Herbal medications for anxiety, depression, pain, nausea and vomiting related to preoperative surgical patients: a systematic review and meta-analysis of randomised controlled trials

Eligibility criteria

The inclusion criteria were:

• Study design: Randomised controlled trials (RCTs) and quasi-RCT.
• Patients: Adults (≥18 years of age) undergoing laparoscopic, obstetrical/gynaecological, or cardiovascular surgeries.
• Time of intervention: During the preoperative period.
• Interventions: Any herbal medications from any of the following plant preparations (whole, powder, extract, crude drug, standardised mixture, drug extract ratio and solvent) which were compared against conventional treatment, placebo, no intervention, other type of complementary and alternative therapy (eg, acupuncture, homoeopathy), or another herbal medication. The following routes of administration were considered: oral (eg, dropping pills, aqueous decocts), topical and intravenous.

The patient-important outcomes (primary outcomes) that we were interested in were: anxiety (Spilberger Anxiety Inventory–Trait Anxiety Inventory and other validated instruments); depression (Depression Scale–Hospital Anxiety and Depression Scale and other validated instruments); PONV (Visual Analogue Scale (VAS) and other validated instruments), or overall pain (VAS and other validated instruments). Secondary outcomes were:

• Adverse events (primarily withdrawals and serious adverse events (eg, death, life-threatening, hospitalisation, disability or permanent damage).
• Number of patients reporting adverse events (as defined above).
• Quality of life (Short Form-36 and other validated instruments).
• Satisfaction with herbal medications.
• Need for rescue medication.

• Duration of symptoms (intervention costs with descriptive analysis).

  The exclusion criteria were:

• Patients: Studies where the majority of participants were HIV positive, or transplant patients.
• Interventions: Studies involving combination of herbal medication regimens as interventions and/or combination of pharmacological medications as control arms were not considered eligible for inclusion.

Data source and searches

We searched Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Ovid EMBASE, LILACS, ISI Web of Science and CINAHL, from their initial inception dates to 30 January 2018. Search terms describing laparoscopic, obstetrical/gynaecological, cardiovascular surgeries and herbal medication interventions were combined (table 1). The search strategy was designed with the assistance of a trained librarian. No restrictions were placed on language, year of publication or publication status.

Searching other resources

In addition to an electronic database search, we made a manual search in the reference lists of every study deemed eligible in order to identify additional trials that were later included; all potentially eligible studies were screened in duplicate. Furthermore, the coauthors leading eligible trials were contacted for additional data and information that could be potentially included.

Selection of studies

Pairs of reviewers independently screened all titles and abstracts identified by the search. Full-text articles for potentially eligible studies were obtained and screened independently by reviewer pairs using the same eligibility criteria as with title and abstract screening. Consensus for both stages of screening, were established by discussion and adjudication by a third reviewer as necessary.

Data extraction and risk of bias assessment

Once a final set of eligible studies were identified, reviewer pairs independently extracted data for the following variables from each study using a pre-standardised data extraction form with: characteristics of the study design; participants; interventions; outcomes event rates (for afore mentioned primary and secondary outcomes) and duration of follow-up.

Reviewers independently assessed risk of bias by using a modified version of the Cochrane Collaboration’s tool. The tool includes nine domains: adequacy of sequence generation, allocation sequence concealment, blinding of participants and caregivers, blinding of data collectors, blinding for outcome assessment, blinding of data analysts, incomplete outcome data, selective outcome reporting and the presence of other potential sources of bias not accounted for in the previously cited domains.36 37

For incomplete outcome data, we considered a loss to follow-up of less than 10% and a difference of less than 5% in missing data in intervention and control groups as low risk of bias. Reviewers discussed with a third party adjudication to resolve disagreements.

Confidence in pooled estimates of effect

The reviewers used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology to rate quality of evidence for each outcome. Quality ratings were assigned as high, moderate, low, or very low.37 Detailed GRADE guidance was used to assess overall risk of bias,38 imprecision,39 inconsistency,40 indirectness41 and publication bias.42 Consensus was established by discussion and adjudication by a third reviewer as necessary, and final results were summarised in an evidence profile (table 5).

