- Top of page
When a woman presents a symptom of vaginal bleeding to a primary care clinician there is a need to relieve the distress the symptom is causing and to determine the likelihood of serious underlying pathology. In premenopausal women the symptoms of endometrial cancer may be heavy menstrual bleeding, intermenstrual bleeding and irregular bleeding. Cervical cancer may also present with irregular bleeding, but the cardinal symptom is postcoital bleeding.[2-4]
Uterine cancer is rare in primary care populations, but is a major cause for concern for the clinician because of the devastating effect that a delay in diagnosis can have on the individual patient. It is predominantly composed of malignancy from two sites: the endometrium of the body of the uterus and the cervix. Endometrial cancer is uncommon in menstruating women, with 852 registrations in the age group of 40–54 years in England in 2010. It is a rare cause of death in this age group, with 35 deaths in England and Wales in 2010. Carcinoma of the cervix is less common in the same age group (617 registrations England 2010), but is a more common cause of death at this time of life (191 deaths in England and Wales 2010). A National Screening programme exists in the UK for its prevention and early detection.
In contrast to the rarity of uterine malignancy in perimenopausal primary care populations, the symptoms of heavy menstrual, intermenstrual, irregular and postcoital bleeding are common.[7-11] The disparity between the rarity of cancer and the frequency of symptoms has caused difficulties in making recommendations as to which women to investigate. Some groups take the view that all women aged over 35 years with abnormal uterine bleeding should undergo investigation,[12-14] whereas others modify the advice and recommend investigation in those with risk factors. The National Institute of Health and Clinical Excellence (NICE) guidance on referral for suspected cancer recognises that the first symptom of gynaecological malignancy may be an alteration in the menstrual cycle, intermenstrual bleeding and postcoital bleeding, but only recommends a consideration of urgent referral for a woman with persistent intermenstrual bleeding and a negative pelvic examination. In contrast, the NICE heavy menstrual bleeding guideline states ‘if appropriate’ a biopsy should be taken for persistent intermenstrual bleeding or failed treatment (of heavy menstrual bleeding) in women ≥45 years. In their guidelines for the management of unscheduled bleeding in women on hormonal contraception, the UK Royal College of Obstetricians and Gynaecologists suggest that in general an endometrial biopsy should be considered in women aged ≥45 years (or in women under 45 years with risk factors) who have persistent unscheduled bleeding after the first 3 months or who present with a change in bleeding pattern. It is recognised that current UK guidelines on the referral of patients with symptoms of potential cancer to secondary care lack a sound epidemiological base.
In order for a woman to make an informed decision on the need for treatment of the symptom, and/or need for investigation for underlying malignancy, the chance that the symptom may resolve spontaneously or that it may be associated with uterine cancer needs to be estimated. We have conducted a linked longitudinal survey–medical record study using survey data from a previously published study on the spontaneous resolution of heavy menstrual bleeding, with the addition of medical record review to determine outcome with regard to malignancy. The aim of this study is to aid decision-making about the management of intermenstrual and postcoital bleeding in perimenopausal women by obtaining estimates of the likelihood that a woman will have spontaneous resolution of the symptom in the next year, and of the chance that she has an underlying malignancy.
- Top of page
The methodology and an analysis of data from the study concerning heavy menstrual bleeding have been published previously. Briefly, the sampling frame was provided by the registered populations of seven general practices that form part of a local research network in the UK. Practices are audited for the quality of their data recording. The practices were selected to cover a mix of urban and rural areas, and a range of affluence. The total registered population was 67 100, and the target population for the study was all women on the registers aged 40–54 years. As more than 99% of the population in England are registered with a general practice, the registers represent the local population. Women are not able to access specialist services (the National Health Service or private) in the UK without going through a primary care provider, except in an emergency. All clinical information was routinely recorded by the practices onto the Egton Medical Information System computerised database with an associated ‘read’ code (a hierarchical coding system of morbidity and practice activity used in the UK, comparable with the International Classification of Diseases).
