Inclusion criteria

Trial participants of any sex and age, diagnosed with chronic hepatitis B, as defined according to the trialists or according to guidelines (see Description of the condition; AASLD 2016; EASL 2017). In addition to chronic hepatitis B, trial participants could have had cirrhosis, hepatocellular carcinoma, concomitant HIV or AIDS, hepatitis C, hepatitis D, or any other concomitant disease.

Exclusion criteria

No exclusion criteria were applied.

Experimental intervention

• Acupuncture

Acupuncture, performed via any of the techniques below, regardless of the style used by the acupuncturist (e.g. traditional Chinese medicine; Japanese, Korean, Western medical) or the treatment regimens provided (e.g. numbers and names of acupoints, depth of insertion, stimulation retention time, times in a day).

• • Manual needle acupuncture: thin needles are inserted at a specific location of the body, known as 'acupoints'. Needling is usually performed perpendicularly or with a tilt of 15 to 45 degrees to a depth of 0.5 to 1 inch, accompanied with or without needle flicking or rotation. Needle retention is commonly allowed from 20 to 30 minutes after insertion of the last one. Manual needle acupuncture is usually administered once or twice a day (Wang 2012; Bardy 2015).

  • Electroacupuncture: the inserted needles are attached to an electric needle instrument (e.g. G6850‐type) that generates continuous electric pulses with small clips to adjust impulse frequency and intensity. Usually, stimulation lasts for 20 to 30 minutes under a continuous wave (e.g. 2 Hz) (NGC 2016; Xu 2018).

  • Laser acupuncture: this technique allows low‐level lasers to stimulate energy and blood circulation within the body. Continuous laser light at a set wavelength (e.g. 685 nm) is used as painless laser needles. Output power may vary from 20 mW to 90 mW per laser needle, and stimulation time may usually last 10 to 40 minutes (Litscher 2007; Jiang 2017).

  • Acupressure: acupressure uses a similar principle to acupuncture. Commonly, it is performed by physical pressure, which can be applied by hand, elbow, or other devices to acupoints and may last from 3 to 20 minutes, once to twice daily (Yazdanpanahi 2017; Ahmedov 2018).

  • Acupoint injection: a development of acupuncture, this treatment injects drugs into acupoints. Acupoint injection is usually implemented with antiviral injections (e.g. peginterferon alfa‐2b 180 μg once a week) (Liang 2011; Lee 2012; Zhang 2012; Jing 2016), or through traditional Chinese medicine injections (e.g. Xiang Dan injection 4 mL once daily) (Zhang 2005; Li 2008; Zhang 2015).

  • Moxibustion: this technique burns or places dry herb moxa (Folium Artemisia argy or mugwort) or another mixture of herbs or therapeutic materials directly or indirectly on acupoints of the body. For indirect moxibustion, an approximate 2‐cm‐long moxa stick is burned at 1 to 2 cm above the skin of the acupoint. A local warm feeling should be obtained (Wang 2017; Kuge 2018). For direct moxibustion, moxa cones are burned directly on the acupoints (Yun 2016; Schlaeger 2018).

  • Acupoint herbal patching: through herbal decoction made into a paste, or grinded herbal powders mixed with water, vinegar, wine, egg white, honey, and vegetable oil, or with solidified oil (e.g. vaseline) or with yellow vinegar into a paste, or ointment, or a small pie, acupoint herbal patching directly covers the acupoint or affected area (Ashi points) from three hours to 24 hours, once daily (Zhang 1990; Hsu 2010; Lee 2016).

Control intervention

• No intervention or sham acupuncture

We allowed co‐interventions in both experimental and control intervention groups only if the co‐intervention was administered equally to all intervention groups.

Primary outcomes

• All‐cause mortality

• Proportion of participants with one or more serious adverse events, that is, any untoward medical occurrences that result in death, are life‐threatening, require hospitalisation or prolongation of existing hospitalisation, result in persistent or significant disability or incapacity, or present as a congenital anomaly or birth defect (ICH‐E2A 1994; ICH‐GCP E6(R2) 2016)

• Health‐related quality of life: any scale used by trialists to assess participants' reporting of their quality of life

Secondary outcomes

• Hepatitis B‐related mortality

• Hepatitis B‐related morbidity (proportion of participants with one or more of the following events: cirrhosis, ascites, variceal bleeding, hepato‐renal syndrome, hepatocellular carcinoma, hepatic encephalopathy, or needed liver transplantation, and who have not died)