Data synthesis and statistical analysis

Pooled risk ratios (RRs) were calculated for dichotomous outcomes and standardised mean differences for continuous variables with the associated 95% CIs using random-effects models with the Mantel-Haenszel statistical method. Absolute effects and 95% CI were calculated by multiplying pooled RRs and 95% CI by baseline risk estimates derived from the largest included RCTs for each respective herbal remedy in our meta-analysis.

Variability was addressed in results across studies by using I² statistic and the p value obtained from the Cochran Q (χ²) test. Our primary analyses were based on eligible patients who had reported outcomes at the last time-point for each study (complete case analysis).

We planned to perform separate analyses to assess publication bias through visual inspection of funnel plots for outcomes addressed in 10 or more studies; however, the information from the included studies was insufficient for performance of any of these analyses.

We avoided double-counting of participants where there were multiple publications in the same population. If there was more than one published report of the same group of patients, the articles were analysed to verify whether or not they reported different outcomes. If they presented the same outcomes we extracted the data from the most recent or most complete article.

We used Review Manager (RevMan) (V.5.3; Nordic Cochrane Centre, Cochrane) for all analyses.43

Patient and public involvement

No patients or members of public were involved in this study.

Search selection

The initial searches identified 7210 titles from the electronic searches. After the duplicates, titles were removed, 6775 potentially relevant articles were retained for further assessment (figure 1). Subsequent to reading titles and abstracts, 6715 of these articles were excluded because they were off-topic, in vitro or animal studies. Sixty articles were retrieved for further assessment. After screening the full texts, 11 (one with two publications) RCTs or quasi-RCT44–55 were included in the qualitative synthesis (figure 1).

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Five45 46 48 52 53 55 of the included trials were published in Chinese. Authors of all included studies were contacted for further clarification regarding items of their methodology for our risk of bias analysis, but none of them supplied us with the requested information.

Study characteristics

Table 2 describes study characteristics related to the design of the study, the setting, number of participants, mean age, gender, inclusion and exclusion criteria, and follow-up. Ten45–55 were RCTs, and one44 were quasi-RCT. Nine44–50 52–54 trials employed a parallel two-arm design. Five trials45 46 48 52 53 55 were conducted in China, three47 51 54 in Iran, two44 49 in Thailand, and one50 in France. The trials sample size ranged from 2050 to 12049 patients. Participants were adults with mean ages ranging from 22.3047 to 63.00 years.50

The majority of the eligible studies among the cardiovascular surgical procedures included patients with rheumatic heart disease of ASA grade II–III.45 46 52 55 For the included studies among the obstetrical/gynaecological procedures the most common inclusion criteria were pregnant patients47 54 and ASA grade I or II49 while for the laparoscopic procedures, patients typically enrolled included non-cancer gynaecological conditions.44 Studies followed participants from 2 hours51 to 15 days50 (table 2).

Table 3 describes study characteristics related to type of surgery, intervention and control groups, and measured outcomes. In relation to the type of surgery, seven45 46 48 50–53 55 included studies evaluated patients undergoing cardiovascular surgical (mostly undergoing heart valve replacement), three47 49 54 obstetrical/gynaecological and, one44 laparoscopic procedure.

Among cardiovascular surgery45 46 48 50–53 55 studies, Ginkgo biloba was used in two45 46 50 studies and Astragalus in two,52 55 and herbal medications were mostly used in the form of mixture48 50 52 53 55 or standardised extract.45 46 Five of these studies reported the use of herbal medication via intravenous,45 46 48 52 53 55 with intravenous normal saline45 46 48 52 53 55 as control group. The measured outcome was biochemical analysis45 46 48 50–53 55 (table 3).

The obstetrical/gynaecological surgery procedures studies used Zingiber officinale (ginger)49 54 and in other Rosa damascena (damask rose),47 in the form of powder47 49 and administered via oral.47 49 54 Placebo was used as the control group.47 49 54 None of the included studies assessed conventional treatment or types of complementary and alternative therapy. The measured outcomes evaluated were pain,47 nausea49 54 and vomiting49 54 (table 3).