Briefly summarising methods from our previous paper, the design was in two phases: first, a baseline cross-sectional postal survey of the target population; and second, a prospective cohort study of naturally menstruating women identified in the baseline survey, and who consented to follow-up questionnaires at 6, 12, 18 and 24 months. The baseline questionnaire was sent out on 5 September 2007.
Table 1 gives the definitions of terms related to vaginal bleeding used in the study derived from the questionnaire. Naturally menstruating women were defined as those who replied yes to the question ‘have you had a period in the last 6 months’ and no to questions concerning use of female hormones, effective treatments for irregular periods, intermenstrual bleeding and postcoital bleeding (gonadotrophin-releasing hormone analogues and antibiotics), an intrauterine device, pregnancy or gynaecological operation in the last 6 months, or ever having had an endometrial ablation.
Table 1. Questions used in the survey and terms used in the text
|‘Have you had a period in the last 6 months?’||‘Yes’||Menstruating|
|In the last 6 months, has not used female hormones, had treatment of known efficacy for irregular periods, intermenstrual and postcoital bleeding, used an intrauterine contraceptive device, been pregnant, had a gynaecological operation or ever had an endometrial ablation or hysterectomy.|| ||Natural|
|‘Over the last 6 months have you bled between periods?’||‘Yes’||Intermenstrual bleeding|
| ||‘Yes, three or more times’||Frequent intermenstrual bleeding|
| ||‘Yes’ on two consecutive questionnaires||Persistent intermenstrual bleeding|
|‘Over the last 6 months have you bled during or after sexual intercourse (making love) when you were not on a period’||‘Yes’||Postcoital bleeding|
| ||‘Yes, three or more times’||Frequent postcoital bleeding|
| ||‘Yes’ on two consecutive questionnaires||Persistent postcoital bleeding|
| ||‘Yes’ or ‘No’||Sexually active|
|For menstruating women, ‘Have you consulted a doctor or nurse during the last 6 months about irregular periods, bleeding between periods or bleeding after sexual intercourse (making love) when you were not on a period?’||‘Yes’||Consulted|
|For non-menstruating women, ‘Have you consulted a doctor or nurse during the last 6 months about bleeding during or after sexual intercourse (making love) or other vaginal bleeding (excluding periods)?’||‘Yes’|| |
The instruments used in the questionnaires to determine the presence or absence of intermenstrual and postcoital bleeding were derived from guidelines,[12-18] and have been validated and used in other studies by members of the study team. If a woman reported either of the two symptoms she was requested to record whether it had occurred once, twice, or three or more times in the previous 6 months. In the case of postcoital bleeding the woman was also offered the option of recording ‘not applicable’, and hence she was deemed not sexually active. If a woman reported the occurrence of intermenstrual or postcoital bleeding three or more times she was defined as both having the symptom and having the symptom frequently, thus frequent intermenstrual and frequent postcoital bleeding are subgroups of intermenstrual and postcoital bleeding, respectively.
The questionnaires contained a number of other items, including other vaginal bleeding symptoms and use of treatments for intermenstrual and postcoital bleeding, and consultation with primary and secondary care, in the previous 6 months.
The cohort of women who were identified from the baseline questionnaires as naturally menstruating provided the sample for analysis. Three outcomes were the focus of attention in the analyses of these women: prevalence and incidence of intermenstrual or postcoital bleeding, resolution of intermenstrual or postcoital bleeding, and use of health care (consultation with a doctor and/or nurse about irregular periods, intermenstrual or postcoital bleeding and, in women who subsequently reported no periods, other vaginal bleeding). Age and cohorts for analysis were defined in the same way as in the previous study at baseline and at 12 months (‘age-defining questionnaire’). If a woman was not of ‘natural’ status (Table 1) for the whole of the analysis time frame or if data were missing for the time frame, she was withdrawn from that cohort, with the exception of the analysis of consultation rates and medical records in which the woman only needed to be of natural status in the first questionnaire. Any one woman could contribute to more than one age-specific calculation if she moved up to the next 5–year age band in the subsequent age-defining 12–month questionnaire. A woman could only contribute once to the the ‘overall’ 15–year age band, and only once to a 2–year analysis, as a woman could not move up an age band.
