Abstract
Two representative Kampo formulas, keishibukuryogan and tokishakuyakusan, are frequently prescribed for patients with dysmenorrhea. We previously constructed a model that could predict which of these 2 formulas was most suitable, which is based on 4 subjective symptoms and 3 objective signs. To evaluate the prognosis of patients with dysmenorrhea using the established prediction model and assess the treatment outcomes between those treated in accordance with the prediction model and those who received various other treatments. In this retrospective, observational study, we included patients with menstrual pain who visited the Kampo Clinic at the Keio University Hospital for the first time between October 2014 and December 2020. These patients were monitored over a 90-day follow-up period. Participants were categorized into 2 groups: model-accordance and various-options. The progression of visual analogue scale (VAS) values was evaluated by determining the slopes from regression analysis between these 2 groups, with changes corroborated by the medical records. The study comprised 57 patients: 37 in the model-accordance group and 20 in the various-options group. Notably, the various-options group reported a significantly higher number of subjective symptoms (P = .03). The VAS value showed a decline, as indicated by the negative slope value of the regression line, across both groups – irrespective of their classification. There were no significant differences in the occurrence of adverse events between the 2 groups. The prognosis of patients with dysmenorrhea and the incidence of adverse events remained consistent, regardless of whether the treatment approach was in accordance with the prediction model or varied. Further studies are warranted to assess the prognosis when Kampo formulas are chosen based on the prediction model in the various-options population.
Keywords: dysmenorrhea, Kampo medicine, Keishibukuryogan, prescription prediction model, prognosis, Tokishakuyakusan
1. Introduction
Dysmenorrhea, prevalent among women of reproductive age, is a common gynecological disorder, transcending racial differences.[1,2] It manifests primarily as pain in the lower abdomen or back, usually commencing before menstruation and persisting through its duration. The release of excess prostaglandins during menstruation, resulting in the contraction and ischemia of the myometrium, has been identified as its primary cause.[3] Dysmenorrhea can be organic (secondary) when linked to diseases like endometriosis, uterine fibroids, and adenomyosis, or functional (primary) when not associated with a specific disease.
Standard interventions for dysmenorrhea encompass nonsteroidal anti-inflammatory drugs (NSAIDs) or oral contraceptives (OCs).[4,5] Still, more than 30% of patients exhibit limited response to NSAIDs, and 10% to 20% of patients unresponsive to both NSAIDs and OCs.[6,7] Given that NSAIDs are contraindicated for patients with gastric and duodenal ulcers, and OCs are unsuitable for those with thrombophilic predisposition, breast cancer, migraine with aura, or pregnancy, alternative therapies have emerged.[8–11]
A notable alternative therapy in Japan is Kampo medicine,[12–15] rooted in ancient Chinese medical traditions but adapted to the Japanese context. This system prescribes medicines based on traditional patients patterns, differentiating its diagnoses from Western medicine.[16] Numerous Kampo formulas have been used to address dysmenorrhea, with keishibukuryogan (KBG, Gui-Zhi-Fu-Ling-Wan, 桂枝茯苓丸) and tokishakuyakusan (TSS, Dan-Gui-Shao-Yao-San, 当帰芍薬散), being primary recommendations.[17,18] Their efficacy has been validated in randomized, placebo-controlled studies.[19,20] Typically, Japan’s national health insurance system covers these fixed-recipe, lower-dosage products prescribed by conventional biomedicine physicians.[21]
KBG comprises Japanese Pharmacopoeia (JP) cinnamon bark, peony root, peach kernel, Poria sclerotium, and Moutan bark.[22] TSS consists of JP Japanese Angelica root, Cnidium rhizome, peony root, Poria sclerotium, Atractylodes lancea rhizome, and Alisma tuber.[22] While both KBG and TSS are prescribed for the same traditional “interior” pattern diagnosis, indicative of internal reactions, such as within bowels and viscera, they cater to distinct sub-patterns. KBG targets patients diagnosed with an excess pattern, tangled cold and heat pattern, yang, and blood stasis; TSS is used in patients diagnosed with a deficiency pattern, cold pattern, yin, blood deficiency, and fluid disturbance. It is necessary to statistically clarify the differences between the patients prescribed these 2 formulations.
In our prior work, we developed a prediction model to determine whether KBG or TSS was more appropriate for patients with dysmenorrhea.[23] The model drew from an e-questionnaire administered to patients, regarding their subjective symptoms, during their initial consultation at Keio University Hospital since 2008. Patients who were treated with either TSS or KBG were asked 128 questions. Using the responses, a logistic regression analysis was conducted for both TSS and KBG groups. We evaluated multicollinearity through the variance inflation factor, ultimately selecting the model with the lowest Akaike’s information criterion. This led to the identification of 4 subjective symptoms: tendency to sweat, leg numbness, cold sensation in the lower back, and lightheadedness, along with 3 objective findings: abdominal strength excess, abdominal strength deficiency, and para-umbilical tenderness/resistance. The model’s performance, as denoted by the area under the ROC curve, exceeded 0.8. Furthermore, the Hosmer–Lemeshow test yielded a P value > .05. Leveraging these 7 factors, our model derived a propensity score ranging between 0 and 1 for both KBG and TSS. Specifically, patients scoring ≥ 0.5 were categorized as “predicted KBG,” while those scoring < 0.5 were labeled as “predicted TSS.” Notably, we observed a congruence exceeding 80% between our model’s predictions and the actual prescriptions given by specialists, albeit with a few discrepancies based on statistical criteria.
