Skip to main content

The effect of berberine on insulin resistance in women with polycystic ovary syndrome: detailed statistical analysis plan (SAP) for a multicenter randomized controlled trial



Although Traditional Chinese Medicine (TCM) has been widely used in clinical settings, a major challenge that remains in TCM is to evaluate its efficacy scientifically. This randomized controlled trial aims to evaluate the efficacy and safety of berberine in the treatment of patients with polycystic ovary syndrome. In order to improve the transparency and research quality of this clinical trial, we prepared this statistical analysis plan (SAP).


The trial design, primary and secondary outcomes, and safety outcomes were declared to reduce selection biases in data analysis and result reporting. We specified detailed methods for data management and statistical analyses. Statistics in corresponding tables, listings, and graphs were outlined.


The SAP provided more detailed information than trial protocol on data management and statistical analysis methods. Any post hoc analyses could be identified via referring to this SAP, and the possible selection bias and performance bias will be reduced in the trial.

Trial registration

This study is registered at, NCT01138930, registered on 7 June 2010.

Peer Review reports


This paper provides the detailed statistical analysis plan (SAP) for “The effect of berberine on insulin resistance randomized, placebo-controlled trial,” comparing glucose disposal rate (GDR) at week 12. Secondary endpoints and GDR at other time points will provide exploratory results.


Polycystic ovary syndrome (PCOS) is an ovarian disorder characterized by hyperandrogenism, ovulatory dysfunction, and polycystic ovaries [1]. It is the most common endocrinopathy affecting women of reproductive age, and has attracted a lot of public attention for its relative high prevalence both in communities and clinical settings [2, 3]. Berberine is the major active component of Rhizoma Coptidis, with a broad array of pharmacological effects [4]. Recently, several studies have shown that berberine has positive effects on type 2 diabetes mellitus, insulin resistance, lipid metabolism, nitric oxide production, and the metabolic syndrome [58]. This study is a multicenter, double-blind, randomized, placebo-controlled clinical trial. The purpose is to compare the insulin resistance in PCOS patients with and without the use of berberine. This study has been approved by the Ethics Committee of the First Affiliated Hospital Heilongjiang University of Chinese Medicine (2010HZYLL-012) and informed consent was obtained from each patient before any study procedures were performed. Full details of the trial design of this randomized controlled trial (RCT) are given in the study protocol which has been registered and published [9]. Patients will be enrolled at four hospitals in mainland China as stated in the protocol.

Despite Traditional Chinese medicine (TCM) having been increasingly used in clinical settings, a major challenge in TCM is to evaluate its efficacy scientifically [10]. The results of several studies have indicated the low quality of RCTs in TCM [11, 12]. For example, a recent study on bias of risk and outcome-reporting in RCTs revealed inconsistencies between information from trial registration and subsequent publications [13]. The inconsistencies occurred mostly in statistical methods, sample size, primary outcomes, safety reporting, and outcome assessor blinding.

In the study protocol we detailed the overall design and approaches. For a clinical trial, bias may also be introduced in the post hoc selection of data management procedures and the statistical analysis methods. The SAP will provide more transparent information for data analysis methods than the protocol. Besides, the possible dispute and queries can be resolved by the published SAP when publishing trial results. These strategies and practices are important to improve the scientific validity in the clinical research of Chinese Medicine interventions. In order to improve the transparency and research quality of our trial, we prepared this SAP.

Inclusion criteria

  1. 1.

    Women aged between 18 and 35 years

  2. 2.

    Confirmed diagnosis of PCOS according to the modified Rotterdam criteria, and all patients must be anovulatory plus have either polycystic ovaries and/or hyperandrogenism

  3. 3.

    Body Mass Index equal to or greater than 23 kg/m2

  4. 4.

    Women must not be pregnant and not expect to become pregnant within 6 months

Exclusion criteria

  1. 1.

    Patients who are treated within the past 3 months with other medications known to affect reproductive function or metabolism, including oral contraceptives, gonadotropin- releasing hormone (GnRH) agonists and antagonists, antiandrogens, gonadotropins, antiobesity drugs, Chinese herbal medicines, antidiabetic drugs, such as metformin and thiazolidinediones, somatostatin, diazoxide, angiotensin-converting enzyme (ACE) inhibitors, and calcium channel blockers

  2. 2.

    Patients with other endocrine disorders including 21-hydroxylase deficiency, hyperprolactinemia, uncorrected thyroid disease, or suspected Cushing’s syndrome

  3. 3.