Prevalence was calculated using baseline questionnaires, as in the previous study, with those with the symptom expressed as a proportion of all those in the age band. The exceptions to this were persistent intermenstrual and persistent postcoital bleeding, which were calculated from responses to the 6–month and 12–month questionnaires, and were expressed as the proportion of women who reported the symptom out of all of the women in the age band, with age defined at the 12–month questionnaire.
Incidence and resolution without recurrence
Incidence and resolution without recurrence were calculated using change in status between the age-defining questionnaire (baseline and 12 months) and subsequent questionnaires, as in the previous study. The 6–month cumulative incidence was defined as women reporting not having the symptom at the age-defining questionnaire but reporting having the symptom at the following questionnaire. The 1–year cumulative incidence was defined as reporting a new symptom at either of the two questionnaires following the age-defining questionnaire, and the 2–year cumulative incidence was defined as reporting a new symptom from baseline on any of the 6-, 12-, 18- or 24–month questionnaires. Thus for an incident case the woman had to develop the symptom at some time in the time frame.
Resolution without recurrence for 6 months was defined as reporting the symptom at the age-defining questionnaire (baseline and at 12 months) but not at the next questionnaire. Resolution without recurrence for 12 months was defined as reporting the symptom at the age-defining questionnaire but not at the next two questionnaires. For resolution without recurrence for 24 months, women had to report the symptom at baseline but not at 6, 12, 18 or 24 months. Thus for resolution without recurrence the woman had to be symptom free for the whole of the time frame following the age-defining questionnaire.
Similar analysis was undertaken for the annual cumulative incidence of recalling a consultation with a doctor and/or nurse during the last 6 months.
The use of the given time frames (6, 12 and 24 months) is supported by the analysis of optimal sampling strategies concerning menstrual function. The definitions of cumulative incidence and resolution were felt to represent the information requirements of women and doctors during a consultation, and provide a variety of rates that may be applied to individuals to meet their own personal values and needs.
Medical record review
In this study medical record review was used exclusively to identify whether a woman was diagnosed with uterine cancer or not. Women who consented to medical record review had their records reviewed for the occurrence of uterine cancer [Read codes B40.. or B41.. or B43.. or ((BB5j2 or BB5jE) without B440.)] from the date of sending out the baseline questionnaire to 2 years after the 24–month questionnaire. A further analysis was undertaken in those who reported intermenstrual and postcoital bleeding, or, for those who subsequently reported no periods, ‘other vaginal bleeding’. Women were classified on the basis of their first reported symptom in the longitudinal study, irrespective of whether they developed other symptoms subsequently. For each symptom the ‘index first symptom’ was the first time the symptom was reported by the woman during the study period. If a woman reported two symptoms simultaneously on the first symptomatic questionnaire she was classified as an index case for both symptoms. If other vaginal bleeding (no period) was reported at the same symptomatic questionnaire as intermenstrual and/or postcoital bleeding, the former symptom was not analysed. In addition, a separate analysis for persistent intermenstrual and persistent postcoital bleeding was undertaken. For persistent intermenstrual or postcoital bleeding the index first symptom was the first time it was recorded on two consecutive questionnaires.
Analyses were performed using spss 18 and confidence intervals for all relevant rates were calculated using confidence interval analysis. Chi-squared tests were used to test differences between proportions (including trend by age), and we report the P values (where a value less than 0.05 is assumed to indicate statistical significance).
- Top of page
We have found a high prevalence, incidence and spontaneous rate of resolution of intermenstrual and postcoital bleeding in naturally menstruating women during the perimenopausal years. Approximately a fifth of naturally menstruating women aged 40–54 years report intermenstrual and/or postcoital bleeding in the previous 6 months. Postcoital bleeding is much less common than intermenstrual bleeding.
Among the 82% of naturally menstruating women aged 40–54 years who have not had intermenstrual bleeding in the past 6 months, about a quarter will develop it at some time in the next 2 years. Among the 90% of women aged 40–54 years who have not had postcoital bleeding in the past 6 months, 8% will develop it at some time in the next 2 years.
