Table 2 Summary of randomised controlled trials for five herbal medicines in oligo/amenorrhoea, hyperandrogenism and PCOS

Kilicdag [63]

Randomised comparative effectiveness trial.

Eighty women, 40 with hyperprolactin-aemia, 40 with cyclical mastalgia.

Comparison of difference between Vitex agnus-castus and Bromocriptine for serum prolactin concentration on days 5–8 of the menstrual cycle.

Mean prolactin concentration before and after in the V.agnus-castus arm; 946mIU/L (±173.5) to 529mIU/l (±279.7), p < 0.0001. In the Bromocriptine arm; 885.0 mIU/l (±177.5) to 472.68mIU/L (±265.6), p < 0.0001.

All participants completed the trial. Adverse reactions; zero reported in V. agnus-castus group; 12.5% of participants reported adverse reactions in the Bromocriptine group (nausea and vomiting).

Treatment for 3 months.

Normal range 25.2mIU/l - 628.5 mIU/l.

Equivalence demonstrated for the significant reduction of serum prolactin for V. agnus-castus and Bromocriptine (P = 0.96).

Small sample sizes with 2 sub-groups. Insufficiently powered to correctly identify the effects; 377 participants were required (±5%, 95% confidence).

Gerhard I, Patek A, et al. [61]

Randomised, double blind, placebo controlled trial.

Ninety-six women with fertility disorders and confirmed infertility (2 years).

Spontaneous menstruation, luteal phase length, serum hormone concentrations and pregnancy rates.

Non-significant improvement in clinical parameters in 57.6% of women in treatment group versus 36.0% in placebo group, P = 0.069.

Numbers too small for statistical significance in clinical outcomes.

Three months. Follow up at 2 years

Secondary amenorrhoea, n = 38; luteal insufficiency, n = 31; idiopathic infertility, n = 27.

Hormonal data from 32 cases. In the third treatment month 66 complete data sets were available.

In a subgroup of women with luteal insufficiency (n = 21) there were significant improvements in clinical parameters in the treatment group compared to placebo (p = 0.023).

Preparation ‘Mastodynon’ contains V agnus-castus plus other herbal extracts which may have confounded outcome measures.

Reasons were as follows; 4 due to drug reactions and 15 due to pregnancy.

15 women conceived in the treatment group compared to 8 in placebo group in the first 3 months (while women were treated).

Inconsistencies in data assessment include the recommendation for treatment with Mastodynon over 3–6 months yet it was tested for 3 months.

Four withdrew for unknown reasons.

All pregnant women were withdrawn from the study. 4 women had miscarriages, all in the active arm. After 2 years there were 21 more pregnancies with 2 miscarriages – evenly spread over active and placebo groups.

Women with infertility were included in this study however data from women who conceived were excluded. This may have led to an underestimation of treatment effect (type 1 error).

Bergmann J, Luft B, et al. [62]

Randomised, placebo controlled double blind study. Three months or 3 menstrual cycles.

Women with fertility disorders, (n = 67). Two sub-groups.

Primary outcome for participants with amenorrhoea: at least one spontaneous menses.

Oligomenorrhoeic subgroup - clinical outcomes were significantly improved in the treatment arm at 82% compared to 45% in placebo arm P = 0.021. When the amenorrheic group were included in analysis, differences were not significant p = 0.19.

Diagnosis for anovulatory amenorrhoea is not well described. Non-statistically significant take home baby rates were complicated by insufficient sample size. 366 patients are required to have a 95% chance, as significant at the 5% level, an increase in take home baby rates from 6% in the placebo group to 18% in the experimental group. The authors conclude that this preparation may be useful if given 3–6 months, yet they only tested for 3 months.

1.oligomenorrhoea, n = 37

For progesterone <1 ng/mL: an increase to >5 ng/mL at the end of 3rd cycle

Mid luteal progesterone concentration in oligomenorrhoeic sub-group was significantly higher than the placebo group p = 0.0479

2. amenorrhoea n = 30. Oligomenorrhoea group: Treatment n = 17. Placebo n = 20. Amenorrhoea group. Treatment n = 16. Placebo n = 14.

For oligomenorrhoea: Shortened menstrual cycle of at least 4 days. Earlier ovulation of at least 3 days. For anovulatory oligomenorrhoea: Mid luteal progesterone increase (>50% 5–10 days before menstruation. Secondary clinical outcomes, pregnancy rates and take home baby rates.

