Andrographis paniculata (Chuān Xīn Lián) for symptomatic relief of acute respiratory tract infections in adults and children: A systematic review and meta-analysis

Introduction Antimicrobial resistance (AMR) is a substantial threat to public health. Safe and effective alternatives are required to reduce unnecessary antibiotic prescribing. Andrographis Paniculata (A. Paniculata, Chuān Xīn Lián) has traditionally been used in Indian and Chinese herbal medicine for cough, cold and influenza, suggesting a role in respiratory tract infections (RTIs). This systematic review aimed to evaluate the clinical effectiveness and safety of A. Paniculata for symptoms of acute RTIs (ARTIs). Materials and methods English and Chinese databases were searched from their inception to March 2016 for randomised controlled trials (RCTs) evaluating oral A. Paniculata without language barriers (Protocol ID: CRD42016035679). The primary outcomes were improvement in ARTI symptoms and adverse events (AEs). A random effects model was used to pool the mean differences and risk ratio with 95% CI reported. Methodological quality was evaluated using the Cochrane risk of bias tool; two reviewers independently screened eligibility and extracted data. Results Thirty-three RCTs (7175 patients) were included. Most trials evaluated A. Paniculata (as a monotherapy and as a herbal mixture) provided commercially but seldom reported manufacturing or quality control details. A. Paniculata improved cough (n = 596, standardised mean difference SMD: -0.39, 95% confidence interval CI [-0.67, -0.10]) and sore throat (n = 314, SMD: -1.13, 95% CI [-1.37, -0.89]) when compared with placebo. A. Paniculata (alone or plus usual care) has a statistically significant effect in improving overall symptoms of ARTIs when compared to placebo, usual care, and other herbal therapies. Evidence also suggested that A. Paniculata (alone or plus usual care) shortened the duration of cough, sore throat and sick leave/time to resolution when compared versus usual care. No major AEs were reported and minor AEs were mainly gastrointestinal. The methodological quality of included trials was overall poor. Conclusions A. Paniculata appears beneficial and safe for relieving ARTI symptoms and shortening time to symptom resolution. However, these findings should be interpreted cautiously owing to poor study quality and heterogeneity. Well-designed trials evaluating the effectiveness and potential to reduce antibiotic use of A. Paniculata are warranted.


Introduction
Antimicrobial resistance (AMR) is a substantial threat to public health. Safe and effective alternatives are required to reduce unnecessary antibiotic prescribing. Andrographis Paniculata (A. Paniculata, Chuān Xīn Lián) has traditionally been used in Indian and Chinese herbal medicine for cough, cold and influenza, suggesting a role in respiratory tract infections (RTIs). This systematic review aimed to evaluate the clinical effectiveness and safety of A. Paniculata for symptoms of acute RTIs (ARTIs).

Materials and methods
English and Chinese databases were searched from their inception to March 2016 for randomised controlled trials (RCTs) evaluating oral A. Paniculata without language barriers (Protocol ID: CRD42016035679). The primary outcomes were improvement in ARTI symptoms and adverse events (AEs). A random effects model was used to pool the mean differences and risk ratio with 95% CI reported. Methodological quality was evaluated using the Cochrane risk of bias tool; two reviewers independently screened eligibility and extracted data.

Results
Thirty-three RCTs (7175 patients) were included. Most trials evaluated A. Paniculata (as a monotherapy and as a herbal mixture) provided commercially but seldom reported manufacturing or quality control details. A. Paniculata improved cough (n = 596, standardised mean difference SMD: -0. 39