Rates of resolution of intermenstrual and postcoital bleeding are also high at these ages. Overall, nearly 60% of women with intermenstrual and/or postcoital bleeding will experience its spontaneous resolution for the next 6 months. In a half of women with postcoital bleeding the resolution will be maintained for 2 years, but in those with resolved intermenstrual bleeding only about a third will be maintained without recurrence for 2 years. There is considerable variation with age in the rates of resolution and maintenance of the symptom-free state, with those in the younger age groups having lower rates in general.
The variation in the prevalence of single episodes of intermenstrual bleeding with age (with highest rates for women aged 45–49 years), in contrast to no significant variation with age for persistent and frequent intermenstrual bleeding, and all forms of postcoital bleeding, is consistent with the idea that in general the former is a natural physiological phenomenon. There is, however, a trend for the other symptoms to follow the same variation with age, suggesting that a physiological process may also be acting in these symptoms in some women, although this pattern in our study failed to achieve statistical significance.
The strength and weakness of this community-based longitudinal survey have been discussed previously. The use of practice registers ensured representation of the female population aged 40–54 years as a whole. The UK locality in which the study took place is slightly more deprived than the national average, and the representation of ethnic minority groups is low. The response rates were consistent with previously published studies of this type. The study was based on self-reporting, which may be subject to recall and observer bias, but is comparable with clinical practice in the UK in primary care. The time frame used of 6 months creates an imprecision in the rates of resolution, as the duration of resolution for an individual may be up to 6 months longer than reported here, depending on when the last event occurred in the previous 6 months. The questionnaire as a whole had not been validated in previous studies, the performance characteristics of the individual questions are unpublished and the questionnaire requires a degree of literacy to complete. The effect of this potential source of bias is unknown.
An earlier study on menstruating women gave a prevalence of intermenstrual bleeding of 13% and postcoital bleeding of 2.1% in women aged 45–54 years. The rates in our study are higher (Table 2), but unlike the earlier study excluded women using female hormones and treatments for intermenstrual and postcoital bleeding and endometrial ablation. Other studies using subjective assessment have given a prevalence of intermenstrual bleeding of between 2% of all women and 17% of menstruating women.[23, 8] One community study reported a prevalence of postcoital bleeding in a family planning clinic of 5%. The 12–month cumulative incidences of 12.0 and 3.4% for intermenstrual and postcoital bleeding, respectively, in the age group 45–54 years in the earlier study are comparable with those in our study (Table 3).
The statutory medical registration body in the UK requires that specialist doctors provide clinical information to the patient's general medical practitioner concerning all episodes of care. It is possible that some women with cancer were not recorded in the general practice records, but we estimate the chances of this to be small because of the healthcare system and the audited quality of routine data collection by the research practices. The expected number of cases of uterine cancer for the study population derived from National Statistics was 2, which is comparable with the three cases that developed.
An estimate of the positive predictive value of the so called ‘red flags’ for uterine cancer of intermenstrual, frequent intermenstrual, persistent intermenstrual, postcoital, frequent postcoital and persistent postcoital bleeding indicates that for community populations they are poorly predictive for uterine cancer. Of the 785 women with at least one of these symptoms in our study, only one subsequently developed uterine malignancy during at least 2 years of follow-up. Two other women who did develop malignancy had not reported red flag symptoms. This would suggest that advice to seek health care if a woman in the perimenopausal years develops such a symptom is inappropriate. A systematic review supports this view, reporting the probability of a woman in a community population with postcoital bleeding as having cervical cancer as 1 in 2800 aged 35–44 years and 1 in 2400 aged 45–54 years. As the prevalence and incidence of intermenstrual bleeding is much higher than that for postcoital bleeding, and the rate of endometrial cancer is only slightly higher than cervical cancer in the perimenopausal years, predictive values are even lower for intermenstrual bleeding.
The positive predictive value of a symptom for cancer increases as an individual moves along the pathway from community to primary to secondary care. As this study concerned a community population, applying these results to women who consult general practice or hospital clinicians should be undertaken with caution. The rate at which these women consult and the rarity of cancer suggests that even in these environments the symptoms are likely to have poor predictive value.