Polycystic Ovarian Syndrome: Marked Differences Between Endocrinologists and Gynaecologists in Diagnosis and Management
Andrea J. Cussons; Bronwyn G. A. Stuckey; John P. Walsh; Valerie Burke; Robert J. Norman
Clin Endocrinol. 2005; 62 (3): 289-295. Â©2005 Blackwell Publishing
Background: Women with polycystic ovarian syndrome (PCOS) commonly consult endocrinologists or gynaecologists and it is not known whether these specialty groups differ in their approach to management.
Objective: To compare the investigation, diagnosis and treatment practices of endocrinologists and gynaecologists who treat PCOS.
Design and Setting: A mailed questionnaire containing a hypothetical patient’s case history with varying presentations oligomenorrhoea, hirsutism, infertility and obesity was sent to Australian clinical endocrinologists and gynaecologists in teaching hospitals and private practice.
Results: Gynaecologists were less likely to assess glucose homeostasis but more likely to use a glucose tolerance test to do so. Diet and exercise were chosen by most respondents as first-line treatment for all presentations. However, endocrinologists were more likely to use insulin sensitizers, particularly metformin, for these indications. In particular, for infertility, endocrinologists favoured metformin treatment whereas gynaecologists recommended clomiphene.
Conclusions: There is a lack of consensus between endocrinologists and gynaecologists in the definition, diagnosis and treatment of PCOS. As a consequence, women may receive a different diagnosis or treatment depending on the type of specialist consulted.
Polycystic ovarian syndrome (PCOS) is a common disorder affecting 510% of women of reproductive age.[1,2] Hyperandrogenism and chronic anovulation are characteristic features, but there has been much debate as to the essential diagnostic criteria.[3-5] PCOS is also reported to be associated with obesity, insulin resistance and type 2 diabetes, dyslipidaemia, hypertension, cardiovascular disease and endometrial carcinoma.[6-8] The treatment of PCOS is controversial, as few randomized controlled trials have been conducted. Women with PCOS commonly consult endocrinologists or gynaecologists, and it is not known whether the management of the disorder differs between these specialties. We surveyed endocrinologists and gynaecologists on their usual practice in the investigation, diagnosis and treatment of PCOS.
Materials and Methods
A questionnaire was sent to all clinical members of the Endocrine Society of Australia ( n = 334) and to 334 gynaecologists (selected randomly from the membership of the Royal Australian and New Zealand College of Obstetrics and Gynaecology) in October 2002. An index case was presented as follows: ‘A 23 years old woman presents with longstanding oligomenorrhoea. She has had a period approximately once every 6 months since menarche at age 13. Her last period was about 6 months ago. She has mild facial hirsutism. There has been no recent change in weight. On examination she has mild central adiposity with coarse hairs on upper lip and abdomen, and a BMI of 28. There is no clinical evidence of Cushing’s syndrome or overt virilization. The clinical picture is consistent with the polycystic ovarian syndrome.’ Respondents were asked to select from a checklist the investigations they would order for this patient. They were then asked which criteria they considered essential to diagnose PCOS, again selecting their responses from a checklist. They were told that the results of their investigations were consistent with PCOS, and respondents were asked which treatment they would advise if the patient’s primary concern was, in turn, oligomenorrhoea, hirsutism, infertility or obesity, selecting options from a list in the order in which they would use them. Finally, a variation on the index case was presented, in which the patient was aged 40 with two previous children by ovulation induction. Respondents were asked whether investigations would differ from the index case, and, if so, what changes they would make.
Categorical data were examined for differences in response between endocrinologists and gynaecologists using Ï‡2-tests, Fisher’s exact test or the normal approximation to the binomial, as appropriate. Differences were considered to be significant if P < 0Â·050. To compare responses for the index case and the variation, Kappa values were calculated for each question. Excellent agreement was indicated with a Kappa value of > 0Â·75, fair agreement if 0Â·4-0Â·75, and poor agreement if < 0Â·4.  Results are presented as percentages after exclusion of missing or ambiguous responses. The study was approved by the Human Research Ethics Committee of Sir Charles Gairdner Hospital, the Endocrine Society of Australia and the Royal Australian and New Zealand College of Obstetricians and Gynaecologists. Results Response Rate and Demographics The overall response rate was 58%, comprising 176 endocrinologists (53% response rate) and 210 gynaecologists (63% response rate). After excluding responses from those who did not manage patients with PCOS, responses from 138 endocrinologists and 172 gynaecologists were analysed. Respondents were predominantly in private practice (endocrinologists 82%, gynaecologists 87%) and/or teaching hospitals (endocrinologists 60%, gynaecologists 33%). Investigation of the Index Case The investigations requested for the index case are shown in Table 1 . Endocrinologists were more likely to request 17-hydroxyprogesterone (17-OH progesterone) and dehydroepiandrosterone sulfate (DHEAS). More endocrinologists than gynaecologists requested urine free cortisol, but this accounted for a minority in each group. Ovarian ultrasound was requested by almost all gynaecologists (91%) but by less than half of the endocrinologists (44%) ( P < 0Â·001). Gynaecologists were also more likely to request endometrial ultrasound. Surgical investigations such as hysteroscopy and laparoscopy were requested by small numbers in both groups, although at a higher rate in the gynaecology group. There were significant differences between specialty groups when assessing glucose and lipid metabolism. Endocrinologists were more likely to undertake any form of glucose assessment. However, when gynaecologists did assess glucose metabolism they were more likely than endocrinologists to ask for an oral glucose tolerance test (OGTT), 45% vs . 23% ( P < 0Â·001). Endocrinologists were twice as likely as gynaecologists to request fasting lipid profiles, 67% vs. 34% ( P < 0Â·001). Diagnostic Criteria for PCOS Endocrinologists were most likely to select menstrual irregularity (70%) with either clinical or biochemical androgenization as essential for the diagnosis of PCOS (see Table 2 ). Less than half (47%) of the gynaecologists believed that menstrual irregularity was an essential diagnostic criterion ( P < 0Â·001). Gynaecologists were less likely to include androgenization as a diagnostic criterion, although when this was selected the majority also believed that either clinical or biochemical androgenization was adequate for diagnosis. Sixty-one per cent of the gynaecology group compared with 14% of the endocrinology group felt that polycystic ovaries on ultrasound were essential for the diagnosis of PCOS ( P < 0Â·001). The LH/FSH ratio was more likely to be included as essential for diagnosis by gynaecologists ( P < 0Â·001). In this survey, only 38% of endocrinologists and 9% of gynaecologists would have used the 1990 National Institutes of Health (NIH) criteria, indicating menstrual irregularity and androgenization as the essential criteria ( P < 0Â·001). Furthermore, only 41% of endocrinologists and 16% of gynaecologists were in line with the more recent Rotterdam consensus criteria, with polycystic ovaries on ultrasound added as one of the possible criteria ( P < 0Â·001)  (see Table 3 ). Treatment Choices Table 4 shows the top four choices for treatment for the varying presentations, showing the percentage of respondents who chose these modalities among their first three treatment options. When the primary concern of the index patient was oligomenorrhoea, diet and exercise were recommended as first-line treatment by the majority of endocrinologists and gynaecologists. The oral contraceptive pill (OCP) was most commonly selected as second-line therapy. Endocrinologists were more likely than gynaecologists to select metformin as a second- or third-line agent, whereas the reverse applied to the use of cyproterone acetate combined with OCP. For hirsutism, diet and exercise were again the most widely recommended first-line treatment in each group. The second most popular treatment choice for hirsutism in both groups was cyproterone acetate with OCP. Metformin was ranked in the top three treatment choices by 22% of endocrinologists and 10% of gynaecologists ( P = 0Â·004). Local treatment for hirsutism was recommended infrequently, with 15% of endocrinologists and 8% of gynaecologists ranking this option in their top three treatment choices ( P = 0Â·051). Diet and exercise were the most common first-line treatment selected by both groups for infertility. Metformin was the most frequent second-line agent selected by endocrinologists while gynaecologists preferred clomiphene. Rates of ovarian surgery were low, although more commonly selected by gynaecologists (11% vs . 2% of endocrinologists, P = 0Â·002). Of the endocrinologists, 87% compared with 62% of gynaecologists stated they would refer the case on to a fertility specialist ( P < 0Â·001). Thiazolidinediones were selected as a treatment choice by 9% of endocrinologists and 1% of gynaecologists, although this was only ranked in the top three treatment choices by 7% of endocrinologists and 1% of gynaecologists ( P = 0Â·05). When the primary concern was obesity, diet and exercise were recommended as first-line treatment by most respondents in both groups. The most common second-line therapy in both groups was metformin, although the use of this agent was significantly greater in the endocrinology group (74% vs. 45%, P < 0Â·001). Endocrinologists were also significantly more likely to use obesity drugs, most commonly selected as a third-line agent (48% vs . 9%, P < 0Â·001). A comparison of the investigations ordered for the index case and the variation is shown in Table 5 . The biochemical investigations selected for the variation case did not differ significantly from those presented in Table 1 for the index case. In particular, there was no significant change in assessment of cardiovascular risk factors such as lipids or glucose metabolism for the variation case compared with the index case. However, there was a significant increase in endometrial assessment by ultrasound by endocrinologists for the variation case (Kappa value 0Â·291, Kappa significance < 0Â·01). Discussion This study showed a lack of consensus between endocrinologists and gynaecologists in their investigation, diagnosis and treatment of PCOS. The response rate was relatively high, and similar to previous surveys of this type. Although the hypothetical nature of the case takes no account of patient preferences that would influence management decisions in clinical practice, the results are likely to be representative of the opinions of two expert groups on this topic. The approach to investigation of the index case was broadly similar in both groups; however, endocrinologists were more likely than gynaecologists to investigate adrenal aetiologies for hyperandrogenism. There were also differences in the investigation of glucose homeostasis. Women with PCOS have significantly increased rates of diabetes, impaired glucose tolerance and impaired fasting glycaemia, especially in association with risk factors such as obesity and increased age. The use of a glucose tolerance test, rather than fasting glucose, to assess glucose homeostasis is supported by data suggesting that fasting glucose may be a poor predictor of diabetes in women with PCOS. Dyslipidaemia is a significant, modifiable risk factor for future cardiovascular disease. In one study comparing women with PCOS and weight-matched controls, total serum cholesterol levels were similar in both groups but high density lipoprotein (HDL)2-cholesterol was reduced in both obese and lean women with PCOS. By contrast, other studies have found that elevated LDL is the predominant lipid abnormality in PCOS, independent of obesity.[14,15] Although endocrinologists were more likely to request a lipid profile on the index case, there was surprisingly no significant difference between the rate of lipid assessment between index and the older variation case in either respondent group, despite the increased risk of cardiovascular disease in older subjects. The disparity over use of ovarian ultrasonography is indicative of the controversy over the inclusion of the ultrasound findings in the diagnostic criteria for PCOS. In asymptomatic women, the appearance of polycystic ovaries on ultrasound has been shown to have no significant impact on fertility. In addition, ultrasonographic evidence of polycystic ovaries was found in 22% of 257 healthy volunteers in one study, none of whom had found it necessary to seek medical attention for gynaecological symptoms. However, in that study, irregular menses and hirsutism were found at higher rates in association with polycystic compared with normal ovaries. Furthermore, transvaginal ovarian ultrasound for the diagnosis of PCOS has significant intraobserver and interobserver variability and, as such, may be considered subjective. The predictive value of ultrasonographic evidence of polycystic ovaries on future reproductive and metabolic health remains unclear. For the diagnosis of PCOS, the only criterion considered essential by a majority of respondents from both consultant groups in our survey was androgenization. There was significant lack of consensus about other criteria. Menstrual irregularity was favoured by endocrinologists and ultrasound findings by gynaecologists. Some reliance on an elevated LH/FSH ratio as an essential diagnostic criterion persists; however, this should be abandoned. Although a high LH/FSH ratio was commonly seen in PCOS when older polyclonal radioimmunoassays were used, modern monoclonal LH assays have made this criterion unreliable. Table 3 outlines the previous and current consensus statements regarding the diagnosis of PCOS. It can be seen from the responses to our survey that there is a lack of consensus with even the latest consensus statement. Consequently, comparison of data and the application of research findings to patient groups is often impossible. With regard to treatment, diet and exercise were favoured by both endocrinologists and gynaecologists as the therapy of first choice for oligomenorrhoea, followed by the OCP. Lifestyle modification including diet and exercise has been shown in the short term to improve cycle frequency in obese women with PCOS. The long-term challenge has been to sustain the effort and the effect. The use of the OCP has beneficial effects in endometrial protection, reduction of androgens and regularization of menses. However, concerns have been raised about use of the OCP in obese women with PCOS in the light of evidence that it reduces insulin sensitivity.[21-23] For this reason metformin has been enthusiastically embraced as an alternative drug therapy for this indication. Metformin rated highly as treatment for oligomenorrhoea. A review of nine uncontrolled studies of metformin monotherapy has found that metformin regularized menses in approximately 62% of women with PCOS. Results of randomized studies also confirm a beneficial, although unquantified, effect of metformin on cycle regularity.[26-28] Very few respondents recommended cyclical progestogens as a treatment for this indication, despite evidence that this prevents endometrial hyperplasia in women with PCOS. Hirsutism in the setting of PCOS is often difficult to treat. There are no randomized controlled trials assessing the efficacy of diet and exercise as a treatment for hirsutism. Despite this, the most common treatment option selected in this survey for the treatment of hirsutism was diet and exercise, followed by the oral contraceptive pill plus spironolactone or cyproterone acetate. Endocrinologists were more likely to favour spironolactone while gynaecologists more commonly used cyproterone acetate. Metformin was chosen by a minority of respondents and the recommendation of local therapies was surprisingly minimal. There are data to support the use of spironolactone, cyproterone acetate and metformin for hirsutism in the setting of PCOS.[30-33] For the treatment of infertility, diet and exercise were selected by both groups as the first treatment choice. Although there are few randomized studies of dietary intervention and fertility, there are several studies using women as their own historical controls.[34-36] Lifestyle modification in the short term improves ovulatory frequency via the proposed mechanisms of reduction in circulating insulin and thereby a reduction in ovarian androgen production. Following lifestyle advice, clomiphene therapy was chosen by gynaecologists and metformin by endocrinologists. Clomiphene may achieve ovulation in up to 80% of anovulatory women with PCOS but the ongoing pregnancy rate is much lower, at 38%. Factors that predict a lower chance of success are higher androgen index, higher body mass index (BMI) and greater age. It has been suggested that the chances may be improved by correcting the two former factors with diet and exercise or with metformin. There is evidence for a significant effect on ovulation of metformin compared to placebo, when used alone or with clomiphene. An ongoing randomized study is comparing metformin with clomiphene as the primary method of ovulation induction. Twelve endocrinologists and two gynaecologists chose thiazolidinediones as their second, third or fourth choice of therapy. Although studies using thiazolidinediones have shown improvement in ovulation rates, at present their use for this indication is ahead of prescribing guidelines.[39-41] The current Food and Drug Administration (FDA) recommendation is that contraception should be used by patients taking thiazolidinediones and they are classed as Category C. Ovarian wedge resection has fallen into disfavour following evidence of ovarian scarring, but some form of ovarian surgery is still among the choices made by 46 gynaecologists and five endocrinologists. More recently, ovarian surgery has been 'refined' to include drilling by diathermy or laser via the laparoscopic approach; however, even with this approach, adhesion formation is common and ovarian injury and atrophy is possible. It has been suggested that limited ovarian surgery is a cost-effective alternative to gonadotrophin therapy. A review of studies comparing gonadotrophins with ovarian drilling has concluded that there is insufficient evidence, on efficacy alone, to prefer one treatment over the other. However, a recent randomised study demonstrated metformin to be superior to laparoscopic ovarian drilling. Obesity was managed first with diet and exercise by over 90% of all respondents. The use of obesity drugs was significantly higher in the endocrinology group (48% vs . 9%, P < 0Â·001). The use of metformin for obesity was also significantly higher among endocrinologists. Weight reduction in obese PCOS subjects through a low calorie diet alone has been shown to improve anthropometric indices, ovarian morphology and spontaneous pregnancy rates. Metformin in the setting of a hypocaloric diet has been shown to result in a greater reduction of body weight and abdominal fat in PCOS subjects compared with placebo. There are no randomized controlled trials regarding the use of antiobesity drugs such as sibutramine in the setting of PCOS. However, a recent study evaluating the effectiveness of sibutramine therapy alone and in combination with ethinyl oestradiolcyproterone acetate in obese women with PCOS found that both treatments exhibited positive effects on clinical and metabolic risk factors for cardiovascular disease. The variation case was intended to explore the assessment of possible complications of PCOS in older patients. In fact, there was no significant difference in the assessment of glucose metabolism or lipids for the older hypothetical patient compared to the index case. The true incidence of cardiovascular disease in women with PCOS compared with the general population is unclear, partly because of the variation in diagnostic criteria between studies. One large study of women classified as having PCOS by ovarian morphology alone found increased rates of diabetes and cerebrovascular disease but not cardiovascular disease. When corrected for BMI there was no difference in rates of diabetes or hypertension between the groups, suggesting that the increased incidence of cardiovascular risk factors in women with polycystic ovaries was in part obesity related. By contrast, more recent studies have demonstrated increased rates of coronary artery calcification and vascular stiffness in women with PCOS.[49,50] Although cardiovascular risk factors are more prevalent in the PCOS population, definitive evidence of increased cardiovascular disease, independent of obesity, is lacking. It therefore remains unclear whether all women with PCOS should be specifically targeted in cardiovascular prevention programmes. The increased use of endometrial ultrasound may reflect the observation that long-standing PCOS is associated with a risk of endometrial cancer, although there is a lack of well-designed studies to quantify the relative risk. In conclusion, there are significant differences between endocrinologists and gynaecologists in their investigation, diagnosis and treatment of PCOS. This immediately raises the prospect that an individual woman's diagnosis may depend on which specialist she sees and which investigations are undertaken. It also makes the comparison of research studies on PCOS difficult when different diagnostic criteria are used for study entry. Despite attempts to reach consensus on the diagnosis of PCOS, it is clear that only a minority of endocrinologists and gynaecologists adhere to NIH or Rotterdam diagnostic criteria. The possibility remains that, by their practice, endocrinologists and gynaecologists are identifying different subgroups of a heterogeneous population. Further research into the possible subgroups within the population of women classified as having PCOS and the potential differences in long-term health consequences is needed. http://www.medscape.com/viewarticle/500870_print