The only included study44 that evaluated laparoscopic procedure used Zingiber officinale in the form of powder by oral route (capsules), while placebo was used as the control group. The measured outcomes were nausea and vomiting (table 3).

Risk of bias assessment

Figure 2 and table 4 describe the risk of bias assessment. Only the domain blinding of data analyst was rated as high risk of bias in all studies.44–55 However, other domains such as blinding of caregivers,44–46 48 52 53 55 blinding of data collectors44–46 48 50 52 53 55 and blinding of outcome assessment44–46 48 50 52–55 were rated mostly as high risk of bias due to the lack of information in the included studies.

Primary outcomes

Vomiting

Results from three RCTs44 49 54 with a total of 272 participants suggested a statistically significantly reduction in vomiting with the use of Zingiber officinale compared with the control group (ie, placebo and tap water) in both laparoscopic and obstetrical/gynaecological surgery (RR 0.57, 95% CI 0.38 to 0.86; p=0.008; I²=0%, p=0.67) (figure 3). Certainty in evidence was rated down to very low because of risk of bias (due to lack of reporting of allocation concealment,44 lack of blinding of caregivers,44 data collectors,44 data analyst,44 49 54 outcome assessment44 54), indirectness and imprecision (fewer than 300 to 400 events) (table 5).

Table 5

GRADE evidence profile for RCTs: Herbal compared with placebo

VomitingNauseaPainNeed for rescue medication for pain

Quality assessment						Summary of findings
No of participants (studies) Range follow-up time	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Study event rates		Relative risk (95% CI)	Anticipated absolute effects Over 24 hours		Certainty in estimates OR Quality of evidence
						Placebo	Herbal		Placebo	Herbal
272 (3) 4–24 hours	Serious limitation*	No serious limitations	Serious limitations†	Serious imprecision‡	Undetected	42/136	24/136	0.57 (0.38 to 0.86)	466 per 1000	200 fewer per 1000 (288 fewer to 205 fewer)	⊕ΟΟΟ VERY LOW
212 (2) 4–24 hours	Serious limitations§	No serious limitations	Serious limitations†	Serious imprecision‡	Undetected	42/106	29/106	0.69 (0.50 to 0.96)	666 per 1000	207 fewer per 1000 (333 fewer to 27 fewer)	⊕ΟΟΟ VERY LOW
92 (1) 24 hours	Serious limitations¶	Undetected	Serious limitations†	Serious imprecision‡	Undetected	42/46	6/46	0.14 (0.07 to 0.30)	913 per 1000	785 fewer per 1000 (849 fewer to 639 fewer)	⊕ΟΟΟ VERY LOW
272 (3) 6–24 hours	Serious limitations**	Serious limitations††	Serious limitations†	Serious imprecision‡	Undetected	86/136	45/136	0.52 (0.13 to 2.13)	666 per 1000	320 fewer per 1000 (580 fewer to 752 more)	⊕ΟΟΟ VERY LOW

*Serious limitations related to allocation concealment,44 lack of blinding of caregivers,44 data collectors,44 data analyst44 49 54 and outcomes assessment.44 54

†Serious limitations related to surgery where the results are not applicable for cardiac surgery.‡Serious imprecision related to outcome (fewer than 300 to 400 events).

§Serious limitations related to lack of blinding of data analyst,49 54 and outcomes assessment54 and selective outcome reporting.49

¶Serious limitations related to random generation, allocation concealment, lack of blinding of data analyst and selective outcome reporting.47

**Serious limitations related random generation,47 allocation concealment,44 47 lack of blinding of caregivers,44 data collectors,44 data analyst44 47 49 and outcomes assessment,44 selective outcome reporting.47 49

††Serious limitation related to inconsistency (I²=92%).

Figure 3

Meta-analysis comparing herbal versus placebo on vomiting for laparoscopic or obstetrical-gynaecological.