Yet, it remains uncertain whether the agreement rate between our prediction model physician prescriptions directly influences patient outcomes. As a response, we’ve been compiling data on the prognosis of the chief complaint and overall symptom burden since 2014.
This study endeavors to evaluate the prognosis of patients with dysmenorrhea based on the retrospective application of our prediction model, laying the groundwork for an ensuing confirmatory study.
2. Methods
2.1. Participants
This study, a subset of a prospective, observational study, included patients with menstrual pain as the chief symptom, as indicated in the e-questionnaire, who visited the Kampo Clinic at Keio University Hospital for the first time between October 2014 and December 2020. The recruitment period overlapped with our previous study by 1 year and 2 months.[23] The follow-up period was set at 90 days or approximately 3 menstrual cycles. Exclusion criteria comprised missing data, age ≥ 50 years, no second consultation within 90 days, and initiation of new treatment, other than Kampo therapy, within 90 days. This study was conducted in compliance with the World Medical Association’s Declaration of Helsinki. All registered participants provided written informed consent and the study design was approved by the appropriate institutional review board at Keio University (Keio University School of Medicine Ethics Committee approval number: 20100144).
2.2. Survey items and patient grouping
The study utilized patient responses from the e-questionnaire, complemented by data from their medical records. Our current protocol employs a tablet-based e-questionnaire to record subjective symptoms of patients, chronological change in the visual analog scale (VAS) score for the chief symptom, physical findings, conventional diagnosis, and the prescribed Kampo formulas. The e-questionnaire consists of 86 symptom-related items recorded during the patient’s initial consultation. Through the questionnaire, we analyzed the following patient characteristics: number of subjective symptoms among the 86 items at the initial consultation, age at initial examination, duration of menstrual cycle (25–38 days, ≤24 days, ≥39 days), bleeding duration (3–7 days, ≤2 days, ≥8 days), infertility status, as well as the cumulative number of deliveries, abortions, and miscarriages. The presence of organic diseases, as well as the status of standard treatments with NSAIDs, OCs, or other analgesics, were confirmed through patient interviews and medical records.
After reviewing each patient’s medical records from the observation period, we applied the prescription-prediction model based on symptoms and findings from their initial consultation, and all patients were classified into model-accordance and various-options groups. Here, the model-accordance group included patients who were prescribed KBG and were predicted for KBG prescription (KBG-KBG) and those who were prescribed TSS and predicted for TSS prescription (TSS-TSS). The various-options group included the rest of the patients.
2.3. Patient prognosis
To determine if the alignment between the physician’s prescription and the prediction model’s results affected patient prognosis, we analyzed 4 key factors: changes in the VAS scores (0–100 mm) for menstrual pain, improvement in perimenstrual symptoms according to the medical records, continuation period of the same Kampo formula, and incidence of adverse events.
2.4. Changes in the VAS scores for menstrual pain
To investigate the variation in the VAS scores, patients were asked to rate their menstrual pain using the VAS in the e-questionnaire at the time of the initial visit and at every subsequent visit. A VAS score of 0 (left-most on the scale) indicated no pain, and a score of 100 (right-most on the scale) indicated the worst imaginable pain. The VAS scores were obtained during the period of continuous prescription of the same medication. We constructed linear regression models and compared the slopes to evaluate the change in the VAS scores for menstrual pain, creating fixed-effects and mixed-effects models with per-patient errors (random effects).
2.5. Improvement in symptoms according to medical records
We examined symptom improvement based on medical records, which encompassed menstrual pain as well as any other gynecological, dermatological, gastrointestinal, and neuropsychiatric symptoms.[24] The duration from the initial hospital visit to when a favorable symptom description was recorded in the medical records was noted, and Kaplan–Meier curves were plotted accordingly.
2.6. Continuation period of the same Kampo formulas
Similarly, to determine how long a particular Kampo formula was used, we calculated the number of days between the initial consultation and when a new Kampo formula was prescribed as per the medical records and plotted Kaplan–Meier curves. Additionally, whenever possible, we examined the medical records to understand the reasons behind medication changes. The reasons categorized into 5 groups: occurrence of adverse events, occurrence of other symptoms, worsening of symptoms, additional prescription for temporary symptoms, and improvement in symptoms.
