Participants were current smokers who wanted to quit smoking. They were selected from among patients who had undergone initial examination but had not yet begun treatment at the SC outpatient departments at the National Hospital Organization (NHO) facilities or from among outpatients who had missed their SC treatment for at least one year after their last treatment. The inclusion criteria were as follows: (1) depressive tendency with a score between 39 and 59 on the self-rating depression scale (SDS) test [20,21,22] (questionnaire test for depression assessment); (2) nicotine dependence score of at least 5 on the Fagerstrom Test for Nicotine Dependence (FTND) [23,24,25]; (3) not currently undergoing pharmacotherapy at a department of psychiatry or taking psychosomatic medicine; and (4) age 20–80 years at the time of obtaining consent. Patients with a score of 53 or more on the SDS test or those clinically judged to require consultation by their attending physician were referred to a department of psychiatry/psychosomatic medicine to check whether they required pharmacotherapy. Patients were excluded from this study regardless of the SDS test score when psychiatry/psychosomatic physicians determined that they required drug administration. Patients were also excluded from this study based on a pre-trial examination and when their attending physician found that they met at least one of the following conditions: (1) administration of Yokukansan would be inadvisable based on the patient’s general condition (e.g., shock, current severe infectious disease, decompensated congestive heart failure, clinical findings that clearly indicated interstitial pneumonia, end-stage cancer, marked gastrointestinal weakening etc.); (2) complications, such as severe liver disease (Child–Pugh classification, stage C), end-stage kidney disease (creatinine clearance < 15 mL/min), and uncontrolled endocrine disease; (3) consistent use of herbal medicines containing licorice and glycyrrhizic acid; (4) history of Yokukansan allergy; (5) current symptoms of allergy to a medication; (6) current pregnancy, breast-feeding, or intention to become pregnant during the study period; (7) participation in this study would be inappropriate for any other reason (e.g., inability to understand the contents of the study because of dementia).

This study was a multicenter, placebo-controlled, double-blind, randomized, parallel-group comparison trial. The study protocol was approved by an ethical committee (the NHO Headquarters Central Review Board) on May 13 2016 (H26-EBM(Intervention)-02). A total of 12 NHO facilities participated in this study. Before the start of the trial, the researcher in charge of statistics used a random number table to assign numbers to either Yokukansan or placebo and shared the results with the researcher in charge of group assignments, who assigned numbers alone to the boxes containing the packets of the test drug (Yokukansan or placebo). Thus, any other persons should not be able to know which one is which. The statistician was not involved in any enrollment, data entry, or data management of the clinical trial and was blinded.

The attending physicians provided detailed written explanations of the contents of this trial to all patients who fulfilled the inclusion criteria. Written informed consent was obtained from all participants. The study subject data were then entered into the electronic data capture system at the data center of the NHO headquarters, which handled patient enrollment in this trial. Immediately after enrollment, the electronic data capture system randomly assigned the subjects to the two groups (Yokukansan or placebo) in a 1:1 ratio using the dynamic allocation method, after which the attending physicians were informed of the number of trial drugs they had to administer to the patients. Based on the number indicated by the data center, the attending physicians administered the trial drugs without knowing whether they were administering Yokukansan or placebo. The group assignments were adjusted at the time of enrollment for confounding factors (age, sex, the FTND score, the SDS score, and drug use during the initial SC treatment) using the minimization method on dynamic allocations.

The active study medication, Yokukansan (Tsumura Yokukansan Extract Granules for Ethical Use, TJ-54), [15] and placebo were purchased from Tsumura & Co. (Tokyo, Japan). Yokukansan 7.5 g contains dry extract 3.25 g of mixed crude drug of the following seven medicinal herbs: (1) Atractylodes lancea De Candolle (Compositae), (2) Poria cocos Wolf (Polyporaceae), (3) Cnidium officinale Makino (Umbelliferae), (4) Uncaria rhynchophilla Miquel (Rubiaceae), (5) Angelica acutiloba Kitagawa (Umbelliferae), (6) Bupleurum falcatum Linné (Umbelliferae), and (7) Glycyrrhiza uralensis Fisher (Leguminosae). This product also contains Japanese Pharmacopoeia Lactose Hydrate and Japanese Pharmacopoeia Magnesium Stearate. The placebo granules are composed of Japanese Pharmacopoeia Lactose Hydrate, Japanese Pharmacopoeia Corn Starch, Japanese Pharmacopoeia Dextrin, Japanese Pharmacopoeia Magnesium Stearate, Caramel, Yellow No. 4 Aluminum Lake, Iron Sesquioxide, Blue No. 1 Aluminum Lake, and identical in appearance.