Nausea

Results from two RCT49 54 with a total of 212 participants suggested a statistically significantly reduction in nausea with the use of Zingiber officinale compared with the control group (ie, placebo and tap water) in obstetrical/gynaecological surgery (RR 0.69, 95% CI 0.50 to 0.96; p=0.03; I²=0%, p=0.39) (figure 4). Certainty in evidence was rated down to very low because of risk of bias (due to lack of blinding of data analyst49 54 and outcome assessment,54 selective outcome reporting49), imprecision (fewer than 300 to 400 events), and indirectness in both studies (table 5).

Figure 4

Meta-analysis comparing herbal versus placebo on nausea for obstetrical-gynaecological.

Need for rescue medication for pain

Results from three RCTs44 47 49 with a total of 272 participants suggest a non statistically significantly reduction in the need for rescue medication for pain between Rosa damascena and Zingiber officinale powder capsules compared with placebo in laparoscopic and obstetrical/gynaecological surgery (RR 0.52, 95% CI 0.13 to 2.13; p=0.36; I²=92%, p=0.00001) (figure 5, panel A). A plausible worse case sensitivity analysis excluding Gharabaghi et al47 study yielded results that were consistent with the primary analysis and fail to show a difference in the effects of herbal medications compared with placebo (RR 0.87, 95% CI 0.66 to 1.14; p=0.31; I²=0%, p=0.53; I²=0%) (figure 5, panel B). Certainty in evidence was rated down to very low because of risk of bias (related to random generation,47 allocation concealment,44 47 lack of blinding of caregivers,44 data collectors,44 statistician44 47 49) and outcomes assessment,44 selective outcome reporting,47 49 indirectness, imprecision (fewer than 300 to 400 events), and inconsistency (table 5).

Figure 5

Meta-analysis comparing herbal versus placebo on need for rescue medication for pain. Panel A: primary analysis considering laparoscopic or obstetrical/gynaecological surgeries. Panel B: sensitivity analysis excluding Gharabaghi et al study considering laparoscopic or obstetrical/gynaecological surgeries.

Secondary outcomes

Main findings

From laparoscopic and obstetrical/gynaecological surgeries, based on 212 surgical patients evidence suggests a statistically significant reduction in both vomiting and nausea favouring Zingiber officinale and in the need for rescue medication for pain favouring both Rosa damascena and Zingiber officinale. We also found favourable results for Rosa damascena and Zingiber officinale for pain47 associated with obstetrical/gynaecological surgery, with the overall certainty in evidence rated as very low (table 5).

Regarding the herbal medication Zingiber officinale, it is widely used around the world for nausea, vomiting and motion sickness.44 49 54 In a systematic review that included six RCTs,56Zingiber officinale was evaluated for nausea and vomiting. Three of these RCTs evaluated PONV, with two of them suggesting that Zingiber officinale was superior to placebo and equally effective as metoclopramide (an antiemetic drug). The pooled absolute risk reduction for the incidence of postoperative nausea, however, indicated a non-significant difference between Zingiber officinale (dose: 1 g/day) and placebo when taken prior to surgery (absolute risk reduction 0.05 (95% CI 0.08 to 0.18). These studies collectively favoured Zingiber officinale over placebo.

In another systematic review57 that evaluated Zingiber officinale in the treatment of pregnancy-associated nausea and vomiting, 12 RCTs involving 1278 pregnant women were included. Zingiber officinale was compared with placebo and significantly improved the symptoms of nausea (mean difference (MD) 1.20, 95% CI 0.56 to 1.84, p=0.0002, I²=0%). Zingiber officinale did not significantly reduce the number of vomiting episodes, when compared with placebo, although there was a trend towards improvement (MD 0.72, 95% CI 0.03 to 1.46, p=0.06, I²=71%). Zingiber officinale is thought to act peripherally, within the gastrointestinal tract, increasing the gastric tone and motility due to anticholinenergic and antiserotonergic actions58 and it has also been reported that Zingiber increase gastric emptying.59 These activities may explain the ability of Zingiber officinale to relieve symptoms of gastrointestinal disorders, such as abdominal pain, and nausea, which is often associated with decreased gastric motility.59 There is little available in the literature on potential adverse effects associated with Zingiber officinale, with some data suggesting that its components may be mutagenic.60 61