2.7. Incidence of adverse events
To examine the incidence of adverse events, we calculated the duration between the initial visit and the reported date of adverse event occurrence, using medical records, and plotted Kaplan–Meier curves. Based on our previous study, we examined 7 common adverse events, which include soft stool/diarrhea, constipation, gastric symptoms, worsening of menstrual pain, skin rash, drowsiness, and lightheadedness, among other events.[24]
2.8. Statistical analysis
R software version 4.1.0[25] was used for all statistical analyses. Comparisons of patient backgrounds between the model-accordance and various-options groups were performed using Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for categorical variables. Linear regression models were estimated using the lme4 and lmerTest packages.[26] The types of mixed-effect models created are shown in Additional File 1, Table S1, Supplemental Digital Content, http://links.lww.com/MD/K789. We used the survminer R package[27] to plot Kaplan–Meier curves and conduct log-rank tests for both the model-accordance and various-options groups, respectively. This was done to assess the differences in the agreement rate between the results of the prediction model and the prescriptions of the physician.
Based on the Kaplan–Meier curve for menstrual pain, we determined the necessary sample size. Assuming a 75% improvement rate in menstrual pain in the model-accordance group and a 50% improvement rate in the various-options group, a significance level of α = 0.05, power of β = 0.80, and the same number of patients in both groups, a total of 124 patients was required in the model-accordance and various-options groups. P < .05 was considered to indicate statistical significance.
3. Results
3.1. Study participants
A total of 63 patients, presenting with the chief symptom of dysmenorrhea were registered at the Kampo Specialty Clinic at Keio University (Fig. 1). Upon exclusion of 6 ineligible participants (3 with missing data, 2 without a second consultation within 90 days, and one initiating a new treatment apart from Kampo therapy within 90 days), 57 participants remained eligible. Among these, 10 had participated in the study by Yoshino et al[23] The prediction model results identified 31 patients in the “predicted KBG” group and 26 in the “predicted TSS” group. Furthermore, 37 patients were in the model-accordance groups (KBG-KBG and TSS-TSS), while the various-options groups (KBG-TSS, TSS-KBG, KBG-others, and TSS-others) included 20 patients. Notably, 78.7% of those prescribed either KBG or TSS matched the prescription-prediction-model-accordance group.
Figure 1.
Flowchart of patient exclusion process. KBG = Keishibukuryogan, TSS = Tokishakuyakusan.
The characteristics of the eligible participants are summarized in Table 1. A significant 93% had regular menstrual and bleeding patterns and consumed antipyretic analgesics like NSAIDs and acetaminophen when needed. The various-options groups had a notably higher percentage of patients not undergoing conventional treatment compared to the model-accordance groups (P = .006). Conversely, the model-accordance groups had a significantly higher percentage of patients using both NSAIDs and acetaminophen than the various-options groups (P = .017). No patient was observed using other analgesics, such as opioids. Additionally, the various-options groups recorded a significantly higher number of subjective symptoms compared to the model-accordance groups (P = .032).
Table 1.
Characteristics of patients with dysmenorrhea.
| Patient characteristics | Total (n = 57) | Prediction | Prediction and prescribed | |||
|---|---|---|---|---|---|---|
| KBG (n = 31) | TSS (n = 26) | Model-accordance (n = 37) | Various-options (n = 20) | |||
| Age at first consultation (years) | 32.1 ± 7.1 | 33.1 ± 6.4 | 30.2 ± 7.5 | 31.4 ± 6.9 | 33.1 ± 7.1 | |
| Menstrual cycle | ||||||
| Less than 24 days | 1 (1.8) | 0 (0.0) | 1 (3.9) | 1 (2.7) | 0 (0.0) | |
| 25–38 days | 53 (93.0) | 28 (90.3) | 25 (96.1) | 35 (94.6) | 18 (90.0) | |
| More than 39 days | 3 (5.3) | 3 (9.7) | 0 (0.0) | 1 (2.7) | 2 (10.0) | |
| Bleeding period | ||||||
| Less than 2 days | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
| 3–7 days | 53 (93.0) | 28 (90.3) | 25 (96.1) | 33 (89.2) | 20 (100.0) | |
| More than 8 days | 4 (7.0) | 3 (9.7) | 1 (3.9) | 4 (10.8) | 0 (0.0) | |
| Conventional treatment | ||||||
| No | 7 (12.3) | 6 (19.4) | 1 (3.9) | 1 (2.7) | 6 (30.0) | ** |
| OC | 6 (10.5) | 4 (12.9) | 2 (7.7) | 6 (16.2) | 0 (0.0) | |
| NSAIDs + acetaminophen | 53 (93.0) | 25 (80.7) | 24 (92.3) | 35 (94.6) | 14 (70.0) | * |
| BMI (kg/m²) | 20.4 ± 3.4 | 21.1 ± 3.4 | 19.6 ± 3.2 | 20.4 ± 3.2 | 20.8 ± 3.5 | |
| Diagnosed organic disease | ||||||
| No | 38 (66.7) | 17 (54.8) | 21 (80.8) | 25 (67.6) | 13 (65.0) | |
| Endometriosis | 11 (19.3) | 7 (22.6) | 4 (15.4) | 7 (18.9) | 4 (20.0) | |
| Adenomyosis | 4 (7.0) | 3 (9.7) | 1 (3.9) | 2 (5.4) | 2 (10.0) | |
| Leiomyoma | 9 (15.8) | 7 (22.6) | 2 (7.7) | 5 (13.5) | 4 (20.0) | |
| Ovarian cyst | 8 (14.0) | 6 (19.4) | 2 (7.7) | 5 (13.5) | 3 (15.0) | |
| Infertility (primary and secondary) | 5 (8.8) | 5 (16.1) | 0 (0.0) | 2 (5.4) | 3 (15.0) | |
| Previous deliveries, abortions, and miscarriages | 12 (21.8) | 5 (16.1) | 7 (26.9) | 9 (24.3) | 3 (15.0) | |
| Number of subjective symptoms | 29.2 ± 11.0 | 29.9 ± 12.3 | 28.4 ± 9.3 | 27.4 ± 11.1 | 34.7 ± 10.0 | * |
Values are expressed as numbers (%) or mean ± SD.