    Patients with known severe organ dysfunction or any mental illness

Sample size

It is assumed that women with PCOS would have an insulin resistance comparable to women with type 2 diabetes. Based on a previous study [4], GDR was 7.42 ± 2.37 mg/(kg × min) in the berberine group and 6.06 ± 2.21 mg/(kg × min) in the control group. To detect this difference between the two arms, the sample size required was 52 per group, given a two-sided α = 0.05 (type I error) and β = 0.1 (type II error).

Randomization and allocation concealment

Eligible participants had been randomized into each of the two arms: berberine (0.5 g, three times per day) or placebo. Berberine or placebo medications were administrated orally for 12 weeks. Patients, investigators, and physicians were blind to the assignment.

Data management

From the beginning of this study, a Data Management Group has been formed (J Sun, YJ Zhang, K Zhang). They were involved in setting up studies, collecting or entering data, cleaning data, and managing accrual data until the study could be considered ready for analysis. The data in Case Report Form (CRF) all came from the original files and the consistencies between the two were monitored periodically. Data should be timely, accurately, and integrally and veritably filled in the CRF. The web-based data management system was also available at:

Guidelines for analysis

This SAP aims to detail the presentation and analysis methods for the main paper(s) reporting results of this study. Therefore, biases related to selective reporting from our study could be reduced. International Council on Harmonization (ICH) guidance on statistical principles for clinical trials (E9) [14] and the Consolidated Standards of Reporting Trials (CONSORT) Statement [15] for the reporting results were followed.

Statistical analysis set

Intention-to-treat set

The intention-to-treat (ITT) set is defined as all patients in the intervention (berberine or placebo) arms to which they were randomized, regardless of the availability of data at follow-up, and the real intervention that they received during the period of the trial. Patients will be included in the group to which they were randomized.

The ITT analysis will be reported mainly for efficacy evaluation of primary and secondary outcomes. In addition, ITT will be used in the balance testing of demographic data, disease history, and laboratory examinations at baseline.

Per-protocol set

The per-protocol set (PPS) includes patients who have completed prescribed treatments in the whole trial period without severe protocol violation.

The severe protocol violations will include (but not be limited to) the following conditions:

  1. 1.

    Not meeting the inclusion criteria

  2. 2.

    The use of concomitant medication which may add confounding to the estimation of efficacy and safety

  3. 3.

    Being beyond the time window, loss to follow-up, or dropout

Any patients with varying degrees of protocol violation will be confirmed in the blinded data review meeting.

Efficacy analyses will be carried out for both ITT and PPS. If the results of the ITT and PPS are inconsistent, sensitivity analyses will be conducted, and possible causes will be investigated. Subgroup analyses by patients’ characteristics, such as age, weight, Body Mass index, waist/hip circumference ratio, etc. will be carried out.

Safety set

The safety set (SS) includes the patients who received at least one-time medication treatment. The SS is the main set of safety assessment in this trial.

Baseline demographic characteristics

Discrete variables will be summarized by frequencies and percentages. Continuously distributed variables will be summarized using either mean ± standard deviation (SD) for data with normal distribution, or median and interquartile range for nonnormally distributed data.

Primary outcome and hypothesis

The primary outcome is the change of GDR from baseline to week 12. The GDR is defined as the amount of glucose required to maintain stable blood glucose concentrations during the last 30 min of the hyperinsulinemic-euglycemic clamping. The mean value of change (μ), will be computed. The relative efficacy of berberine on increasing GDR compared to placebo will be tested at the one-sided significance level of 0.025 under the null hypothesis and alternative hypothesis as below:

  • H0: μ Berberine = μ Placebo,

  • H1: μ Berberine  > μ Placebo.

Statistics relevant to the primary outcome, including eligible number, mean, SD, median, interquartile range, and minimum and maximum, will be calculated. A paired Student’s t test/signed rank test will be used to compare the difference between baseline and post treatment within each group. The Analysis of Covariance (ANCOVA) model will be constructed in estimating the difference of GDR between baseline and post treatment. Covariates will be the baseline value of GDR considering the effects of different sites and groups. Based on this model, we will calculate least squares means (LSMEANS) of the difference accompanying 95 % confidence interval between the two groups. To explore the consistency of results across centers, the interaction between treatment effect and centers or groups will be added into the model. The interaction effect will be considered statistically significant when the P value for the interaction test is equal to or less than 0.10. If the interaction is statistically significant, further descriptive investigation will be conducted in subgroups by sites.

Secondary outcomes and statistical methods

The secondary outcomes will be analyzed based on both ITT and PPS:

  1. 1.

