Study selection

As shown in Fig. 1, a total of 780 relevant literatures were found in the seven databases mentioned above in the process of literature retrieval and screening. After removing duplicate articles, 437 papers remained. Then, 357 publications were excluded as most of them did not meet the inclusion criteria. After browsing and checking, the remaining 80 articles were screened again according to the research direction.80 articles were further excluded: 5 reviews; 46 theoretical research projects; 2 systematic reviews; 12 experimental research projects. Finally, 15 studies were assessed to be eligible and included in our review.

Fig. 1

Flow diagram of study selection process

Study characteristics

Table 1 showed the basic characteristics of the finally included research objects. Three of the articles were in English and the rest were in Chinese. There were a total of 10390 COVID-19 patients (male: 47.29%) in the 15 studies included. All clinical controlled trials have been conducted in China since 2019. The disease stages of the patients included in the study were mild, moderate, severe, and critical. No trial utilized placebo of QFPD.

Table 1 Basic characteristics of included studiesAssessment of risk of bias

As shown in Table 2 and Fig. 2, the quality assessment of the included studies adopted the risk of bias (ROB) assessment tool provided by the Cochrane Handbook. 7 trials (7/15, 46.7%) [14, 18, 19, 23,24,25, 27] reported random sequence generation. Allocation concealment was unclear for it was not described in this review. 4 trials reported no application of blinding. 6 trials reported blinding of participants and personnel, and 2 trials [19, 21] reported blinding of outcome assessment. We evaluated each of the included studies as a whole according to these seven criteria: 5 trials were medium risk and the other 10 were at low risk.

Table 2 The risk of bias of included trialsFig. 2Description of QFPD

QFPD consists of 21 traditional Chinese herbs as shown in Table 3. Ephedra, containing ephedrine and pseudoephedrine, has the effect of epinephrine, and it excites sympathetic nerves, promotes sweating, and removes evil to go out. Glycyrrhiza uralensis Fisch invigorates the stomach and reconciles various drugs. In addition, glycyrrhizic acid has the effect of corticosteroids and plays a strong anti-inflammatory effect. Pericarpium citri Reticulatae, agastache rugosus, and ginger contain volatile oil, sweating while promoting intestinal peristalsis. Its luteolin (Fig. 9b) can reduce the production of pro-inflammatory cytokines and inflammatory mediators, and regulate immunity [30]. Asarum contains kaempferol, anti-inflammatory, anti-cancer, anti-oxidation, antiviral, and anti-bacterial. It can enhance the body's immunity and has other pharmacological effects. Kaempferol showed concentration-dependent inhibition in non-cytotoxic concentrations [31]. Yam promotes digestion and nourishes the body, and a large amount of starch helps dissolve the effective components of raw gypsum.

Table 3 Components of Qingfei Paidu decoction

In summary, QFPD can repair lung injury, enhance the immunity of the body, and play a role in regulating multiple targets and signaling pathways. Pharmacological studies have shown that QFPD inhibits the mRNA translation of the SARS-COV-2 virus by acting on multiple ribosomal proteins [32]. The dosage forms of QFPD include decoction (14/15, 93.3%) and granule (1/15, 6.7%).

Overall outcomes assessmentClinical cure rate

Thecriteria for clinical cure rate must meet the four conditions:

a)

The body temperature has returned to normal for more than 3 days;

b)

Respiratory symptoms improved significantly;

c)

Lung imaging showed significant absorption improvement of acute inflammation;

d)

Two consecutive negative nucleic acid tests of respiratory tract specimens (The sampling interval must be at least one day) (COVID-19 Diagnosis and Treatment Protocol Trial Version 9) [33].

The clinical cure rate of QFPD was reported in 8 studies [14, 17,18,19,20,21,22,23]. There were 473 patients in the experimental group and 436 in the control group. Meta-analysis showed that QFPD had a significant improvement on clinical cure rate, compared to CWM (8 trials, n = 909; RR = 1.15; 95%Cl 1.10–1.20; I2 = 0%, P < 0.00001; Fig. 3a).

Fig. 3

Forest plot of (a) Clinical cure rate, (b) Lung CT, (c) Ranging from mild to critical condition, (d) Length of hospital stay

Lung CT

Five trials assessed the efficacy of QFPD on lung CT [11, 14, 17, 18, 22]. There were 222 patients in experimental group and 243 in control group. QFPD exhibited a significant improvement on lung CT (5 trials, n = 465; RR = 1.22; 95% Cl 1.12–1.33; I2 = 0%, P < 0.00001; Fig. 3b).

Ranging from mild to critical condition

In order to evaluate the effects of QFPD on ranging from mild to critical condition, four trials were enrolled in this study [14, 15, 17, 21]. QFPD had obvious important effect on ranging from mild to critical condition (4 trials, n = 347; RR = 0.35; 95%Cl 0.21 to 0.60; I2 = 0%, P = 0.0001; Fig. 3c).