BMI = body mass index, KBG = Keishibukuryogan, OC = Oral contraceptives, TSS = Tokishakuyakusan.
P < .01;
P < .05.
3.2. Prescribed Kampo formulas
The Kampo formulas prescribed to the patients are detailed in Additional File 1, Table S2, Supplemental Digital Content, http://links.lww.com/MD/K790. Out of the 57 eligible patients: 21 were prescribed KBG (4 received only KBG, while 7 were given a combination of KBG and other Kampo formulas), 26 were given TSS (with 19 receiving only TSS, and 7 getting a mix of TSS and other Kampo formulas), and 10 were administered combinations of different Kampo formulas. These patients were prescribed various dosage forms, including decoctions and powders, sourced from multiple pharmaceutical companies.
3.3. Results of the prediction model
The correlation between the prediction model results and the prescriptions of physicians is shown in Additional File 2, Figure S1, Supplemental Digital Content, http://links.lww.com/MD/K786. We found no consistent relationship between the agreement of predictions and prescriptions and the predictive score’s value. Discrepancies in predictions occurred even when the prediction score was near the extremes of 1 or 0.
In Additional File 1, Table S3, Supplemental Digital Content, http://links.lww.com/MD/K791, the proportion of patients displaying symptoms associated with each item utilized in the prediction model is presented. A notable symptom, the “tendency to sweat,” indicative of a tangled cold and heat pattern, was prevalent among patients in the KBG-KBG, KBG-others, TSS-KBG, and TSS-others groups. A cold sensation in the lower back was also frequently reported in the KBG-KBG and KBG-others groups. Over half of the patients in the “predicted KBG” groups (KBG-KBG, KBG-TSS, and KBG-others) showed signs of para-umbilical tenderness and resistance, indicative of blood stasis. The symptom of lightheadedness, suggesting fluid disturbance, was reported in approximately half of the patients across all groups.
3.4. Patient prognosis
3.4.1. Changes in the VAS scores for menstrual pain.
To evaluate the changes in the VAS scores for menstrual pain, we analyzed the regression results of the fixed-effects model for both the model-accordance and various-options groups. Data from more than 2 VAS scores for menstrual pain were available for 46 patients (comprising 29 from the model-accordance and 17 in the various-options group). The results of the mixed-effects regression for each graph are presented in Additional File 1, Table S1, Supplemental Digital Content, http://links.lww.com/MD/K789. Models incorporating random effects into the intercept could not be generated. In every regression model, the absolute value of the “day” term – which indicates the slope of the regression line – was larger in the model-accordance group than in the various-options group. The “day” values of the model-accordance and various-options groups in the fixed-effects model were −0.405 and −0.247, respectively. The outcomes of the mixed-effects regression mirrored those of the fixed-effects model (see Additional File 1, Table S1, Supplemental Digital Content, http://links.lww.com/MD/K789).
3.4.2. Improvement in symptoms according to the medical records.
To assess symptom improvement as detailed in the medical records – including menstrual pain and any other gynecological, dermatological, gastrointestinal, and neuropsychiatric symptoms – we constructed Kaplan–Meier curves and conducted log-rank tests. These results are depicted in Figure 2A and B, and the specific improved symptoms are listed in Additional File 1, Table S4, Supplemental Digital Content, http://links.lww.com/MD/K792.
Figure 2.
Improvement and continuation rates based on medical records. (A) Rate of menstrual pain improvement. (B) Rate of Improvement for symptoms other than menstrual pain. (C) Rate of discontinuation of the initial Kampo formula. (D) Incidence rate of adverse events.
While no significant difference emerged in the improvement rate of menstrual pain between the model-accordance and various-options groups, a tendency for a higher rate of improvement was observed in the model-accordance group (Fig. 2A). Comparative results within the “predicted KBG” groups and “predicted TSS” groups are shown in Additional File 2, Figures S2, Supplemental Digital Content, http://links.lww.com/MD/K787 and S3, Supplemental Digital Content, http://links.lww.com/MD/K788, respectively. Across these prediction groups, no notable difference in the improvement rate of menstrual pain was identified.
Similarly, there was no significant difference in the improvement rate of symptoms other than menstrual pain, both between the model-accordance and various-options groups (Fig. 2B) and within the prediction groups (as seen in Additional File 2, Figures S2, Supplemental Digital Content, http://links.lww.com/MD/K787 and S3, Supplemental Digital Content, http://links.lww.com/MD/K788).