Introduction
Respiratory tract infections (RTIs) are one of the most common reason for primary care consultations in the UK [1]. Treatments for RTIs are mainly symptomatic [2], and often include analgesics, antipyretics [3], mucolytics, expectorants, decongestants [4], and educational interventions [5], although evidence supporting currently used symptomatic treatment is still limited [6]. Antibiotics are frequently prescribed in primary care settings in Europe [7] with 60% of all antibiotic prescribing in the UK occurring in primary care [1]. Research has suggested RTIs are predominantly of viral aetiology [8], and that antibiotics are of very limited benefit in the majority of uncomplicated infections [9,10]. Systematic reviews to date have failed to provide evidence for the effectiveness of antibiotics for RTIs [11]. Antibiotics showed no benefit in symptom improvement for acute RTIs (ARTIs) such as colds [12], persisting acute purulent rhinitis [12], or acute laryngitis [13]; and suggested little absolute benefits for reducing symptom duration or complications in sore throat [14], bronchitis [15,16], sinusitis [17] and acute otitis media [18]. Antimicrobial resistance (AMR) is an evolving major global threat to public health [19]. A recent Public Health England report showed a 6% increase in total antibiotic use in England between 2010 and 2013 and it remains an important government priority to reduce antibiotic prescribing [20,21]. The marginal benefit of antibiotics for ARTIs are outweighed by increasing AMR and common adverse reactions [3] leading to unnecessary increases in healthcare costs [22][23][24].
Research is urgently needed to explore other treatments that may be offered for symptomatic relief to reduce unnecessary antibiotic prescribing. In order to facilitate rapid translation of research into clinical practice, there has been much interest in researching options currently available to the general public. This has involved over the counter (OTC) pharmacological treatments such as paracetamol as well as herbal alternatives. Evidence from previous systematic reviews suggested promising but limited evidence for Chinese herbs in influenza [25], common colds [26], upper RTI [27], and cough [28].
A. Paniculata (Burm.f.) Wall ex Nees (Acanthaceae family), also known as nemone chinensi, Chuān Xīn Lián, has traditionally been used in Indian and Chinese herbal medicine. It is traditionally used as an antipyretic for relieving and reducing the severity and duration of symptoms of common colds and alleviating fever, cough and sore throats, or as a tonic to aid convalescence after uncomplicated RTIs [29] [30]. There is encouraging evidence to demonstrate the potential mechanistic for effects of A. Paniculata for RTIs. The active constituents of A. Paniculata include the diterpene, lactones commonly known as the andrographolides which have shown anti-inflammatory, antiviral, anti-allergic, and immune-stimulatory activities [31]. They inhibit platelet-activating factor mediated inflammatory response [32], reduces expression of pro-inflammatory proteins such as cyclooxygenase-2 [33,34], and demonstrates analgesic effects as well as antipyretic effects comparable to paracetamol [35]. A. Paniculata has also been shown, in vitro, to be effective against avian influenza A (H9N2 and H5N1) and human influenza A H1N1 viruses, possibly through blocking the binding of viral hemagglutinin to cells [36], or by inhibiting H1N1 virus-induced cell death [37].
Two previous systematic reviews showed that A. Paniculata alone or in combination with A. senticosus is superior to placebo for reducing symptom severity in upper RTIs [38,39]. However, the clinical evidence for A. Paniculata for symptoms of lower RTI has not yet been systematically evaluated and would be important to review prior to conducting further research in this area. Furthermore, previous systematic reviews have been limited to Englishlanguages searches and given that A. Paniculata is used in Indian and Chinese herbal medicine, an up-to-date systematic review without language restrictions is warranted.
This systematic review therefore evaluated the clinical efficacy, effectiveness and safety of A. Paniculata for of the treatment of ARTIs.

Materials and methods
This systematic review followed PRISMA reporting guidelines (S1 Table). A protocol of this review has been registered (CDR: CRD42016035679, S1 File). Ethics statement: N/A.

Search strategy and study selection
MEDLINE, EMBASE, AMED, Cochrane Library, CINAHL, China National Knowledge Infrastructure (CNKI), Wan Fang, Sino-Med Database, and Chinese Science and Technology Journal Database (VIP) were searched from their inception to March 2016. A range of freetext words and indexed terms related to "Andrographis Paniculata" and "respiratory tract infection" were searched. The reference lists of studies meeting the inclusion criteria were searched to identify additional relevant studies. A detailed search strategy and search term alternatives for each database are available as supporting information; see S2 File. There were no exclusions made based on language. Literature searching (XYH, RHW) was followed by independently screening with at least two authors (XYH, RHW, ML). Study authors were contacted to obtain relevant missing data if necessary and where resources allowed.