With the start of the SC treatment based on the standard procedures, [2] administration of the trial drug (Yokukansan or placebo) was also initiated. This timing was designated as zero time point. The trial drugs were administered orally at a dose of one packet per administration, thrice per day (7.5 g/day), before breakfast, lunch, and dinner, or between meals, for 12 weeks. In subjects over 75 years of age and those with a bodyweight ≤ 50 kg, the dose was reduced to one packet per oral administration, twice per day (5.0 g/day), before breakfast and dinner, or between meals. Medication compliance was confirmed by recording the number of remaining packets for each subject at the end of the trial. The subjects were considered compliant if 70% or more of the prescribed doses had been administered.

Standard SC therapy has been performed as previously described [2,3,4]. The patients were treated with transdermal nicotine patches or oral varenicline, and examined on their first visit and two, four, eight, and twelve weeks after the first visit. On their repeated visits, maintenance of SC was checked, and specific advice regarding the continuation of SC was given by a nurse and a doctor.

The FTND is a world standard test used to assess the physical dependence on nicotine [20,21,22]. The following tests were conducted during the pre-administration screening and at the 12-week time point: the SDS test (test to measure depression) [4, 23,24,25] and the POMS-short form test (test of mood profile) [26, 27]. The subjects self-completed the test questionnaires as much as possible. If a subject failed to respond to an item or indicated clearly incorrect data, the subject was interviewed orally by a medical professional to confirm the data.

The SDS test has been employed in numerous clinical research studies on SC treatment and is reliable and valid [4, 23,24,25]. The POMS-short from test is commercially available, and the necessary parts were purchased for this study (Kanebo Shobo Co., Tokyo, Japan). It has six subscales: tension-anxiety (score range, 0–20), depression-dejection (score range, 0–17 for men, 0–19 for women), anger-hostility (score range, 0–18 for men, 0–19 for women), fatigue (score range, 0–20), vigor (score range, 0–20), and confusion (score range, 0–17 for men and 0–18 for women). The POMS-short form test is reliable and valid, and higher scores reflect higher levels of the construct being measured by the subscale [26, 27].

Each participant’s height, weight, and blood pressure were measured before taking Yokukansan or the placebo and again at the end of 12 weeks. They were also asked whether they had stopped smoking. Their exhaled carbon monoxide concentration was subsequently measured after the 12-week period by a nurse using the Pico-plus SmokerlyzerR (Harada Cooperation, Osaka, Japan).

The primary endpoint was the SC treatment’s success rate at the 12-week time point. This success rate (in %) was calculated by dividing the number of subjects who successfully stopped smoking by the total number of study subjects; “successful cessation of smoking” meant a subject had not smoked even a single cigarette over the past seven days and had an expiration CO concentration of 7 ppm or less.

The following were the two secondary endpoints: (1) the degree of change in the SDS and POMS scores from the pre-drug administration screening to the 12-week time point (assessment of post-SC depression), and (2) adverse events (AEs) during the trial drug administration period.

In our previous study, the SC treatment’s success rate was 35% for patients with an SDS score of 39 or more; however, it was 69% for patients with an SDS score of less than 39 [4]. To determine the target sample size based on these data, we assumed a success rate of 35% for the placebo group and estimated that the success rate should increase by 12.5% for the Yokukansan group. Using a one-tailed significance level of 5% and a power of 80% yielded a target sample size of 210 cases in each group. Assuming 4%–5% ineligible cases, our final enrollment target for each of the two groups was 220, or a total of 440 cases.

All statistical analyses were performed by a professional statistician using the Statistical Package for Social Sciences software program, version 24.0, for Windows (IBM Japan, Ltd., Tokyo, Japan). The normality of data was confirmed using the Shapiro–Wilk test. Continuous parametric data were expressed as the mean (standard deviation [SD]). Various baseline data were compared between the placebo and Yokukansan groups using Fisher’s exact probability test, the independent t-test, or the Mann–Whitney U test (Table 1). An intention-to-treat criterion was used to compare the rate of successful SC at the 12-week time point. AE frequencies between the two groups were compared using Fisher’s exact probability test (Table 2). For patients with good adherence to the trial drugs, clinical data, including the SDS and POMS scores (all scales) at the baseline and the 12-week time points, were compared using the paired-sample t-test in each treatment group separately (Tables 3 and 4). Additionally, concerning changes in data from the baseline to the 12-week time point, we analyzed the interaction between the placebo and Yokukansan groups using a two-way analysis of variance (Table 5). Statistical significance was set at P < 0.05.