• Propotion of participants with one or more adverse events considered not to be serious: any untoward medical occurrence in a participant that does not meet the above criteria for a serious adverse event is defined as an adverse event considered not to be serious (ICH‐E2A 1994; ICH‐GCP E6(R2) 2016)

Allocation sequence generation

• Low risk of bias: the study authors performed sequence generation using computer random number generation or a random numbers table. Drawing lots, tossing a coin, shuffling cards, and throwing dice were adequate if an independent person not otherwise involved in the study performed them

• Unclear risk of bias: the study authors did not specify the method of sequence generation

• High risk of bias: the sequence generation method was not random. We planned to include such studies only for assessment of harms

Allocation concealment

• Low risk of bias: the participant allocations could not have been foreseen in advance of, or during, enrolment. A central and independent randomisation unit controlled allocation. The investigators were unaware of the allocation sequence (e.g. whether the allocation sequence was hidden in sequentially numbered, opaque, and sealed envelopes)

• Unclear risk of bias: the study authors did not describe the method used to conceal the allocation, so the intervention allocations may have been foreseen before, or during, enrolment

• High risk of bias: it is likely that investigators who assigned the participants knew the allocation sequence. We planned to include such studies only for assessment of harms

Blinding of participants and personnel

• Low risk of bias: either of the following: blinding of participants and key study personnel ensured, and it was unlikely that the blinding could have been broken; or rarely, no blinding or incomplete blinding, but the review authors judged that the outcome was not likely to be influenced by lack of blinding, such as mortality

• Unclear risk of bias: either of the following: insufficient information to permit judgement of 'low risk' or 'high risk'; or the study did not address this outcome

• High risk of bias: either of the following: no blinding or incomplete blinding, and the outcome was likely to be influenced by lack of blinding; or blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome was likely to be influenced by lack of blinding.

Blinding of outcome assessment

• Low risk of bias: either of the following: blinding of outcome assessment ensured, and unlikely that the blinding could have been broken; or rarely, no blinding of outcome assessment, but the review authors judged that the outcome measurement was not likely to be influenced by lack of blinding such as mortality

• Unclear risk of bias: either of the following: insufficient information to permit judgement of 'low risk' or 'high risk'; or the study did not address this outcome

• High risk of bias: either of the following: no blinding of outcome assessment, and the outcome measurement was likely to be influenced by lack of blinding; or blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement was likely to be influenced by lack of blinding

Incomplete outcome data

• Low risk of bias: missing data were unlikely to make treatment effects depart from plausible values. The study used sufficient methods, such as multiple imputation, to handle missing data

• Unclear risk of bias: information was insufficient for assessment of whether missing data in combination with the method used to handle missing data were likely to induce bias on the results

• High risk of bias: the results were likely to be biased due to missing data

Selective outcome reporting

• Low risk of bias: the trial reported the following predefined outcomes: all‐cause mortality; serious adverse events; and health‐related quality of life outcomes. If the original trial protocol was available, the outcomes should be those called for in that protocol. If the trial protocol was obtained from a trial registry (e.g. www.clinicaltrials.gov), the outcomes sought were those enumerated in the original protocol if the trial protocol was registered before or at the time that the trial was begun. If the trial protocol was registered after the trial was begun, those outcomes were not considered to be reliable

• Unclear risk of bias: not all predefined outcomes were reported fully, or it was unclear whether or not data on these outcomes were recorded

• High risk of bias: one or more predefined outcomes were not reported

Other bias

• Low risk of bias: the study appeared to be free of other factors that could put it at risk of bias

• Unclear risk of bias: the study may or may not have been free of other factors that could put it at risk of bias

• High risk of bias: other factors in the study could put it at risk of bias

Overall risk of bias

• Low risk of bias: we planned to classify the outcome result at overall 'low risk of bias' only if we could classify all of the bias sources described in the above paragraphs at 'low risk of bias'

• High risk of bias: we planned to classify the outcome result at 'high risk of bias' if we could classify any of the risk of bias sources described above at 'unclear risk of bias' or 'high risk of bias'

We tried to reach consensus through discussion. We planned that JPL would arbitrate in cases of disagreement.

Meta‐analysis

We performed the analyses according to the instructions provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), as well as the Cochrane Hepato‐Biliary Group Module. We analysed data using Review Manager 5 (Review Manager 2014).

We assessed our intervention effects via both fixed‐effect and random‐effects model meta‐analyses, and we reported results of both when results differed (e.g. one giving a significant intervention effect, the other no significant intervention effect). We put greater weight on the estimate closest to the zero effect (the highest P value) (Jakobsen 2014).