At 6 months following conclusion of treatment, the take home baby rate with treatment was 18.7% compared to 6.4% in placebo group. Not statistically significant.

Milewicz A, Gejdel E, et al. [64]

Randomised placebo controlled, double blind, trial. Three months.

52 women with latent hyperprolactinaemia and luteal phase defects. Participants stratified for cycle length, height (cm) and weight (kgs) and randomised. Baseline differences between arms were not significant p = 0.63, p = 0.48 and p =0.37 respectively. 37 complete case reports: Treatment arm n = 17, placebo n = 20.

Serum prolactin concentration at 15 and 30 minutes following intra venous TRH (200mcg) stimulation. Luteal phase length, number of days. Measurements on menstrual cycle days 5 to 8 and 20 for FSH, LH, oestradiol, progesterone, DHEAs, thyroid stimulating hormone (TSH), T3, T4, testosterone.

No significant changes in prolactin before and after in either group.

In this study 52 women were eligible to participate, statistical analyses were performed on data from 37 women.

Length (number of days) of the luteal phase before and after; treatment group 3.4 (±5.0) to 10.5 (±4.3) (p < 0.005), placebo 3.4 (±5.1) to 5.5 (±5.2), p = 0.22.

There is missing data due to the presence of luteinised, unruptured follicles (9 women). These data were not included in analyses. Six women did not present for further investigation.

Mid luteal (day 20) serum progesterone concentration before and after; treatment arm 2.46 (±0.70) to 9.69 (±6.34), p < 0.001. Placebo 1.99 (±0.65) to 2.34 (±0.59) p = 0.08.

No description of the distribution of drop-outs or missing data. This suggests the potential imbalance between intervention and control and a possible over-exaggeration for treatment effect.

Mid-cycle oestradiol; treatment arm 131.6 (±25.0) to 151.6 (±25.4), p < 0.05. Placebo: 119.5 (±26.0) to 131.1 (±33.2) p = 0.22. Pregnancies in treatment group n = 2.

Intention to treat analysis was not performed.

Unaccounted confounding factors include medications, fertility status, duration of latent hyperprolactinaemia.

Shahin et al. [65]

Randomised controlled trial using with an active control arm for comparative effectiveness. One menstrual cycle.

147 women aged less than 35 years with un-explained infertility and recurrent clomiphene resistance for ovulation induction. Anovulatory participants were excluded (n = 28). Anovulation was diagnosed by serum oestradiol < 200 ng/ml and absence of a dominant ovarian follicle on day 9 of the menstrual cycle. Complete data sets available for 119 women.

Pregnancy rate measured as increasing serum human chorionic gonadotropin (HCG) over two days. Clinical pregnancy defined as detection of gestational sac with embryonic heart-beat. Endometrial thickness measured by ultrasound concurrent with follicle maturation monitoring. Number of days to ovulation (trigger injection) Serum concentration for FSH oestradiol and LH. Luteal progesterone measured on days 21–23 of the menstrual cycle. Miscarriage and multiple pregnancy rates.

Pregnancy rate in clomiphene alone group was 20.3% and 43.3% in the clomiphene plus Cimicifuga racemosa group (P < 0.01). Clinical pregnancy rate in the combination group was 36.7% versus 13.6% in the clomiphene alone group (P < 0.01). Endometrial thickness in combination group was 8.9 (±1.4) versus 7.5 (±1.3) (p < 0.001). Days to ovulation in clomiphene alone group was 13.0 ± 1.1 and in the clomiphene plus Cimicifuga racemosa group 14.2 ± 1.3 (n.s.). Luteal progesterone peak (ng/ml) in combination group was 13.3 (±3.1) versus 9.3 (±2.0) in clomiphene alone group (p < 0.01). All other hormone measures were not significantly different

No detailed current baseline criteria for other causes of infertility. Confounding factors include current male fertility status. This may have caused an imbalance between the two groups. There is no description of the distribution of excluded (anovulatory) participants between groups.

Kamel [67]

Randomised controlled trial with an active control group. Comparative effectiveness trial for ovulation induction in women with PCOS. Three menstrual cycles.

Women aged 21–27 with primary or secondary infertility. Diagnosis of PCOS by ultrasound and clinical history (n = 100). Gynaecology outpatient clinic. Two groups. Group one (n = 50) received Clomiphene citrate 100 mg days 2–7 of the menstrual cycle; group two (n = 50) received 20 mg Cimicifuga racemosa for days 2–12 of the menstrual cycle.