Data extraction and management
A data extraction spreadsheet was designed and piloted with appropriate changes made for this review. The form identified trial characteristics, characteristics of trial population and conditions, details of interventions in all trial arms according to the consolidated standards of reporting trials (CONSORT) herbal extension in terms of features of herbal intervention [40], details of concomitant interventions, quality assessment, and findings on efficacy, effectiveness and AEs. Two reviewers extracted study data independently for Chinese-language (XYH, RHW, LL) and English-language (ML, CB) trials, with findings compared and agreed.

Eligibility criteria
This review included published and unpublished randomised controlled trials (RCTs). Quasi-RCTs, crossover trials, controlled before and after studies, interrupted time series (ITS) studies, and non-experimental studies were not included due to their potential high risk of bias.
Studies of human participants of all ages, with symptoms of ARTIs. A clinical diagnosis of ARTI was the main inclusion criteria. Diagnoses of upper or lower ARTIs include acute common cold, influenza, rhinosinusitis, laryngitis, tonsillitis, pharyngitis, croup, acute otitis media, bronchitis, pneumonia, and acute exacerbations of chronic obstructive pulmonary disease (COPD). Symptoms of ARTIs are defined as having symptoms such as cough, sore throat, fever, runny nose and discoloured sputum for a duration of less than four weeks. Trials were excluded if they recruited participants with asthma, had active or previous peptic ulceration, were hypersensitive to analgesics, had psychosis, or were severely depressed. Exclusion also applied to trials that included patients who required hospital admission (for example, for meningitis, severe pneumonia, epiglottitis, or Kawasaki disease), had a known immune deficiency, or were pregnant or breastfeeding [41].
Examples of herbal mixture include: products containing A. Paniculata in combination with Scutellaria baicalensis, or in combination with Lonicera japonica, Forsythia suspense, and Aster trinervius. No limitation was imposed concerning dosage, methods of dosing or duration of administration.
We included comparisons such as placebo or no intervention; usual care such as analgesics, antivirals, antibiotics, anti-inflammatories, steroids or corticosteroids; or other herbal remedies. Studies comparing different preparations of A. paniculata, e.g. comparing tablet with granule, were also included in this review.

Outcome measures
The following primary outcome measures were included in this review: 1. Participant self-reported or clinician/observer assessment on overall ARTI symptoms; or two target symptoms cough and sore throat. Commonly used measures included: • Changes on visual analogue scales (VAS) • Changes in symptoms scored on a Likert-type scale • Global assessment of symptom improvement by the patient • Global assessment of symptom improvement by treating clinician 2. AEs: This included any anaphylactic, allergic reactions, hypersensitivity reactions, or complications of A. Paniculata, such as rash, nausea, fatigue, or worsening of ARTIs symptoms. We also collected information regarding AEs due to interactions among A. Paniculata in combination with other remedies, or potential interactions with medications patients had for their co-morbidities.
We defined serious AEs according to the International Council on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidelines as any event that leads to death, is life-threatening, requires hospitalisation or leads to persistent or significant disability; biochemistry results such as electrolytes, liver and kidney function tests (alanine aminotransferase and creatinine) [42].
Secondary outcome measures included: 1. Mean time to reported remission or resolution of symptoms. This may be measured directly, through patient or clinician/observer report or indirectly as the time to return to normal activities.
2. Reduction in reported antibiotic usage, e.g. number of scripts issued immediately at the time of consultation and uptake of delayed prescriptions. Although the Chinese government launched a special campaign to promote the rational use of antimicrobials in healthcare settings in the 2011 healthcare reform, this has yet to be implemented in many places in China [43]. Antibiotics are prescribed on patients' initial visit if there were suspicions of bacteria inflammation, therefore scripts immediately issued at the time of consultation was recorded.
Trials that did not report either our primary and or secondary outcome measures were excluded from this review.
Timing of effect measures: Some studies may have used a repeated measures approach. Timings of measures for each included trial were documented with commonly reported time points explored if there was sufficient data available. All outcome measures were assessed at baseline and data for all time points were extracted with the aim to pooling those trials that collected data at similar time points. Otherwise, data at the most appropriate follow-up point were assessed.