3.4.3. Duration of continuation with the initial Kampo formula.
Figure 2C shows the discontinuation rates for the initially prescribed Kampo formula. The reasons for medication alterations are listed in Table 2. Although there was no significant difference in the rate of medication adjustments between the model-accordance and various-options groups, the frequency of medication change tended to be lower in the model-accordance group (as shown in Fig. 2C, and further detailed in Additional File 2, Figures S2, Supplemental Digital Content, http://links.lww.com/MD/K787 and S3, Supplemental Digital Content, http://links.lww.com/MD/K788).
Table 2.
Reasons for prescription changes, and adverse events.
| Total (n = 57) | Prediction | Prediction and prescribed | |||
|---|---|---|---|---|---|
| KBG (n = 31) | TSS (n = 26) | Model-accordance (n = 37) | Various-options (n = 20) | ||
| Reasons for prescription changes | |||||
| Adverse event | 1 (1.8) | 1 (3.2) | 1 (5.0) | ||
| Additional symptoms | 1 (1.8) | 1 (3.8) | 1 (2.7) | ||
| No improvement | 6 (10.5) | 5 (16.1) | 1 (3.8) | 3 (8.1) | 3 (15.0) |
| Improvement | 1 (1.8) | 1 (3.2) | 1 (5.0) | ||
| Adverse events | |||||
| Soft stool/diarrhea | 4 (7.0) | 2 (6.5) | 2 (7.7) | 3 (8.1) | 1 (5.0) |
| Drowsiness | 1 (1.8) | 1 (3.2) | 1 (5.0) | ||
| Constipation | 1 (1.8) | 1 (3.8) | 1 (5.0) | ||
Values are expressed as numbers (%).
KBG = Keishibukuryogan, TSS = Tokishakuyakusan.
One patient, initially prescribed kamishoyosan, switched to TSS after experiencing soft stools and abdominal discomfort. This patient showed improvement in menstrual pain, and shimotsuto was added for anemia which was reported in her initial consultation. Among the patients whose prescriptions were changed due to insufficient symptom relief, KBG was added for 1 patient in the TSS-TSS group, while the remaining 5 had accompanying prescriptions (outside of KBG and TSS) either introduced or modified. For the patient who adjusted their medication due to symptom improvement, the initial combination had been TSS and hochuekkito, and only hochuekkito was discontinued.
3.4.4. Incidence of adverse events.
Figure 2D presents the occurrence of adverse events, further detailed in Table 2. Adverse events reported included soft stool/diarrhea (n = 4; 7.0%), drowsiness (n = 1; 1.8%), and constipation (n = 1; 1.8%). No severe adverse events were observed in this study. Only 1 patient, who experienced soft stool/diarrhea, had their prescription changed. The study found no significant differences in the incidence of adverse events between agreement (as seen in Fig. 2) and prediction groups (refer to Additional File 2, Figures S2, Supplemental Digital Content, http://links.lww.com/MD/K787 and S3, Supplemental Digital Content, http://links.lww.com/MD/K788).
4. Discussion
In this study, we explored the variations in the prognosis of patients with dysmenorrhea after receiving Kampo formula treatment. This analysis focused on the correlation between the outcomes of the prescription-prediction model and the physician’s choice to prescribe either KBG or TSS. Our findings indicate that the prognosis of patients remained consistent regardless of whether their treatments aligned with the model-accordance or fell into the various-options category.
Significantly, none of the prognostic outcomes showcased notable differences. These outcomes included the improvement rate of both menstrual pain and other related symptoms, the duration of continuation with the same Kampo formula, and the incidence of adverse events. A previous review of 51 different clinical trials reported that 82% women reported relief of menstrual pain with NSAIDs.[28] Our study demonstrated that 71.9% (41 out of 57 patients) experienced an alleviation in menstrual pain and related symptoms. This improvement is notable, considering that the majority of these patients had previously used NSAIDs and a third had diagnosed organic diseases. The observed incidence rate of adverse events of 10.5% (6 out of 57 patients) further underscores the safety of Kampo medicine as a therapeutic option. In contrast, prior research indicated an adverse event rate of 27.4% and side effects rate of 2.4% with ibuprofen,[29,30] and an adverse event occurrence of 50.6% with OCs.[31] These comparisons suggest that Kampo formulas, when appropriately prescribed by specialists, can improve patient prognosis irrespective of their alignment with prediction model results. Additionally, Kampo treatments exhibit a lower propensity for adverse events.
