3.1. Study Selection and Description
Our search generated a total of 312 potentially relevant studies, finally, 10 RCTs (English databases: n = 6; Chinese databases: n=1; Korean databases: n=3) [18-27] met our inclusion criteria (Fig. 1). The characteristics of the included RCTs are shown in Table 1. We converted each name of herb into the scientific name.
Three [18, 20, 23] of the 10 trials were conducted in Korea and published in Korean. For the remaining seven studies, two were conducted in Iraq and published in English [25, 26], and five were conducted in Italy , Iran , Pakistan , India  and China , respectively.
A total of 656 participants were included in 10 trials. The sample sizes in each group ranged from 15 to 113 in the EAHM group and from 10 to 120 in control group. The median sample sizes per arm were 33 in the EAHM group and 32 in the control group. One study  reported cheeks of same patients with AV received EAHM (containing Hippophae rhamnoides) versus placebo, and EAHM (containing Cassia fistula) versus placebo respectively.
3.3.1. Types of EAHM forms
Various types of EAHM form were used in 10 included studies. Of these, three studies [18, 25, 26] applied EAHM lotion to AV patients. Cream [19, 24] and pack [20, 27] types were utilized in two studies. One study was for gel  and emulsion  respectively. One study  used three types of forms including foaming cleanser, toner and ample, with same ingredients on AV patients.
3.3.2. Amount of EAHM Used
The amount of EAHM used was reported in only two studies: one study  utilized total 500 mg of EAHM. A total 170 ml of EAHM foaming cleanser, 200 ml of EAHM toner and 60 ml of EAHM ampule were used in the other study .
3.4.1. EAHM versus Placebo
Of the eight studies [18-25] that compared EAHM with placebo interventions, six studies [18-20,23-25] provided data for statistical pooling.
Four studies [21-24] compared primary outcome ‘global assessment’ between EAHM and placebo; two studies [23, 24] of meta-analysis showed significant effect on improving the global assessment [Fig. 2(a), MD = -2.62, CI = -4.84 to -0.40, P=0.02] with high heterogeneity (P = 0.002, I2 = 90%); in the study by Enshaieh et al.  reported significant difference in two groups (P < 0.001); we did not assess result of one trial due to insufficient data.  Park et al.  divided face into six parts, and assigned scores for each part.
For primary outcome ‘lesion count of acne’, data extracted from two studies [19, 20] showed a significant superior improvement in EAHM compared to placebo in inflammatory lesion count of acne [Fig. 2(b), MD = -1.25, CI = -1.68 to -0.83, P < 0.00001, I2 = 0%] and non- inflammatory lesion count of acne [Fig. 2(c), MD = -1.32, CI = -1.75 to -0.90, P < 0.00001, I2 = 0%]. As meta-analysis was impossible in the other two trials, we described the result of the studies as follows; one trial  showed significant improvement in inflammatory lesion count of papules (P < 0.05), inflammatory lesion count of pustules (P < 0.01) and non-inflammatory lesion count of comedones (P < 0.001); in another trial , EAHM treatment significantly reduced inflammatory lesion count of papules (P < 0.05) and pustules (P < 0.05) compared with placebo.
Among eight studies that contrasted EAHM with placebo, four trials assessed secondary outcome ‘sebum of skin’; meta-analysis on three studies [18, 19, 23] showed no significant difference between the groups [Fig. 2(d), MD = -0.21, CI = -0.53 to 0.11, P = 0.20, I2 = 0%]; in the study by Khan et al.  reported EAHM group was more effective than placebo group in sebum of skin (P < 0.05).
The pooled data from two trials [20, 25] indicated no significant effect in secondary outcome ‘patient-reported changes in symptom’ [Fig. 2(e), RR = 2.56, CI = 0.43 to 15.22, P = 0.30, I2 = 90%].
In the trial of Baek et al. , the effect was not significant in secondary outcome ‘anti-bacterial activity test’. Anti-bacterial activity was observed in EAHM group but not in placebo group. There were no significant differences between groups in secondary outcome ‘quality of life’. 
184.108.40.206. Adverse Events
Seven studies [18-23,25] reported AEs; Baek et al.  reported minor AEs, such as itching and irritation in EAHM group; we found itching, irritation and pigmentation as minor AEs of EAHM treatment ; one trial  reported pruritus, burning sensation and scaling occurred in the EAHM group, pruritus and burning sensation presented in the control group; Khan et al.  reported four cases of itching and irritation occurred in the EAHM group, and seven cases of itching and irritation presented in the placebo group; AEs did not occur in another three studies [19, 23, 25]. One study  did not mention AEs. We pooled data from seven trials [18-23,25], EAHM group reported more AEs than placebo group but no statistically significant differences were observed between groups [Fig. 3(a), RR = 1.37, CI = 0.46 to 4.09, P = 0.58, I2 = 39%].
3.4.2. EAHM versus CM
We summarised the results of two studies [26, 27] compared EAHM with CM intervention because data pooling was not assessable.
Two trials [26, 27] evaluated the effectiveness and safety of EAHM compared with CM; in the study by sharquie  that compared EAHM lotion (containing Camellia sinensis) with zinc sulphate reported an improvement in global assessment. A significant improvement at acne count of inflammatory lesion (P < 0.001) and non-inflammatory lesion (P < 0.001) were observed in EAHM group but not in control group; the results of study  that contrasted EAHM pack with benzoyl peroxide indicated significant difference in acne lesion count and patient-reported changes in symptom (P < 0.05).
220.127.116.11. Adverse Events
There were no severe AEs in all participants; one study  reported five cases of itching occurred in the EAHM group, and two cases of itching and five cases of burning sensation represented in the control group; another study  mentioned two cases of AEs occurred in the treatment group, and three AEs presented in the control group. The results of AEs did not show a significant difference between the groups (Fig. 3(b), RR = 0.71, CI = 0.31 to 1.66, P = 0.43, I2 = 0%).
3.5. Cochrane risk of bias assessment
Overall, the included RCTs had a low methodological quality (Table 2). Only one trial  reported adequate random sequence generation using computer number generator, whereas two trials [19, 23] reported inappropriate method (sequence generated by patients entered the study).
Allocation was properly concealed in two trials. One study [19, 21] reported that allocation was properly conducted using identical appearance that is sealed randomization code. Another study  reported that allocation was performed by and independent person.
The participant and practitioner were blinded in five trials [18-21,23]; Double-blinded RCTs were conducted in two trials [19, 21] and same form of intervention was used in the EAHM and control groups. [18, 20, 23] Single-blinded studies were employed in four studies [22,24-26] and one study  used different types of intervention form in both groups. Only one trial  reported the details of the blinding of outcome assessment.
Most of the included studies [18,19,21,23,24-27] had low risk of bias addressing incomplete outcome data (e.g., no missing outcome data or missing outcome data, but the drop-out rate did not exceed 20% for short-term follow-up). Two trials [20, 22] had high ROB; one  had missing data and the drop-out rate exceeded 20% for short-term follow-up, and another  reported that the number of patients was different when we compared pre-treatment with post-treatment in outcome measure of global assessment.
Regarding selective outcome reporting, only one  studies reported their protocol before conducting the RCTs.