Assessment of risk of bias in included studies
The risk of bias of the included RCTs was assessed independently by two reviewers using the tool developed by Higgins and Green in the Cochrane Handbook for Systematic Reviews of Interventions [44]. We assessed bias over the following domains: selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of researchers conducting outcome assessments), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other sources of bias. A judgement of 'low risk' of bias, 'high risk' or bias, or 'unclear risk' of bias was provided for each domain. Any disagreements were resolved by discussion or by involving a third reviewer until consensus was reached.

Dealing with missing data
Where data was missing or incomplete, we contacted study authors to obtain this where possible. If the means were reported without standard deviations, we calculated the standard deviation from the information reported such as p-values, F-values or confidence intervals. As far as possible, we utilised intention to treat (ITT) analysis data for all outcomes. However, most included trials reported complete cases only; and complete case data were the primary analysis dataset. For each outcome, the number of participants whose data was available at baseline and at follow up, and the rate of loss to follow-up were recorded.

Assessment of heterogeneity
Between-study heterogeneity was assessed using the I 2 -statistic which describes the percentage of variation across studies that is due to heterogeneity rather than chance. Rules of thumb for interpretation of this statistic suggest that I 2 >30% equates to moderate heterogeneity, I 2 >50% equates to substantial heterogeneity and I 2 >75% equates to considerable heterogeneity [46]. For all I 2 values above 50%, we investigated potential sources of heterogeneity. Although this threshold is widely used, it is somewhat arbitrary and therefore if the I 2 value was below 50% but the direction and magnitude of treatment effects suggest important heterogeneity, we investigated the potential sources in a sensitivity analysis and took this into account when interpreting the findings. As high levels of heterogeneity were expected due to complexity in the form of A. Paniculata (e.g. monotherapy or herbal mixture, capsule or liquid), it was planned to use a random effects model to pool the overall effects [46].

Assessment of reporting biases
Funnel plots were created to investigate potential reporting bias where this was feasible and there were sufficient studies [49]. Funnel plot tests for asymmetry were conducted separately in STATA software version 14 using the metabias command.

Sensitivity analysis
Sensitivity analyses were conducted for the primary outcomes to determine whether the review conclusions would have differed if eligibility was restricted to trials without high or unclear risk of bias for either in sequence generation or allocation concealment domains) [46]; and if eligibility was restricted to trials that provided any detail on authentication or standardisation of the herb.

Subgroup analysis
If there was sufficient available data, several subgroup analyses were planned a priori to compare the effect estimate between studies that evaluated: • Patients with upper ARTIs versus lower ARTIs; • Adults versus children (younger than 18); • A. Paniculata as monotherapy versus as fixed combinations; • A. Paniculata in different preparation, e.g. granule versus tablet or other forms

Interventions
Experimental interventions included A. Paniculata as a monotherapy and as an herbal mixture in combination with other herbs. Table 6 presents the characteristics of A. Paniculata reported Included trials seldom reported manufacturing or quality control details. Three reported method of measuring andrographolide proportion using HPLC technique [76][77][78]; and only one reported that the product was produced, analysed and bottled according to good manufacturing practice (GMP) standard [59]. Three trials reported added materials [57,76,78] but only one [78] provided clear description (200 mg of micro crystalline cellulose). Extract solvents used included methanol [78], polyethylene glycol [80], and two used methanol for HPLC extraction [76,77]. Only one trial provided extract solvent concentration details [78].

