 (11–254) days. The median number of symptoms was 2.8 (1–9). The most frequent symptoms were respiratory manifestations (149/286, 52.1%), followed by fatigue (147/286, 51.4%), smell disorder (101/286, 35.3%), headache/joint pain (98/286, 34.3%), and hair loss (76/286, 26.6%). Sleep difficulties were ascertained in 30/286 (10.5%) individuals, whereas 34/286 (11.9%) patients presented cognitive disturbances, also called brain fog. Fortunately, no serious complications from our treatments were observed.

3.2 Chief complaints may change in the clinical course of long COVID

According to the χ² test, respiratory symptoms tended to be a major complaint in the initial phase, followed by fatigue, fever, and headache/joint pain, whereas hair loss was the major complaint in the late phase (Table 3). Thus, our clinicians treating patients with long COVID should understand that the chief complaint of the patients may vary over time. Additionally, our data revealed no significant relationship between COVID-19 severity and the number of symptoms of long COVID (data not shown).

3.3 Variation in therapeutic efficacy among long COVID-related symptoms

Compared with previous studies, our data showed that the recovery period of pulmonary manifestations in patients with long COVID was undoubtedly shortened if they used inhaled budesonide.^22-25 Using fatigue symptoms as the basis for comparison of the recovery period, pulmonary symptoms presented both a tendency towards early recovery and better therapeutic effect in most patients; subsequently, fever and headache/joint pain were comparatively well managed based on the rapid recovery in more patients (Figure 1). Conversely, hair loss was the worst symptom, with less quick recovery and more refractory cases, although hair loss tended to occur in the later phase. Brain fog, which has no effective therapeutic intervention, and smell disorder took second place in slow recovery (Figure 1).

3.4 Independent risk factors influencing the recovery period of long COVID

As shown in Table 4, the duration of recovery from long COVID was significantly shorter in the mild COVID-19 group than in the moderate/severe COVID-19 group. Considering both our data and previous studies showing a significant difference in the association between COVID-19 severity and long COVID outcomes,^{12, 26-28} COVID-19 severity undoubtedly exerts a negative effect that cannot be ignored. Next, we focused on the effects of the number of complaints on the recovery period. Consistent with previous studies describing that multiple symptoms at the onset of COVID-19 were associated with long COVID occurrence,^{9, 14, 29} our results also revealed that the simple complaint group showed significantly faster improvement from long COVID compared with the multiple complaint group. An undeniable relationship between the number of manifestations and long COVID was observed. Moreover, earlier intervention for long COVID-related symptoms potentially led to more rapid recovery (Table 4). Thus, the sooner patients start treatment, the better the chances of recovery from long COVID.

3.5 Improvement of long COVID may be delayed by smoking

Strikingly, a smoking habit was an independent risk factor for slowing the recovery period from long COVID, whereas BMI, sex, age, and comorbidities were not (Table 5). Since smoking is a well-known major factor exacerbating COVID-19,^30-32 our data support the medical rationale for promoting smoking cessation to prevent not only COVID-19 aggravation but also the delay in long COVID recovery.

3.6 Increased serum transaminase levels may be a surrogate marker of “fatigue”

Finally, we assessed the relationship between serum transaminase levels and fatigue symptoms in 164 patients from whom blood samples were selectively collected in case of blood test consent had been given. It is generally accepted that the liver is a major organ in the pathogenesis of fatigue because it regulates much of the storage, release, and production of substrate for energy generation,³³ and it has been reported that ALT activity may be important not only as a marker of liver diseases but also as an indicator of general health.³⁴ According to Specified Medical Checkups and Specified Health Guidance in Japan, ALT > 50 is a judgment value of recommendation for medical examination. Therefore, we selected the cutoff points of 50 and 75 IU/L from a clinical perspective. Interestingly, the correlation between fatigue symptoms and serum ALT levels was shown using a χ² test, although AST levels showed no significant difference (Table 6). Because liver damage may occur during COVID-19 progression, regardless of pre-existing liver disease, our finding is consistent with that of a previous study.³⁵ Namely, increased serum transaminase levels, especially ALT, may provide a convenient indication of fatigue.