We aimed to evaluate the prognosis of patients after receiving a Kampo formula that differed from the model’s recommendation.[23] Notably, KBG and TSS are not diametrically distinct in terms of their component crude drugs or the traditional medicine patterns they target. Both KBG and TSS share 2 common ingredients: peony root and Poria sclerotium. Moreover, both formulas cater to patients with blood stasis and fluid disturbance patterns and have demonstrated anti-inflammatory effects.[2,32–34]
Our revalidation of the prediction model showed an accuracy rate of approximately 80%, using a population similar to a previous study.[23] This rate indicates that the prognosis is analogous regardless of whether it’s based on the model’s prediction or the specialist’s choice of formula. However, this agreement rate does not factor in patients in the various-options group who received Kampo formulas other than TSS and KBG. One possible reason for the mismatch between the results of the prediction model and the prescription by the physician might have been the number of subjective symptoms.[35] The model used in this study takes into account only 4 subjective symptoms, overlooking others during the prediction process. As displayed in Table 1, patients in the various-options group presented with more symptoms than those in the model-accordance group. For patients with multiple complaints in the various-options group, Kampo specialists might have emphasized a broader array of symptoms than the prediction model accounted for, leading them to prescribe a formula that was different from the results of the prediction model. In our previous study, patients diagnosed with a qi stagnation pattern often reported multiple complaints.[36]
Kampo specialists consider a broader range of patient information that just the independent variables (4 symptoms and 3 findings) outlined in the prediction model. They also assess a wide array of potential Kampo formulas, beyond just TSS and KBG, to determine the most suitable prescription for each patient. As shown in Additional File 1, Table S2, Supplemental Digital Content, http://links.lww.com/MD/K790, 13 patients (6 from the model-accordance group out of 37 and 7 from the various-options group out of 20) received prescriptions during their initial consultations that aligned with Kampo formulas for treating qi stagnation, and they were diagnosed with this condition. However, the prognosis for the various-options group was not significantly divergent from that of the model-accordance group. Interestingly, the prognosis trended poorer in the various-options group compared to the model-accordance group. A comparative interventional study between the prediction model and formula selection by Kampo specialists is still required. For a more comprehensive comparison, when designing an ideal various-options group as a control, we should first apply the prediction model to the participants. Subsequently, those recommended for KBG should receive TSS and vice versa.
This retrospective observational study has several limitations that should be noted. First, being an observational study, our data collection was incomplete, especially regarding VAS scores. A subsequent analysis, wherein the missing scores were inputted, hinted at possibly lower treatment efficacy than originally estimated (see Additional File 1, Table S1, Supplemental Digital Content, http://links.lww.com/MD/K789). This suggests that the treatment’s efficacy might have been overestimated if every patient had consistently reported their VAS scores. Second, the study’s sample size was limited and imbalanced. The various-options group was smaller than the model-accordance group, which could raise the risk of a β-error. Third, we included participants who were administered a range of Kampo formulas besides KBG and TSS. This inclusion introduces potential complications in interpreting our results. Our predictive model solely targets 2 main Kampo formulas: KBG and TSS. In clinical scenarios, a myriad of other Kampo formulas are also employed (see Additional File 1, Table S2, Supplemental Digital Content, http://links.lww.com/MD/K790). Furthermore, the study did not account for patient compliance. Fourth, there were discrepancies between changes in VAS scores and medical record descriptions. For instance, in the KBG-KBG group, even with a declining VAS score trend, only 30% of patients had documented symptom improvements in their medical records. In the TSS-TSS group, despite a milder decline in the VAS score, 80% of patients reported improvement in their medical records. This might be attributable to varying definitions or perceptions of “improvement in symptoms.”[37] Given these limitations, future interventional studies are warranted to ascertain any differences in prognoses between the model-accordance and various-options groups more conclusively.
5. Conclusion
The prognosis of patients with dysmenorrhea did not differ based on whether the prescriptions aligned with the prediction model (model-accordance) or diverged from it (various-options) when Kampo formulas were prescribed by specialists. To fully assess the efficacy of Kampo formulas chosen by the prediction model within the various-options group, prospective studies with a larger sample size and a control group are warranted.
Acknowledgments
We would like to thank the physicians who supported data acquisition in the clinic and Editage (www.editage.com) for English language editing.
Author contributions
Conceptualization: Tetsuhiro Yoshino, Kenji Watanabe.
Data curation: Tetsuhiro Yoshino.
Formal analysis: Ayako Maeda-Minami, Ayako Kawamoto.
Funding acquisition: Kenji Watanabe.
Project administration: Tetsuhiro Yoshino, Masaru Mimura.
Resources: Kenji Watanabe.
Supervision: Yuta Yokoyama, Sayo Suzuki, Yuko Horiba, Tomonori Nakamura, Masaru Mimura, Kenji Watanabe.
Validation: Tetsuhiro Yoshino,
Visualization: Ayako Maeda-Minami, Ayako Kawamoto.
Writing – original draft: Ayako Maeda-Minami, Ayako Kawamoto.
Writing – review & editing: Tetsuhiro Yoshino, Yuta Yokoyama, Sayo Suzuki, Tomonori Nakamura, Kenji Watanabe.