Outcome measurements
The most commonly reported primary outcome measure was global assessment on overall symptoms improvement (Tables 1-5). Although not clearly reported in every trial, it is assumed this outcome was measured by the practitioner. Apart from one study [63], all Chinese-language trials reported four-category scores in symptoms of ARTIs, among which 11 [55, 57, 58, 62, 71-73, 75, 80-82] reported data based on the CMCRG [中药新药临床研究指 导原则]. The CMCRG is a four-category scoring system to evaluate overall treatment effects based on: 1). Cured: a). no temperature in 3 days, b). no symptom or sign of RTIs, c). accumulated score decreases !95%; 2) Markedly effective: a). no temperature in 3 days, b). most symptoms and signs of RTIs disappear, c) accumulated score decrease between 70% to 95%; 3). Effective: body temperature decreased in 3 days, b). most of key symptoms and signs of RTIs disappear, c). accumulated score decrease between 30% to 70%; 4). Ineffective or worsening: a). no decrease or increased body temperature, b). no improvement in key symptoms and signs of RTIs or even getting severe, c). accumulated score decreases less than 30%. Accumulated score was calculated as: (baseline score-endpoint score)/baseline score X100%. Scores were given based on: 1). Symptoms of ARTIs, e.g. symptoms: fever, sore throat, cough, nasal congestion, runny nose, headache, sweating, sneezing, thirst, 2). Signs of ARTIs, e.g. aversion to wind, and changes in tongue appearance and pulse; and 3). Laboratory checks, e.g. chest radiography, circulation, faeces, blood, urine, liver and kidney function, electrocardiogram (ECG). In this review, the combined cure and markedly effective (CCME) rate was considered as improved by the review authors. Symptom score on severity of cough [59,63,66,75,76], sore throat [59,75], and overall symptoms (commonly a list of 8-12 ARTI symptoms) [69,70] were reported in seven trials. Secondary outcome measures reported in the included trials were: time to resolution of cough [55,56,59,67], of sore throat [55,56,59], and of overall symptoms [58,61]; only one trial reported reduction in reported antibiotic usage [70].
A few trials used a repeated measures approach [50,56,69,71,80]. Apart from one trial on acute pharyngitis which followed-up at 20 days [51], the most common end point follow-up that was reported ranged from 3 to 7 days and the outcome data for the end points closest to an average of 5 days were extracted and assessed (Tables 1-5).
Most trials (24 of 33) had a high risk of bias in blinding of the participants and personnel as they assessed two interventions that were different in dosage, or form of preparation, or two Andrographis paniculata for acute respiratory tract infections types of interventions, or compared A. Paniculata plus usual care versus usual care, without any blinding information given. Two trials comparing A. Paniculata with placebo control had low risk of bias as both patients and evaluator [76] or investigator and pharmacist [78] were blinded to group assignment and could not distinguish between the two interventions. The remaining trials [52,55,59,71,74,77,79] provided no information regarding similarities between interventions, or provided no information to confirm whether or not blinding of personnel was conducted. Most included trials failed to provide enough information to determine whether blinding of outcome assessment was achieved. Nine trials were judged to be at high risk of bias as they assessed subjective outcome measures and the patients or practitioners knew that which intervention they had been assigned to (i.e. A. Paniculata plus usual care versus usual care) [62][63][64][65][66][67][68][69][70].
Twenty-six included trials reported no attrition. Among the 7 trials that had dropouts, three trials reported 3-8% dropout and conducted ITT by assuming no effect for dropouts. No per protocol analysis was performed for those three trials [56,58,73]. Two trials reported dropouts (1% [78] and 6% [52]) without ITT analysis. Another trial reported 25% dropout and provided both ITT and per protocol analysis findings [76]. The author suggested that the dropout rate in two groups were equal and that the potential reason for large dropout may have been related to three weeks' winter holiday. One trial did not clarify how missing data was dealt with [70].
One trial [79] published a protocol containing information on outcome measures and follow-up points that were consistent with the main trial report. All remaining trials did not have a protocol available. Four trials [65,71,75,82] reported selected findings that were not fully consistent with the outcome measures set in the methods. Only one trial had no obvious risk of other bias [80] and this was the only trial that stated that there was no conflict of interests. None of the other included trials stated whether or not a conflict of interest existed and three trials included one or more author who worked for the pharmaceutical company of the product being evaluated as an intervention [59,71,77]. The most common reasons for high risk of other bias were: 1). In 12 trials, diagnostic criteria were not applied at recruitment and there were no inclusion or exclusion criteria specified [53, 54, 58, 60-62, 65, 67, 68, 74, 81, 82]; 2). Four trials provided either no condition-related baseline data [63,75,81,82], or no sociodemographic characteristic baseline [59,79], or neither [69]; and 3). Two trials reported discrepancies between permitted co-intervention(s) for the intervention and control groups: in one trial, paracetamol was given if body temperature > 39 in the treatment group but 38-38.5 in the control group [59]; the other trial allowed no additional treatment for the intervention group only [61]. One third of the trials reported informed consent [55,56,59,64,66,[69][70][71][78][79][80].
Funnel plot for one comparison was performed to investigate potential publication bias (Fig 3). There was no evidence (p = 0.870) of small-study effects.