We assessed three primary outcomes with P ≤ 0.025 as significant, and three secondary outcomes with P ≤ 0.025 as significant to secure a family‐wise error rate below 0.05 (Jakobsen 2014). For exploratory outcomes, we considered P < 0.05 as significant because we view these outcomes as only hypothesis‐generating outcomes. Whether we presented our data synthesis as a meta‐analysis or in a narrative way depended on our assessment of the statistical and clinical heterogeneity of the meta‐analysed trial data per comparison.

We did not impute any missing data in our primary analysis; however, we planned to impute missing values in our sensitivity analysis of continuous and dichotomous data (see Sensitivity analysis; Jakobsen 2014).

We planned to use Fisher's exact test for dichotomous data (Fisher 1922), as well as Student's t‐test for continuous data when data from only one trial were available (Student 1908).

Trial Sequential Analysis

As cumulative meta‐analysis involves risk of producing random errors due to sparse data and repetitive testing, we performed Trial Sequential Analysis. To control random errors, we calculated the required information size (i.e. the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect) (Wetterslev 2008; Thorlund 2011a; TSA 2011). The required information size calculation should also account for the diversity present in the meta‐analysis (Wetterslev 2008; Wetterslev 2009; Wetterslev 2017). A more detailed description of Trial Sequential Analysis can be found at www.ctu.dk/tsa (Thorlund 2011a; TSA 2011).

We controlled the risks of type I errors and type II errors for both dichotomous and continuous outcomes (Brok 2008; Wetterslev 2008; Brok 2009; Wetterslev 2009; Thorlund 2010; Castellini 2017; Wetterslev 2017). For dichotomous outcomes, we estimated the diversity‐adjusted required information size (DARIS) based on the event proportion in the control group, a relative risk reduction of 15%, an alpha of 2.5% for primary and secondary outcomes and 5.0% for exploratory outcomes, a beta of 10% (Castellini 2017), and diversity suggested by the trials in the meta‐analysis (Wetterslev 2009; Jakobsen 2014). We intended to include participants with different severity of chronic hepatitis B, but it happened that no participants in the control group died. Therefore, we conducted three post‐hoc Trial Sequential Analyses based on assumed proportions of participants dying being low (4%; young participants with mild disease), moderate (20%; middle‐aged participants with mild disease), and high (40%, middle‐aged participants with severe disease) within one year in the control group. For continuous outcomes, we intended to estimate the DARIS based on the SD observed in the control group, a minimal relevant difference of 50% of this SD, an alpha of 2.5%, a beta of 10% (Castellini 2017), and diversity suggested by the trials in the meta‐analysis (Wetterslev 2009; Jakobsen 2014).

We tested statistical significance using trial sequential monitoring boundaries for benefit and harm, along with futility using futility boundaries (Thorlund 2011a). If the Z‐curve crosses the trial sequential monitory boundaries for benefit or harm before reaching DARIS, the effect of the intervention is considered superior or inferior to that of the control intervention. If the Z‐curve crosses the futility monitoring boundaries before reaching the DARIS, this would mean that the intervention does not possess the postulated effect, and further randomised trials might be futile. Furthermore, if the trial sequential monitoring boundaries are not surpassed, and if the trial monitoring boundaries for futility are not crossed, it is probably necessary to continue conducting trials to detect or reject a certain intervention effect (Wetterslev 2008; Thorlund 2011b). In our cases where the monitoring boundaries were not reached, we also displayed the Trial Sequential Analysis‐adjusted CI.

Summary of findings

We constructed a 'Summary of findings' table to show our results and confidence in the evidence for Primary outcomes and Secondary outcomes. We displayed information on assumed control group risk, corresponding intervention group risk, relative effect, MD, CI, statistical significance of relative effect, number of participants, and quality of the evidence. We calculated the corresponding risk (and its 95% CI) using the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). Using GRADEpro GDT software (community.cochrane.org/help/tools‐and‐software/gradepro‐gdt), we assessed five factors of the evidence referring to limitations in the study design and implementation that suggested the quality of evidence: within‐study risk of bias, indirectness of the evidence (population, intervention, control, outcomes), unexplained heterogeneity or inconsistency of results (including problems with subgroup analyses), imprecision of results, and risk of publication bias (GRADEpro GDT; Balshem 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g; Guyatt 2011h; Guyatt 2013a; Guyatt 2013b; Guyatt 2013c; Guyatt 2013d; Mustafa 2013; Guyatt 2017).

We classified the evidence as follows.

• High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

• Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.

• Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.