Adverse effects

The adverse effects were reported in 6 trials [14, 17, 18, 25,26,27]. Meta-analysis showed that QFPD had a significant improvement on reducing adverse effects (6 trials, n = 3865; RR = 0.8; 95%Cl 0.68–0.95; I2 = 21%, P = 0.01; Fig. 3d).

Fig. 4

Forest plot of (a) Adverse effects, (b) Death, (c) Time for nucleic acid conversion

Death

Three trials evaluated the effects of QFPD on death [20, 23, 27]. Compared to CWM, a significant reduction in death was observed by QFPD (3 trials, n = 9448; RR = 0.23; 95%Cl 0.17–0.33; I2 = 0%, P < 0.00001, Fig. 4a).

Time for nucleic acid conversion

The effect of QFPD on time for nucleic acid conversion was reported in 5 trials [15, 24,25,26, 28].QFPD shortened the time for nucleic acid conversion obviously (5 trials, n = 502; WMD = − 4.08; 95%Cl − 5.14 to − 3.02; I2 = 76%,P < 0.00001, Fig. 4b).

Length of hospital stay

9 studies reported length of hospital stay [14, 15, 17, 18, 20, 21, 24,25,26]. There were 391 patients in experimental group and 386 in control group. Meta-analysis showed a significant improvement on length of hospital stay by QFPD (9 trials, n = 777; WMD = − 2.89; 95% Cl − 3.04 to − 2.73; I2 = 98%,P < 0.00001; Fig. 4c).

Clinical symptoms assessmentCough

In the field of disappearing time of cough, 3 studies were enrolled in the review [15, 17, 29]. A significant improvement on disappearing time of cough between QFPD and CWM was identified in this study (3 trials, n = 175; WMD:-1.63; 95% CI − 1.89 to − 1.37; I2 = 0%, P < 0.00001; Fig. 5a).

Fig. 5

Forest plot of impact of QFPD on (a) Cough (b) Fever (c)Fatigue

Fever

Three studies reported the symptom of fever [15, 17, 29]. In the field of fever reduction time, the results suggested no significant difference on the time of fever reduction between QFPD and CWM (3 trials, n = 175; WMD: − 1.48; 95% CI − 1.84 to − 1.13; I2 = 47%, P < 0.00001;Fig. 5b).

Fatigue

The effect of QFPD on fatigue was evaluated in 2 studies [15, 29]. There were 67 patients in experimental group and 48 in CWM group. Improvement on disappearing time of fatigue was identified in QFPD group compared to CWM group (2 trials, n = 115; WMD: − 1.47; 95% CI − 2.19 to − 0.75; I2 = 10%, P < 0.0001; Fig. 5c).

Laboratory indicatorsWBC

For the number of WBC, 5 trials [14, 18, 22, 25, 29] involving 288 patients were enrolled. No significant difference on WBC was identified, compared to CWM (5 trials, n = 288; WMD: 0.50; 95%Cl − 0.69 to 1.69; I2 = 80%, P = 0.41; Fig. 6a).

Fig.6

Forest plot of impact of QFPD on (a) WBC, (b) CRP, (c) PCT

CRP

Four trials evaluated the efficacy of QFPD on the level of CRP [14, 19, 22, 26]. Meta-analysis showed a significant improvement between QFPD and CWM on the number of CRP in patients with COVID-19 (4 trials, n = 315; WMD:-4.39; 95%CI − 6.58 to − 2.20; I2 = 92%, P < 0.0001; Fig. 6b).

PCT

To evaluate the effects of QFPD on the PCT number, 3 trials were enrolled in this study [19, 20, 22]. There were 60 patients in the experimental group and 52 in the control group. QFPD had obvious important effect on PCT (3 trials, n = 112; WMD =− 0.15; 95% Cl − 0.81 to − 0.12; I2 = 0%,P < 0.00001; Fig. 6c).

Results of sensitivity analysis

The sensitivity analysis of length of hospital stay, WBC and CRP was shown in Fig. 7, and the results showed that omitting individual studies exhibited no significant effects on the pooled results compared with the results of the original forest map, indicating that the study results were stable.

Fig. 7

Sensitivity analysis of (a) Length of hospital stay, (b) WBC, (c) CRP

Publication bias

Publication bias was detected by plotting the funnel plots of included trials. The asymmetry showed a mild publication bias. (Fig. 8).

Fig. 8

Funnel plot of (a) Clinical cure rate, (b) Lung CT, (c) Length of hospital stay

GRADE assessment

According to the GRADE assessment for the efficacy and safety of QFPD, different quality levels of evidence were reported in Table 4. There were moderate evidences in clinical effective rate, ranging from mild to critical, adverse effects and CRP. The criteria of “risk of bias” ranked “serious” for all outcome measures, leading the evidence strengths to moderate. Even worse, significant publication bias was also detected in terms of lung