Effect estimates
The included trials featured five comparison groups: A. Paniculata versus placebo (4 trials); A. Paniculata versus usual care (12 trials); A. Paniculata plus usual care versus usual care alone (9 trials); A. Paniculata versus other active herbal interventions (5 trials); and A. Paniculata pillule versus A. Paniculata tablet (3 trials). Subgroup analyses were performed for two of the planned subgroups: monotherapy or herbal mixture and different forms of preparation of A. Paniculata. These were conducted for primary outcome measures in A. Paniculata versus usual care and A. Paniculata plus usual care versus usual care. Subgroup analysis in other comparison groups and subgroup analysis on upper or lower ARTIs, and adults versus children were not performed due to insufficient data.
A. Paniculata vs placebo (n = 4). Evidence from four trials (three had low or medium RoB [76,78,79] (Fig 4) [76][77][78][79]. One trial showed a statistically significant effect in favour of A. Paniculata as a single herb in tablet compared to placebo as measured by patient reported rate of improvement in overall symptoms (n = 50, RR: 2.80, 95%CI [1.19, 6.30]) [77]. No data was available under this comparison for time to symptom resolution or antibiotic medication usage.
A. Paniculata vs usual care (n = 12). Evidence from ten trials showed a statistically significant effect in favour of A. Paniculata compared to usual care as measured in overall symptoms improvement CCME rate (n = 1347, RR: 1.36, 95%CI: [1.18, 1.57], I 2 = 67%) (Fig 5). Heterogeneity for the herbal mixture in capsule subgroup was low when the Wang 2008 trial was removed (p = 0.43, I 2 = 0%). This may be due to: 1). not reporting inclusion/exclusion criteria for recruiting participants and the duration of illness were not clear, therefore there was potentially high population heterogeneity; and 2) lack of authentication. Apart from one subgroup (A. Paniculata as a single herb) failing to show a statistically significant effect [50,51], A. Paniculata as herbal mixture in capsule [53][54][55][56][57][58] and as herbal mixture in tablet [60] and liquid [61] showed statistically significant effects compared to usual care.
Outcomes of three trials in this comparison group were not pooled and were presented narratively: Sun and Zhao also showed significant improvement in overall symptom as measured  [70].
A. Paniculata vs other herbal interventions (n = 5). Evidence from five trials showed a statistically significant effect in favour of A. Paniculata compared to other herbal interventions as measured by improvement rate in overall symptoms (n = 827, RR: 1.44, 95%CI: [1.10, 1.89], I 2 = 89%). Upon removing Zhang 1994 from the analysis, heterogeneity was reduced (I 2 = 66%), while did not greatly change the summary estimates. Possible reasons for this may be that this trial targeted children and that the product evaluated was not authenticated (Fig 9). No data were available for time to resolution or antibiotic medication usage for this comparison group.
A. Paniculata in pillule vs in tablet (n = 3). Evidence from three trials [80][81][82] showed a statistically significant effect in A. Paniculata in pillule compared to A. Paniculata in tablet as measured by improvement rate in overall symptoms CCME (n = 586, RR: 1.14, 95%CI: [1.04, 1.25], I 2 = 86%) (Fig 10). No data was available under this comparison for time to symptom resolution or antibiotic medication usage.