Based on our findings as well as the results of other systematic reviews,56 57Zingiber officinale has potential as a possible alternative anti-emetic and anti-nausea drug for surgical patients, although this must be verified with further research using standardised forms of the herb with the constituents thought to be most active, for instance, 6-gingerol, 8-gingerol, 10-gingerol, and 6-shogaol.62

In relation to pain, Rosa damascena has been tested in pre-clinical studies63 64 for anti-inflammatory and analgesic properties, and in clinical studies for analgesic and antinociceptive effects.65 66 Similar to our findings, a systematic review67 showed promising evidences for its effectiveness and safety in pain relief. Although these positive findings,63–67 these results must be cautiously interpreted. Rosa damascena presents as a promising indication for the effectiveness in pain relief but more studies are needed. Rosa damascena68 petals infusion has been tested for toxicity and it was well tolerated, showing minimal nephrotoxic or hepatotoxic effects, unless it is used at extreme doses.

Another focus of this manuscript was to assess potential adverse events with the use of herbal medication, but none of the eligible trials reported this information. Considering all the data evaluated in the present study, we reiterate the importance of patients continuing to follow the guidance provided by ASA,31 which was previously described in the introduction, which is to discontinue herbal medications 2 weeks prior to an elective surgery.

There is a general perception that herbal medications or drugs are safe and devoid of adverse effects, but this can be misleading. Caution is needed when dealing with herbal medication, because they have been shown to be capable of producing a wide range of undesirable or adverse reactions such as clinically significant drug interactions which may impact the efficacy of standard and proven medications.69 70.

Strengths and limitations

Strengths of this review include a broad search; evaluation of eligibility, risk of bias, and data abstraction independently and in duplicate; use of the GRADE approach in rating the quality of evidence; and focus on both absolute and relative effects of the intervention on patient important outcomes.

Potential limitations are related to the data available for this topic on the current literature. Trials often had outcomes reported incompletely, inadequate reporting of random sequence generation, and often neglected to blind participants and study personnel due to the nature of the intervention. A second limitation of this review is the fact that we were able to include only eleven trials including 693 patients (364 patients in the meta-analysis), thus limiting the statistical power for some of our pre-defined outcomes and as a result we rated down for imprecision. A third limitation was that the trials that used Zingiber officinale for vomiting and nausea, also presented some heterogeneity in their plant preparation, although all of them were administered orally, Apariman et al44 used 1.5 g of powder capsules; Nanthakomon and Pongrojpaw49 used 1.0 g of powder capsules and Zeraati et al54 used 25 drops of liquid extract. A fourth limitation was the inconsistent standardisation of herbal medications components, which may have introduced variation on therapeutic effects.71 Finally, another limitation of this review that one might also consider the possibility that a gastric content may have played a role in the occurrence of vomiting between Apariman et al44 and Zeraati et al54 studies.

Differences between our PROSPERO protocol and our final review minimal, but included the review only on testing the impact of herbal medicine before surgery to evaluate prophylactic effects on anxiety, depression, pain, nausea and vomiting post intervention. We choose to include only preoperative interventions to minimise the potential interaction with the postoperative medications (eg, anti-emetics, painkillers) on the predefined outcomes.

Implications for clinical practice and for research

There is very low-certainty evidence showing that Zingiber officinale is more effective than placebo for the reduction of vomiting (laparoscopic and obstetrical/gynaecological surgery) and nausea (obstetrical/gynaecological surgery) in patients. Similarly, there is very low-certainty evidence showing that Rosa damascena is more effective than placebo for the reduction of pain in patients undergoing obstetrical/gynaecological surgery. Finally, there is also very low-certainty evidence showing that Rosa damascena and Zingiber officinale are more effective than placebo for reducing the need for rescue medication for pain in laparoscopic and obstetrical/gynaecological surgeries.