Supplementary Material
Abbreviations:
- BMI
- body mass index
- KBG
- Keishibukuryogan
- NSAID
- nonsteroidal anti-inflammatory drug
- OC
- oral contraceptive
- TSS
- Tokishakuyakusan
- VAS
- visual analogue scale
This work was supported by a Grant-in-Aid for Research on the Propulsion Study of Clinical Research from the Ministry of Health, Labor, and Welfare to build the questionnaire and collect data. Masaru Mimura received research grant support from Tsumura & Co. and Kracie Pharmaceuticals. The patient recruitment fees for some patients, an article-processing charge, and an English editing fee for this article were paid from the joint research program fund of Keio University and Tsumura Co. The funders were not involved in the study design, collection, analysis, and interpretation of data, writing of the article, or decision to submit it for publication.
This study was conducted in compliance with the World Medical Association’s Declaration of Helsinki, and the study design was approved by the appropriate institutional review board at Keio University (Keio University School of Medicine Ethics Committee approval number: 20100144). All registered participants provided written informed consent to participate in the study.
All registered participants provided written informed consent to the publication of the study results.
Tetsuhiro Yoshino was employed for the joint research program by Tsumura & Co. Kenji Watanabe and Yuko Horiba received lecture fees from Tsumura & Co. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. All other authors have no competing/conflict of interest.
The datasets generated during and/or analyzed during the current study are not publicly available, but are available from the corresponding author on reasonable request.
Supplemental Digital Content is available for this article.
How to cite this article: Maeda-Minami A, Kawamoto A, Yoshino T, Yokoyama Y, Suzuki S, Horiba Y, Nakamura T, Mimura M, Watanabe K. Patient prognosis and prediction model for taking Kampo formulas in dysmenorrhea: An observational study. Medicine 2023;102:48(e36191).
Contributor Information
Ayako Maeda-Minami, Email: aya.kwmt0713@keio.jp.
Ayako Kawamoto, Email: aya.kwmt0713@keio.jp.
Yuta Yokoyama, Email: yokoyama-yt@pha.keio.ac.jp.
Sayo Suzuki, Email: suzuki-sy@pha.keio.ac.jp.
Yuko Horiba, Email: mannta217@keio.jp.
Tomonori Nakamura, Email: nakamura-tm@pha.keio.ac.jp.
Masaru Mimura, Email: mimura.a7@keio.jp.
Kenji Watanabe, Email: watanabekenji@keio.jp.
References
- [1].McKenna KA, Fogleman CD. Dysmenorrhea. Am Fam Physician. 2021;104:164–70. [PubMed] [Google Scholar]
- [2].Zhang S, Lai X, Wang X, et al. Deciphering the pharmacological mechanisms of Guizhi-Fuling capsule on primary dysmenorrhea through network pharmacology. Front Pharmacol. 2021;12:613104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Ferries-Rowe E, Corey E, Archer JS. Primary dysmenorrhea: diagnosis and therapy. Obstet Gynecol. 2020;136:1047–58. [DOI] [PubMed] [Google Scholar]
- [4].Harel Z. Dysmenorrhea in adolescents and young adults: an update on pharmacological treatments and management strategies. Expert Opin Pharmacother. 2012;13:2157–70. [DOI] [PubMed] [Google Scholar]
- [5].Osayande AS, Mehulic S. Diagnosis and initial management of dysmenorrhea. Am Fam Physician. 2014;89:341–6. [PubMed] [Google Scholar]
- [6].Proctor M, Farquhar C. Diagnosis and management of dysmenorrhoea. BMJ. 2006;332:1134–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Oladosu FA, Tu FF, Hellman KM. Nonsteroidal antiinflammatory drug resistance in dysmenorrhea: epidemiology, causes, and treatment. Am J Obstet Gynecol. 2018;218:390–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Fukai-Watabe S, Ishibashi S, Watabe T, et al. Premenstrual syndrome and dysmenorrhea successfully managed with the Japanese traditional medicine Tokito: a case report and literature review. Tradit Kampo Med. 2021;9:61–6. [Google Scholar]
- [9].Guo Y, Liu FY, Shen Y, et al. Complementary and alternative medicine for dysmenorrhea caused by endometriosis: a review of utilization and mechanism. Evid Based Complement Alternat Med. 2021;2021:6663602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Su KH, Su SY, Ko CY, et al. Ethnopharmacological survey of traditional Chinese medicine pharmacy prescriptions for dysmenorrhea. Front Pharmacol. 2021;12:746777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Wang Y, Xu J, Zhang Q, et al. Immediate analgesic effect of acupuncture in patients with primary dysmenorrhea: a fMRI study. Front Neurosci. 2021;15:647667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Proctor ML, Murphy PA. Herbal and dietary therapies for primary and secondary dysmenorrhoea. Cochrane Database Syst Rev. 2001;3:CD002124. [DOI] [PubMed] [Google Scholar]
- [13].Latthe PM, Champaneria R. Dysmenorrhoea. BMJ Clin Evid. 2014;2014:0813. [PMC free article] [PubMed] [Google Scholar]