Sensitivity analysis
Sensitivity analyses were conducted by restricting inclusion in the meta-analysis to trials with low risk of bias in both sequence generation and allocation concealment domains [50,76,78]. The effect of A. Paniculata over placebo was enhanced in overall symptoms (n = 219, SMD: -1.21 [-1.50, -0.92]) and in cough (n = 504, SMD: -0.56 [-0.80, -0.31], I 2 = 46%); while the effect for overall symptoms of using A. Paniculata in pollule over A. Paniculata tablet remained similar. Removal of trials that did not provide authentication or standardisation information [50-55, 57, 60, 61, 63-65, 69, 71, 79, 81] did not greatly change the summary estimates. Data from two trials [58,74] were removed from the meta-analysis with reasons given above.

Summary of evidence
Thirty-three trials involving 7175 patients with ARTIs were included in this review with no language restrictions. Findings suggest limited but consistent evidence that A. Paniculata improved cough and sore throat when compared with placebo. A. Paniculata (alone or plus usual care) has a statistically significant effect in improving overall symptoms of ARTIs when compared to placebo, usual care, and other herbal therapies. A. Paniculata in pillule tended to be more effective in improving overall symptoms over A. Paniculata in tablet. Evidence also suggested that A. Paniculata (alone or plus usual care) has shortened the duration of cough, sore throat and sick leave/time to resolution when compared versus usual care. Reduction in antibiotic usage was seldom evaluated in the included trials.
Although no serious AE was observed and minor AEs were mainly gastrointestinal in the included trials, caution is warranted in interpreting safety before comprehensive safety data is available. The quality of included trials was generally lower than desired as many were poorly designed, underpowered and inadequately blinded. There was high heterogeneity among trials due to variations in population and outcomes.

Variations in A. Paniculata
Form of preparation and dosage. The two commonly prescribed preparations in the included trials were capsules and tablets; there were no decoctions. This may due to the extremely bitter nature of the herb described as the "king of bitters". Findings of this review showed A. Paniculata pillules are superior to tablets in reliving overall symptoms [80][81][82], suggesting a place for pillule preparations.
Most A. Paniculata products have an extraction ratio of 14:1 standardised to contain an average of 35% of andrographolides [27] but solvent extraction ratios were not reported in most included trials. The amount of andrographolide produced from a daily dose of A. Paniculata extract varied from 15.75mg of andrographolide for URTIs [70], 225 mg for bronchiectasis [63], and up to 1200 mg for pharyngo-tonsillitis [52]. The most common treatment length was 5-7 days, ranging from 3 days for an AURTI [56] to 14 days for bronchiectasis [63] requiring administration three times daily. There is limited dose-finding research available documenting recommended percentage of active ingredient, dosage or ceiling effects so dosage is based in traditional use and herbal textbooks.  [87]. Traditional Chinese Medicine (TCM) prescriptions often involve several herbs with synergistic actives which are frequently individualised based on the presenting symptoms and TCM diagnosis. This may result in complex phyto-pharmaceutical interactions and AEs.
The active ingredients of A. Paniculata has not been fully identified in most trials but it is generally assumed to be the andrographolides. Only three trials [76][77][78] provided manufacturing details and chromatographic fingerprints of the herbal preparations to ensure quality and consistency of the products (Table 6). Those studies with inadequate information about the herbal content and manufacturing procedures may not be generalisable to other A. Paniculata studies as bioequivalence is 'assumed' rather than proven. A CONSORT herbal extension checklist is recommended to guide reporting of herbal trials and to assure herbal quality and bioequivalence.