    Oral glucose tolerance test (OGTT): serum for glucose, insulin, and c-peptide levels will be determined.

    The outcomes of the OGTT will be analyzed by the t test/Wilcoxon rank sum test. The trapezoidal method will be applied to calculate the area under the curve (AUC) of OGTT values at different time points [16] using the following formula:

$$ \left({X}_{0 \min }+{X}_{180 \min}\right)/2 + {X}_{30 \min }+{X}_{60 \min }+{X}_{120 \min }. $$
  1. 2.

    Ovarian androgen biosynthesis as measured by human chorionic gonadotropin (hCG), stimulated production of 17-hydroxyprogesterone (17-OHP), androstedione (A2), and testosterone (T) levels

  2. 3.

    Hormonal profile including: testosterone, sex hormone-binding globulin (SHBG), follicle-stimulating hormone (FSH), luteinizing hormone (LH), and dehydroepiandrosterone sulfate (DHEAS) levels

  3. 4.

    Fasting lipid metabolic profile: cholesterol, triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) levels

  4. 5.

    Weight, waist/hip circumference ratio, blood pressure, Ferriman-Gallwey Score, and severity of acne before and after treatment

Student’s t test or the Wilcoxon rank sum test will be used to test the difference of the above outcomes between groups according to the distribution of variables.

Safety outcomes and statistical methods

Safety outcomes, including vital signs, adverse events, and renal and liver function (including three categories: normal, abnormal without clinical significance, and abnormal with clinical significance), will be analyzed based on the SS.

Shift tables will be constructed to map the changes of abnormal renal and liver function test results in each visit. The vital signs comparison between groups at baseline and post treatment and the changes relative to baseline will be described with mean, SD or median and interquartile range and Student’s t test or the Wilcoxon rank sum test will be used accordingly. Adverse events will be listed separately by type (nonserious adverse events, serious adverse events), by visit, and by real intervention.

Handling of missing data

The imputation will be carried out for primary and secondary outcomes only. When analyzing the efficacy in ITT, the Last Observation Carried Forward (LOCF) imputation method will be used.

Interim analysis

No interim analysis was planned. All results and conclusions will be derived from the final analysis of this trial.

CONSORT flow diagram

The profile of patients will be summarized in a CONSORT flow diagram.







Angiotensin-converting enzyme


Analysis of Covariance


Area under the curve


Consolidated Standards of Reporting Trials


Case Report Form


Dehydroepiandrosterone sulfate


Follicle-stimulating hormone


Glucose disposal rate


Gonadotropin-releasing hormone


Human chorionic gonadotropin


High-density lipoprotein cholesterol


The International Council for Harmonization




Low-density lipoprotein cholesterol


Luteinizing hormone


Last Observation Carried Forward


Least mean squares


Oral glucose tolerance test


Polycystic ovary syndrome


Per-protocol set


Randomized controlled trial


Statistical analysis plan


Standard deviation


Sex hormone-binding globulin


Safety set




Traditional Chinese Medicine




  1. Legro RS. Evaluation and treatment of polycystic ovary syndrome [Last update 19 September 2009]. In: De Groot LJ, Beck-Peccoz P, Chrousos G, et al., editors. Endotext [Internet]. South Dartmouth:, Inc.; 2000. Accessed 20 Sept 2016.

  2. Joham AE, Teede HJ, Ranasinha S, Zoungas S, Boyle J. Prevalence of infertility and use of fertility treatment in women with polycystic ovary syndrome: data from a large community-based cohort study. J Womens Health (Larchmt). 2015;24(4):299–307.

    Article  Google Scholar 

  3. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO. The prevalence and features of the polycystic ovary syndrome in an unselected population. J Clin Endocrinol Metab. 2004;89(6):2745–9.

    Article  CAS  PubMed  Google Scholar 

  4. Huang ZJ, Zeng Y, Lan P, Sun PH, Chen WM. Advances in structural modifications and biological activities of berberine: an active compound in traditional Chinese medicine. Mini Rev Med Chem. 2011;11(13):1122–9.

    Article  CAS  PubMed  Google Scholar 

  5. Zhang Y, Li X, Zou D, Liu W, Yang J, Zhu N, et al. Treatment of type 2 diabetes and dyslipidemia with the natural plant alkaloid berberine. J Clin Endocrinol Metab. 2008;93(7):2559–65.