- [14].Isomura M, Sagawa K, Tobe C, et al. Pregnant women with the complication of depression who visit the department of oriental medicine at a Japanese perinatal center. Tradit Kampo Med. 2020;8:55–9. [Google Scholar]
- [15].Suzuki S, Obara T, Ishikawa T, et al. Prescription of Kampo formulations for pre-natal and post-partum women in Japan: data from an administrative health database. Front Nutr. 2021;8:762895. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Wu X, Le TK, Maeda-Minami A, et al. Relationship between conventional medicine chapters in ICD-10 and Kampo pattern diagnosis: a cross-sectional study. Front Pharmacol. 2021;12:751403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Oya A, Oikawa T, Nakai A, et al. Clinical efficacy of Kampo medicine (Japanese traditional herbal medicine) in the treatment of primary dysmenorrhea. J Obstet Gynaecol Res. 2008;34:898–908. [DOI] [PubMed] [Google Scholar]
- [18].Pan JC, Tsai YT, Lai JN, et al. The traditional Chinese medicine prescription pattern of patients with primary dysmenorrhea in Taiwan: a large-scale cross sectional survey. J Ethnopharmacol. 2014;152:314–9. [DOI] [PubMed] [Google Scholar]
- [19].Kotani N, Oyama T, Sakai I, et al. Analgesic effect of a herbal medicine for treatment of primary dysmenorrhea – a double-blind study. Am J Chin Med. 1997;25:205–12. [DOI] [PubMed] [Google Scholar]
- [20].Sun WH, Zhao L, Tian XP, et al. Clinical observation on treatment of 90 patients with primary menalgia with guizhi fuling capsule. Zhongguo Zhong Xi Yi Jie He Za Zhi. 2004;24:1121–3. [PubMed] [Google Scholar]
- [21].Yoshino T, Kashio A, Terasawa Y, et al. The integration of traditional medicine with conventional biomedicine: a narrative review of the Japanese perspective. J Integr Complement Med. 2023;29:372–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Table of the links of Kampo product informations in Japanese Pharmacopoeia (JP) and/or Package Insert. In: Research Center for Medicinal Plant Resources. Available at: http://mpdb.nibiohn.go.jp/kconsort/kconsort.html [accessed December 21, 2021]. [Google Scholar]
- [23].Yoshino T, Katayama K, Horiba Y, et al. The difference between the two representative Kampo formulas for treating dysmenorrhea: an observational study. Evid Based Complement Alternat Med. 2016;2016:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Horiba Y, Yoshino T, Watanabe K. Effectiveness of Japanese Kampo treatment in dysmenorrhea: single-center observational study. Tradit Kampo Med. 2017;5:51–5. [Google Scholar]
- [25].R software version 4.1.0. Available at: http://www.r-project.org. [accessed November 30, 2021].
- [26].Bates D, Maechler M, Bloker B, et al. Package “lme4” linear mixed-effects models using “Eigen” and S4. Available at: https://cran.microsoft.com/snapshot/2015-07-26/web/packages/lme4/lme4.pdf 2015. [accessed November 30, 2021].
- [27].Kassambara A, Kosinski M, Biecek P, Fabian S. Package “survminer” drawing survival curves using “ggplot2.” Available at: https://cran.r-project.org/web/packages/survminer/survminer.pdf [accessed November 30, 2021].
- [28].Owen PR. Prostaglandin synthetase inhibitors in the treatment of primary dysmenorrhea Outcome trials reviewed. Am J Obstet Gynecol. 1984;148:96–103. [DOI] [PubMed] [Google Scholar]
- [29].Furey SA, Waksman JA, Dash BH. Nonprescription ibuprofen: side effect profile. Pharmacotherapy. 1992;12:403–7. [PubMed] [Google Scholar]
- [30].Kellstein DE, Waksman JA, Furey SA, et al. The safety profile of nonprescription ibuprofen in multiple-dose use: a meta-analysis. J Clin Pharmacol. 1999;39:520–32. [PubMed] [Google Scholar]
- [31].Wong CL, Farquhar C, Roberts H, et al. Oral contraceptive pill as treatment for primary dysmenorrhoea. Cochrane Database Syst Rev. 2009;2009:CD002120. [DOI] [PubMed] [Google Scholar]
- [32].He DY, Dai SM. Anti-inflammatory and immunomodulatory effects of Paeonia lactiflora pall, a traditional Chinese herbal medicine. Front Pharmacol. 2011;2:10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Ríos JL. Chemical constituents and pharmacological properties of Poria Cocos. Planta Med. 2011;77:681–91. [DOI] [PubMed] [Google Scholar]
- [34].Zhang Y, Su N, Liu W, et al. Metabolomics study of Guizhi Fuling capsules in rats with cold coagulation dysmenorrhea. Front Pharmacol. 2021;12:764904. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Zhao S, Wu W, Kang R, et al. Significant increase in depression in women with primary dysmenorrhea: a systematic review and cumulative analysis. Front Psychiatry. 2021;12:686514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Wickström K, Edelstam G. Minimal clinically important difference for pain on the VAS scale and the relation to quality of life in women with endometriosis. Sex Reprod Healthc. 2017;13:35–40. [DOI] [PubMed] [Google Scholar]
- [37].Yoshino T, Katayama K, Yamaguchi R, et al. Classification of patients with cold sensation by a review of systems database: a single-centre observational study. Complement Ther Med. 2019;45:7–13. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.