Serum measurements during follicular phase for FSH, LH and FSH:LH ratio. Mid luteal progesterone. Ultrasound observation of endometrial thickness. Pregnancy rates including twin pregnancies. Adverse events including hyperstimulation.

Positive outcomes for Cimicifuga racemosa compared to clomiphene for reduced day 2–5; LH (p = 0.007) and improved FSH to LH ratio (p = 0.06), mid luteal progesterone (p = 0.0001), endometrial thickness (p = 0.0004). Pregnancy rates were higher in the Cimicifuga racemosa group (7/50 compared to 4/50) but not statistically significant (p = 0.1). Adverse events (4 women) and twin pregnancy’s (two women) were not significantly different between groups.

No detail for diagnostic criteria for PCOS. Confounding fertility factors not described. Drop-out reasons were not reported seven in Cimicifuga racemosa group and four in clomiphene group.

Shahin [68]

Non-blinded randomised controlled trial.

Women with PCOS and infertility, n = 194.

Primary outcomes pregnancy rates. Secondary outcomes:

Pregnancy rates were 33 out of 192 cycles (17.2%) for the clomiphene alone group and 71 out of 204 cycles (34.8%) for the clomiphene plus Cimicifuga racemosa group.

Non-blinding compromised the internal validity of the findings in this study. Confounding variables include variations in participant’s and clinicians attitudes and may have led to differences which were unaccounted for between the two groups. However the outcomes are objective with a statistically powered sample size.

Three menstrual cycles each separated by two months of no treatment.

Two groups matched for demographics, age, BMI, primary and secondary infertility and duration of infertility (months). Treatment arm n = 96, control n = 98.

1. Number of days to ovulation (trigger injection). Follicular maturation monitored by ultrasound.

Number of days to trigger injection was 15 (±1.7) for the clomiphene alone group and 12.0 (±1.9) in the clomiphene plus Cimicifuga racemosa group (p = 0.01)

Measures for miscarriages are based on per cycle are not valid. Miscarriages per pregnancy are of greater relevance.

Randomisation for 206 women 12 were excluded due to failure to respond (treatment group n = 7, control n = 5).

2. Endometrial thickness monitored by ultrasound.

Endometrial thickness in the clomiphene alone group was 8.5 mm (±1.9) compared to 12.9 (±2.3) in the clomiphene plus Cimicifuga racemosa group (p < 0.001).

The miscarriage rate per pregnancy for the clomiphene alone group was 5 out of 33 (15.2%) and 6 out of 71 (8.5%) in the clomiphene plus Cimicifuga racemosa group.

3. Serum hormones during follicular phase oestradiol, LH and FSH. Luteal progesterone measured day 21–23 of the cycle.

Serum LH was 8.0 (±0.9) in the clomiphene group and 5.7 (±0.9) in the clomiphene plus Cimicifuga racemosa group (p < 0.001) and oestradiol was 228.3 (±30.2) in the clomiphene alone group and 299.5 (±38.9) \in the clomiphene plus Cimicifuga racemosa group (p = 0.01)

4. Pregnancy outcomes for early miscarriage.

Miscarriages were 5 out of 192 cycles in the clomiphene group and 6 out of 204 cycles in the clomiphene plus Cimicifuga racemosa group (n.s.).

Wang et al. 2008 [66]

Double blinded, placebo controlled randomised trial (pilot). Eight weeks.

15 overweight women with oligo/amenorrhoea and polycystic ovaries on ultrasound. Mean body mass index 28.8 ± 1.3 kg/m2. Mean age 31.1 ± 2.0 years

Primary outcomes: Insulin resistance and sensitivity. Secondary outcomes oestradiol and testosterone concentration. Body mass index (BMI). Before and after treatment comparisons between randomised groups plus comparison between treatment group and normal ovulatory, normal weight women. Adverse events.

Improved insulin sensitivity (QUICKI) in the treatment group. 0.35 to 0.38, (7.7%) p < 0.03. Insulin resistance (HOMO-IR) significantly reduced in treatment group 2.57 to 1.43 (44.5%) p < 0.03. Controls no change insulin sensitivity or insulin resistance. No change in either group for BMI, testosterone and oestradiol. Differences between Cinnamomum cassia group and normal weight and ovulatory controls were not significant. (P < 0.17). No reported adverse reactions.

Small pilot study, the authors report that larger studies are required to confirm findings. Small sample size may explain non-significant comparison with normal weight and ovulating women. Reproductive outcomes were unchanged in this study however the duration of the study was insufficient to demonstrate reproductive changes.