Safety (adverse events and toxicity)
The traditional uses of A. Paniculata are as a liver tonic to help maintain appetite and digestion; alleviate gastro-intestinal upsets and acute diarrhoea; immune function and to support intestinal function [27]. This traditional use may reduce adverse reactions caused by conventional medicines when they are prescribed in conjunction with A Paniculata. Findings of this review showed five cases of minor AEs in A. Paniculata group [71,79,81] (two cases were A. Paniculata plus usual care [64,66]) and 48 cases [51,58,61,64] were reported in control groups in the included trials. Minor AEs were mainly gastrointestinal, while there were two cases of dry mouth (Ribavirin [61]) and six cases of skin reaction (Cefixime [51] and Echinacea purperea [70]) reported. This was not consistent with the recent therapeutic goods administration (TGA) pharmacovigilance analysis, which revealed most common AEs associated with A. Paniculata were hypersensitivity or allergic reactions [29]. The TGA safety report explored association between anaphylactic/allergic type ADRs and A. Paniculata, suggesting that ADRs tend to be related to highly concentrated methanol extracts [29]. Our safety findings are inconclusive as there was an absence of proportionate data on each minor AE in each group thus limiting a comprehensive risk-benefit assessment.
Acute toxicity studies in rats suggested median lethal doses for andrograpolide is more than 40g/kg and 10 mg/kg body weight is when the ADRs became apparent [88]. The European Medicines Agency (EMA) reports no acute or genotoxicity data on Andrographis extracts but there is a possibilty of high doses causing reproductive toxicity, with decreases in sperm counts and motility that were linked to disruption of spermatogenesis in rats [89]. Animal research showed andrographolide-induced induction of CYP1A2, indicating an interaction with theophylline [90]. And Baicalin tends to interact with Omeprazole Chlorzoxazone Losartan [91], Rosuvastatin [92] and Acetaaminophen [93]. Mechanism of actions among herbal mixtures included in this review were not properly documented to support their use.

Implications and future direction
This review suggests that A. Paniculata might act as a safe and effective treatment for ARTIs, either alone or in combination with usual care, as monotherapy or as a herbal mixture. Manufacturing information may be an important factor that differed among these included trials, and we recommend all further trials are based on a consistent, safe and well-defined A. Paniculata product. Pharmacological research exploring correlations between ADRs and manufacturing procedures (with methanol, or aqueous solvent, or aqueous-ethanol mixture) are also needed. Sensitivity analysis showed that higher quality trials suggested an enhanced improvement in overall symptoms and cough. Future well designed trials evaluating effectiveness and safety of oral A. Paniculata in capsule or tablet form and reported according to the herbal CONSORT checklist are vital and may serve to minise antibiotic prescription and AMR. The potential for antibiotic sparing should be studied in future trials.

Strengths and limitations
Cochrane methodology was followed with a protocol of this systematic review registered and published online. A broad search strategy including both English and Chinese databases was adopted without language restrictions. Papers identified were screened and eligible trials extracted independently by two reviewers. We attempted to include grey literature by seeking manufacturers' reports and attempted to contact original authors for missing data. A number of studies including a substantial patient sample were identified; characteristics of the herb were documented following the criteria of CONSORT herbal extension.
Methodological quality of included trials was restricted as randomisation was not well documented; 73% of the trials included were not blinded; where ITT analysis were performed, loss to follow-up data were counted as no effect [56,58,73]; and most trials were published without a protocol available. The diagnostic criteria used in included trials were inconsistent and more than one third provided no inclusion/exclusion criteria. Not all trials were performed in countries where the International Council for Harmonisation (ICH) guidelines were legally binding. The included trials rarely clarified whether the products were GMP certified. However, methodological quality judgements were made on the basis of incomplete reporting the evidence of effectiveness may be undervalued [44]. Chinese-language randomised trials present a prominent excess of significant results that requires cautious interpretation [94]. It was not clear whether some of the trials were conducted with adequate ethical review; whether the products evaluated were not authenticated, or whether these details were poorly reported.
There were heterogeneities among trials included due to the heterogeneity population, clinical setting, variations in the form of A. Paniculata and controlled intervention employed, outcome measures, and different study protocols. Inadequate number of trials were available to allow further subgroup analyses on children or on lower ARTIs. Some included trials were non-inferiority RCTs as placebo control was considered unethical by some researchers. They demonstrated that A. Paniculata was clinically superior to other herbal interventions but failed to provide evidence on the established effect.

Conclusions
A. Paniculata appears to be beneficial and safe for relieving ARTI symptoms and reducing time to symptom resolution. The evidence is inconclusive due to limited methodological quality of included trials and study heterogeneity. Well-designed trials evaluating effectiveness, efficacy and safety of A. Paniculata as a monotherapy, or as an herbal mixture, as well as exploring its potential to reduce antibiotic prescribing in primary care, are warranted.