    Article  CAS  PubMed  Google Scholar 

  6. Lee YS, Kim WS, Kim KH, Yoon MJ, Cho HJ, Shen Y, et al. Berberine, a natural plant product, activates AMP-activated protein kinase with beneficial metabolic effects in diabetic and insulin-resistant states. Diabetes. 2006;55(8):2256–64.

    Article  CAS  PubMed  Google Scholar 

  7. Xu MG, Wang JM, Chen L, Wang Y, Yang Z, Tao J. Berberine-induced upregulation of circulating endothelial progenitor cells is related to nitric oxide production in healthy subjects. Cardiology. 2009;112(4):279–86.

    Article  CAS  PubMed  Google Scholar 

  8. Affuso F, Mercurio V, Ruvolo A, Pirozzi C, Micillo F, Carlomagno G, et al. A nutraceutical combination improves insulin sensitivity in patients with metabolic syndrome. World J Cardiol. 2012;4(3):77–83.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Li Y, Ma H, Zhang Y, Kuang H, Ng EH, Hou L, et al. Effect of berberine on insulin resistance in women with polycystic ovary syndrome: study protocol for a randomized multicenter controlled trial. Trials. 2013;14:226.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Ma Y, Sun S, Peng CK. Applications of dynamical complexity theory in traditional Chinese medicine. Front Med. 2014;8(3):279–84.

    Article  PubMed  Google Scholar 

  11. Yang GY, Luo H, Liao X, Liu JP. Chinese herbal medicine for the treatment of recurrent miscarriage: a systematic review of randomized clinical trials. BMC Complement Altern Med. 2013;13:320.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Su CX, Yan LJ, Lewith G, Liu JP. Chinese herbal medicine for idiopathic sudden sensorineural hearing loss: a systematic review of randomised clinical trials. Clin Otolaryngol. 2013;38(6):455–73.

    Article  PubMed  Google Scholar 

  13. Liu JP, Han M, Li XX, Mu YJ, Lewith G, Wang YY, et al. Prospective registration, bias risk and outcome-reporting bias in randomised clinical trials of traditional Chinese medicine: an empirical methodological study. BMJ Open. 2013;3(7):e002968.

  14. The International Council for Harmonization (ICH): E9 Statistical principles for clinical trials. 1998. Accessed 20 Sept 2016.

  15. Moher D, Hopewell S, Schulz KF, Montori V, Gøtzsche PC, Devereaux PJ, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. Int J Surg. 2012;10(1):28–55.

    Article  PubMed  Google Scholar 

  16. AI-Fozan H, AI-Futaisi A, Morris D, Tulandi T. Insulin responses to the oral glucose tolerance test in women of different ethnicity with polycystic ovary syndrome. J Obstet Gynaecol Can. 2005;27(1):33–7.

    Article  Google Scholar 

Download references


We would like to express our gratitude to Prof. Fujian Song, the Professor of the Department of Population Health and Primary Care of the University of East Anglia, who reviewed the manuscript both for statistics and language. We also deliver many thanks to Suping Lang, the Vice president of GCP ClinPlus Co., Ltd., who gave important professional input for statistical analyses models in the revising stage.


Jian-Ping Liu was funded by Research Projects in 2012 for the National Clinical Trial Base in TCM, a multicentre, placebo-control randomized trial for berberine/tanshinone in women with polycystic ovary syndrome (Grant number: JDZX2012037). Xiao-Ke Wu was funded by Research Projects in 2012 for the National Clinical Trial Base in TCM, a multicentre, placebo-control randomized trial for berberine/tanshinone in women with polycystic ovary syndrome (Grant number: JDZX2012036) and the Program for “Leading Excellent Innovative Talents” of Heilongjiang University of Chinese Medicine. The efficacy and molecular mechanism of berberine/tanshinone for women with polycystic ovary syndrome (2012–2017).

Authors’ contributions

JPL and XKW conceived and designed the study. YZ drafted and critically revised the manuscript, and in addition will perform data analysis for the primary and secondary outcomes. JS drafted the data management plan and will undertake the statistical analysis for other outcomes in this trial. YJZ, QYC, and KZ undertook data cleaning and will participate in data analysis. HLM was responsible for query resolution and the integrity of the data. All authors read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Author information

Authors and Affiliations


Corresponding authors

Correspondence to Xiao-Ke Wu or Jian-Ping Liu.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, Y., Sun, J., Zhang, YJ. et al. The effect of berberine on insulin resistance in women with polycystic ovary syndrome: detailed statistical analysis plan (SAP) for a multicenter randomized controlled trial. Trials 17, 512 (2016).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: