Listening to patients with unexplained menstrual symptoms: what do they tell the gynaecologist?
*Correspondence: Dr A. Garden, Department of Obstetrics and Gynaecology, Liverpool Women's Hospital, Crown Street, Liverpool L8 7SS, UK.
Objectives To describe how women present unexplained menstrual symptoms to gynaecologists; to find out whether presentation reflects how intense their symptoms feel or how much benefit they expect from gynaecological treatment; and to test the prediction that surgical treatment decisions are more likely to follow specific types of presentation.
Design A cross sectional cohort study.
Setting Gynaecological outpatient clinics in a teaching hospital.
Sample Fifty-nine patients, referred for menstrual problems, in whom investigations had excluded physical disease.
Methods Patients indicated symptom intensity and expectations of treatment before consultation on self-completed questionnaires. Audiotape recordings of consultations with the gynaecologist were transcribed and patients' use of specific communication strategies was coded according to a previously reported scheme.
Main outcome measures Treatment decision was noted.
Results In a third to a half of patients, presentation extended beyond symptom report to include catastrophisation about consequences of symptoms, reference to other individuals to substantiate the patient's problems, criticism of previous or possible future interventions or request for hysterectomy. Those with greatest expectations of gynaecological treatment were more likely to catastrophise and request hysterectomy. Those with most intense subjective symptoms were more likely to catastrophise, refer to other individuals and request hysterectomy, and these strategies were more likely to be followed by surgical treatment decisions.
Conclusions Surgical treatment for unexplained menstrual problems is not driven by gynaecologists. Reduction in unnecessary hysterectomies will require training in communication skills that reflects the challenging nature of many patients' presentation in the gynaecology clinic.
The benefits of teaching communication skills to doctors are now widely accepted, particularly in respect of patient interviewing, explaining diagnoses and breaking bad news1. However, one of the most difficult communication problems for doctors, the negotiation of treatment decisions for physical symptoms in the absence of physical pathology, is not yet routinely the subject of communication teaching2. Before effective teaching can be developed in this area, it is necessary to have evidence about how patients describe their problems. The aim of the present study was to gather this evidence for patients with menorrhagia.
Excessive bleeding during menstruation is a common gynaecological problem accounting for up to one-third of gynaecology outpatient referrals3. Although clinical research has defined abnormal menstrual loss as over 80 mL per cycle, demonstrable loss is within recognised normal limits in most patients complaining of menorrhagia4. Moreover, objective measurement of blood loss is not applicable to everyday clinical practice and, although more ‘patient-friendly’ techniques of assessing menstrual loss have been reported5, the diagnosis and management of menorrhagia usually depends entirely on what the woman says about her blood loss. Despite a wide variety of medical techniques shown to be effective in reducing menstrual loss6, more than 50% of women undergo hysterectomy within five years of referral7. Indeed, 70% of hysterectomies carried out on premenopausal women in the UK are for abnormal menstrual loss. Therefore, it is estimated that, by the age of 55 years, one in five women in the UK will have received a hysterectomy8, many in the absence of demonstrable pathology. Techniques such as endometrial ablation have not reduced the demand for hysterectomy9, but have increased the proportion of patients receiving surgical intervention. Information on how patients with menorrhagia present their problems to gynaecologists, and on how decisions for surgery emerge in such consultations, might help to understand apparently high surgical intervention rates, and would provide an evidence base for training gynaecologists in the communication skills needed to ensure that treatment decisions meet patients' needs and avoid unnecessary intervention.
The primary aim of the present study was therefore to use techniques developed previously for analysing audiotape recordings of gynaecological consultations10 to describe how women without evidence of physical pathology communicate their menstrual problems to the gynaecologist. A secondary aim was to test two views as to factors that influence how women present their symptoms. One is that the presentation simply reflects the intensity of symptoms (i.e. those with the worst menorrhagia present most forcefully). The second view arises from evidence that patients have well defined expectations of the benefits of gynaecological treatments—particularly surgery—that go far beyond the relief of menstrual symptoms to include improved psychological and physical wellbeing11. Presentation might therefore be shaped by these expectations; that is, patients might behave as ‘consumers’ such that how they present reflects the degree to which they expect that the gynaecologist can help them. Finally, we examined whether, as has been suggested previously, gynaecologists' surgical treatment decisions were related to—and therefore perhaps influenced by—the ways in which women presented their symptoms10,12.
Consecutive patients, referred to one of three general gynaecologists at a teaching hospital with heavy or painful menstruation who had been investigated and in whom no abnormality had been identified, were approached. Patients who were under 18 years old, had insufficient understanding of English to give consent or complete questionnaires or previous history of major psychiatric illness were excluded.
After ethical approval, suitable patients were identified from case records, were asked for written consent and were assured of confidentiality and anonymity before consultation. Consenting patients then completed a questionnaire providing socio-demographic information and indicating severity of menstrual symptoms on five-point Likert scales: pain (from ‘not painful’ to ‘very painful’); and blood loss (from ‘very light’ to ‘very heavy’). To find out whether presentation was related to patients' expectations, patients completed the Expectations of Gynaecological Treatments Questionnaire10, which provided separate scores to indicate the extent to which the gynaecologist was expected to be able to reduce specifically menstrual symptoms, improve general physical wellbeing and improve psychological wellbeing.
The consultation between the gynaecologist and patient was recorded on audiotape and transcribed anonymously. The gynaecologist completed a simple rating to describe the extent to which the outcome of the consultation was influenced mainly by the gynaecologist or the patient12. Using a procedure developed previously12, the transcript of each consultation was coded for the presence or absence in patients' speech of each of 10 communication strategies (see Table 1), chosen on the basis of their potential to influence the gynaecologist's decision. Strategies are illustrated when reporting results below. This coding procedure has previously been shown to have good interrater reliability12, and its content validity is derived from the qualitative research upon which it was based10. Treatment decision, noted from the transcripts and confirmed from medical records, was used to divide patients into two groups: surgery (hysterectomy or endometrial ablation) and conservative management or no treatment.
Table 1. Number of patients [n (%)] who used each communication strategy. Numbers are shown for the total sample and for those who received conservative or surgical treatment decisions. Odds ratios [95% confidence intervals] and χ2 describe the effect of each strategy on the odds of a surgical decision (Fisher's exact probabilities were used where an expected cell size <5).
|Symptom report|| || || || || |
|Nature of symptoms||51 (93)||40 (93)||11 (92)||0.83 [0.08–18.73]||0.26|
|Improvement||28 (51)||25 (58)||3 (25)||0.24 [0.06–1.00]||4.12*|
|Symptoms unchanged||20 (36)||13 (30)||7 (58)||3.23 [0.86–12.09]||3.20|
|Deterioration||16 (29)||12 (28)||4 (33)||1.29 [0.33–5.10]||0.13|
|Catastrophisation||20 (36)||11 (26)||9 (75)||8.73 [2.00–38.16]||9.90*|
|Criticism of interventions|| || || || || |
|Received||32 (58)||24 (56)||8 (67)||1.58 [0.41–6.06]||0.45|
|Anticipated||21 (38)||15 (35)||6 (50)||1.87 [0.51–6.81]||0.91|
|Request for intervention|| || || || || |
|Nonsurgical||17 (31)||15 (35)||2 (17)||0.37 [0.07–1.93]||1.46|
|Hysterectomy||18 (33)||6 (14)||12 (100)||70.572 [7.92–629.24]||31.55**|
|Invoke other's authority||24 (44)||15 (35)||9 (75)||5.60 [1.31–23.86]||6.14*|
Relationships between categorical variables (decision for surgery vs conservative treatment; whether or not each strategy was used) were examined by χ2 (Fisher's exact test being used to estimate significance as appropriate where expected cell sizes were small). Expectations of the effects of treatment were approximately normally distributed and were examined by t tests. Symptom intensity (pain, bleeding) was measured using the Mann–Whitney U test. Logistic regression analysis was used where it was necessary to identify which of a set of variables was uniquely related to use of a specific strategy or to treatment decision. Analyses were by SPSS10. The criterion for significance was P < 0.05.
Two women were excluded because of inadequate English; four declined to participate or withdrew participation. The final sample contained 59 patients but four of these consultations were not recorded because of equipment failure. Mean age was 37 years (range 18–50; SD 7); 45 were married or living with a partner; 46 had one or more children.
Referrals were for menstrual bleeding (15 women), pain (8 women) or both (36 women). Patients had received a mean of 2.4 previous consultations (range 1–8) and 2.3 investigations (range 1–6) as part of the current referral. Nine women had been previously referred to the study hospital with menstrual symptoms. Investigations were mainly hysteroscopy (25 women), dilatation and curettage (25 women), hormone profile (15 women) and laparoscopy (13 women). Others included ultrasound scan, outpatient endometrial sampling, thyroid function tests and screening for infection.
Of the 59 consultations, 23 (39%) were with senior house officers, 34 (58%) with specialist registrars and 1 (2%) with a consultant. Treatment decisions included: surgery (nine hysterectomy and three endometrial ablation); further investigations or change in conservative management (27); and no change in conservative management or discharge (20). As anticipated, gynaecologists were more likely to view consultations leading to surgery as influenced by the patient (Table 2; Mantel–Haenszel χ2= 4.24, df= 1, P < 0.05).
Table 2. Gynaecologists' ratings of ‘who influenced the treatment decision most’. The number of consultations is shown separately for those leading to conservative and surgical decisions. The scale was omitted for three patients.
|Patient and doctor equally||28||6|
Patients' use of different strategies is detailed in Table 1. The most common was, understandably, to describe the symptoms (e.g. ‘my periods are heavy’), although this was absent in four consultations. More than half of the patients criticised previous interventions (e.g. ‘since the investigations it's a million times worse’). Between a third and a half criticised anticipated interventions (e.g. ‘it wouldn't help me’), described symptoms catastrophically (e.g. ‘I was even going to tell my husband to phone an ambulance’), requested hysterectomy (e.g. ‘as far as I'm concerned, I came here today to opt for a hysterectomy’); all patients who requested endometrial ablation also requested hysterectomy, or referred to another individual as authority for their symptoms (e.g. ‘I showed my boyfriend because it was really, really frightening’) or for their need for treatment (e.g. ‘I've talked to people who've had hysterectomies and they've said they've felt a lot better’).
Details of the responses are shown in Table 3. Predictably, patients who had rated their menstrual pain most intensely were least likely to report symptom improvement (U= 142, P < 0.001) and were more likely to report that symptoms were unchanged (U= 212, P < 0.05). They were also more likely to catastrophise (U= 143, P < 0.001), criticise interventions received (U= 241, P < 0.05), request hysterectomy (U= 106, P < 0.001) and cite external authority (U= 187, P < 0.01). Patients reporting most blood loss were less likely to report symptom improvement (U= 245, P < 0.05) and were more likely to describe deterioration (U= 192, P < 0.05) and to refer to another individual as authority for their symptoms or need for treatment (U= 224, P < 0.05).
Table 3. The intensity of specific symptoms and expectations in patients who used vs patients who did not use specific communication strategies with the gynaecologist. Expectation of improved menstrual symptoms was unrelated to any strategy and is not shown. Scores are arbitrary units. For each strategy, significance levels refer to comparisons of patients who used the strategy vs those who did not. Values are given as mean (and, for skewed variables, median) [SEM].
|Symptoms improved|| || || || |
|Not used||4.44 (5.0) [0.15]||4.26 (4.0) [0.16]||31.41 [1.95]||14.52 [1.26]|
|Used||3.00*** (3.0) [0.26]||3.56* (4.0) [0.24]||25.18* [2.08]||10.39* [1.13]|
|Symptoms unchanged|| || || || |
|Not used||3.41 (3.5) [0.23]||3.59 (3.5) [0.20]||26.17 [1.74]||11.46 [1.13]|
|Used||4.25* (5.0) [0.25]||4.45 (4.5) [0.14]||31.85 [2.54]||14.10 [1.37]|
|Deterioration|| || || || |
|Not used||3.53 (4.0) [0.22]||3.71 (4.0) [0.18]||29.08 [1.65]||12.26 [0.92]|
|Used||4.19 (4.5) [0.26]||4.38* (5.0) [0.24]||26.19 [3.12]||12.81 [2.11]|
|Catastrophisation|| || || || |
|Not used||3.29 (3.0) [0.22]||3.86 (4.0) [0.18]||27.06 [1.68]||10.29 [0.89]|
|Used||4.53*** (5.0) [0.21]||4.00 (4.0) [0.27]||30.30 [2.80]||16.15*** [1.57]|
|Criticism of treatment received|| || || |
|Not used||3.32 (3.0) [0.27]||3.91 (4.0) [0.23]||28.83 [2.13]||12.17 [1.32]|
|Used||4.00* (4.5) [0.23]||3.91 (4.0) [0.20]||27.81 [2.05]||12.59 [1.20]|
|Request for hysterectomy|| || || || |
|Not used||3.27 (3.0) [0.21]||3.78 (4.0) [0.17]||25.35 [1.59]||10.46 [0.89]|
|Used||4.71*** (5.0) [0.17]||4.18 (5.0) [0.30]||34.17** [2.67]||16.44*** [1.64]|
|Invoke individual as authority|| || || |
|Not used||3.26 (3.0) [0.24]||3.68 (4.0) [0.21]||27.87 [1.67]||11.39 [1.07]|
|Used||4.35** (5.0) [0.21]||4.22* (4.0) [0.18]||28.71 [2.65]||13.75 [1.46]|
Patients with high expectations that the gynaecologist could relieve specifically menstrual symptoms were no more likely than others to use any strategy. By contrast, patients who expected the greatest general improvement in physical wellbeing were more likely to catastrophise (t= 3.52, P < 0.001) and request hysterectomy (t= 3.50, P < 0.001) and were less likely to report symptom improvement (t= 2.44, P < 0.05). Similarly, those who expected the greatest improvement in psychological wellbeing were also more likely to request hysterectomy (t= 3.00, P < 0.01) and were less likely to report symptom improvement (t= 2.18, P < 0.05).
It was possible that patients with high expectations of the gynaecologist were more likely to use these communication strategies merely because they had the worst symptoms. Therefore, in logistic regression analyses, the use of each strategy was regressed on expectations after controlling for symptoms by stepwise entry of pain and bleeding. Even after controlling for symptoms in this way, expectation of improved physical wellbeing was associated with catastrophisation (χ2= 4.71, df= 1, P < 0.05; odds ratio and 95% confidence interval associated with unit increase in score: 1.13, 1.01, 1.28) and expectation of improved psychological wellbeing was associated with requesting hysterectomy (χ2= 5.60, df= 1, P < 0.05; odds ratio and 95% confidence interval associated with unit increase in score: 1.09, 1.01, 1.18).
Three strategies were more common in consultations that led to surgery (Table 1): catastrophisation, requesting hysterectomy and citing another individual as authority. Conversely, conservative management was more likely where patients reported symptom improvement. In logistic regression analysis, treatment decision was regressed on all these strategies. Only the patient's request for hysterectomy uniquely predicted the gynaecologist's decision to offer surgery (χ2= 34.79, df= 1, P < 0.001). Because all surgically treated patients had requested hysterectomy, the odds ratio was infinite. To enable calculation of a finite, albeit conservative, odds ratio and confidence interval, we therefore added one to each cell of the contingency table. The resulting odds ratio and 95% confidence interval is shown in Table 1.
In all patients that we studied, physical pathology had been excluded. Therefore, the only information that the gynaecologists had about the patients' problems was what the patients said to them. Many patients presented their problems in ways that extended beyond mere description of symptoms and reports of deterioration or improvement. More than half criticised previous interventions, whether intended to be therapeutic or investigative. Nearly half referred to another person in such a way as to attest to the intensity of symptoms or the need for treatment. More than a third explicitly criticised interventions that were possible in the future or described symptoms in ‘catastrophic’ ways that conveyed the possibility or fear of serious consequences if symptoms remained untreated. Finally, a third of patients explicitly requested hysterectomy.
Although the content of these strategies did not convey information about the nature of the symptoms, three of them were, in effect, indicators of subjective symptom intensity. Patients who had reported most pain or blood loss on symptom rating scales were more likely to catastrophise, request hysterectomy or invoke another person's authority. However, patients also behaved as ‘consumers’: even allowing for the observation that patients with worse symptoms were more likely to use these strategies, patients with highest expectations of what gynaecologists could do to improve their lives were most likely to catastrophise and request hysterectomy.
Some of the ways in which patients presented their problems are likely to be challenging for the gynaecologist. For instance, in requesting hysterectomy, the patient takes on the role of recommending a treatment, which is usually regarded as the doctor's responsibility. Blaming the gynaecologist for failure or harm associated with previous interventions implies the gynaecologist's culpability for the patient's suffering. Catastrophisation and reference to other people to attest to the patient's need for treatment might be perceived as coercive13. There was evidence that these challenging presentations might have influenced the gynaecologists' decisions. Surgery was more likely when patients had catastrophised, invoked another person as authority for their symptoms or need for treatment or had requested hysterectomy. Requesting hysterectomy was the key strategy; others predicted surgical decisions only because the patients who used them also requested hysterectomy.
An obvious hypothesis is therefore that it is by using influential strategies in communication with the gynaecologist that, in the absence of physical disease, patients with greatest symptoms or with greatest expectations achieve surgery. However, our findings are based on a small sample size. Further research, with a larger number of patients, would be necessary to test this hypothesis and to explore whether the patients' strategies have different effects dependent on gynaecologists' characteristics such as experience and gender. It could be anticipated that junior doctors with less experience or confidence in their clinical abilities would be more likely to be influenced by patients using such strategies than more senior doctors.
Our measurement of patients' ways of presenting to gynaecologists is a significant advance on previous ways of coding speech in consultations14 because it focuses on aspects of speech that are likely to influence the outcome of a consultation. Nevertheless, it is limited. Audiotaping excludes information about non-verbal communication. Moreover, patients probably use strategies that we did not measure, and the number of times that specific strategies are used may be more important than simply whether or not they are used. However, our measurements were sufficiently sensitive to show unambiguously that many patients presented in ways that are not addressed in current teaching on communication. Together with a previous report12, they therefore begin to build an evidence base for communication teaching that reflects the reality of gynaecological consultation.
Our findings are inconsistent with the widely held view that casts gynaecologists in the role of exploiting women by administering unnecessary surgery. All patients who were treated surgically had requested hysterectomy. Reduction in the surgery that occurs in the absence of physical disease, and which may therefore be unnecessary, will require communication training for gynaecologists that reflects the challenging nature of many patients' presentation, patient education about the benefits of alternative treatments and a debate about the merits of a consumer-oriented approach to major treatment decisions. Our findings can inform such teaching and debate.
The authors would like to thank the gynaecologists, nurses and patients at the Liverpool Women's Hospital for their enthusiastic participation, Jill Floyd and Penny Jones for their expert organisational assistance and Susupta Chaudhuri for assistance with data collection.