13Gynaecological cancer

Publication Details

General recommendations

1.

A patient who presents with symptoms suggesting gynaecological cancer should be referred directly to a team specialising in the management of gynaecological cancer, depending on local arrangements. [D]

Specific recommendations

2.

A woman’s first suspicion that she may have a gynaecological cancer may be when she experiences alterations in her menstrual cycle, intermenstrual bleeding, postcoital bleeding, postmenopausal bleeding or vaginal discharge. For a patient who presents with any of these symptoms, the primary health care professional should recommend a full pelvic examination including speculum examination of the cervix. [C]

3.

In a patient found on examination of the cervix to have clinical features which raise the suspicion of cervical cancer, referral should be urgent. A cytology test is not required before referral, and a previous negative cytology result is not a reason to delay referral. [C]

4.

Ovarian cancer is particularly difficult to diagnose on clinical grounds as the presentation may be with vague, non-specific abdominal symptoms alone (bloating, constipation, abdominal or back pain, urinary symptoms). In a woman presenting with any unexplained abdominal or urinary symptoms, abdominal palpation should be carried out. If there is significant concern, a pelvic examination should be considered if appropriate and acceptable to the patient. [D]

5.

Any woman with a palpable abdominal or pelvic mass on examination that is not obviously uterine fibroids or not of gastrointestinal or urological origin should have an urgent ultrasound scan. If the scan is suggestive of cancer, or if ultrasound is not available, an urgent referral should be made. [C]

6.

When a woman who is not on hormone replacement therapy presents with postmenopausal bleeding, an urgent referral should be made. [C]

7.

When a woman on hormone replacement therapy presents with persistent or unexplained postmenopausal bleeding after cessation of hormone replacement therapy for 6 weeks, an urgent referral should be made. [C]

8.

Tamoxifen can increase the risk of endometrial cancer. When a woman who presents with postmenopausal bleeding and she is taking tamoxifen urgent referral to a team specialising in the management of gynaecological cancer should be made. [C]

9.

An urgent referral should be considered in a patient with persistent intermenstrual bleeding and a negative pelvic examination. [D]

Vulval cancer

10.

When a woman presents with vulval symptoms, a vulval examination should be offered. If an unexplained vulval lump is found, an urgent referral should be made. [C]

11.

Vulval cancer can also present with vulval bleeding due to ulceration. A patient with these features should be referred urgently. [D]

12.

Vulval cancer may also present with pruritus or pain. For a patient who presents with these symptoms, it is reasonable to use a period of ‘treat, watch and wait’ as a method of management. But this should include active follow-up until symptoms resolve or a diagnosis is confirmed. If symptoms persist, the referral may be urgent or non-urgent, depending on the symptoms and the degree of concern about cancer. [C]

Introduction

The gynaecological cancers considered in these papers are those of the uterus (endometrium), ovary, cervix and vulva. Screening is excluded, and the focus is on patients presenting in primary care with symptoms or signs that might be related to cancer.

Incidence

Cancer of the cervix

In 1997 in England and Wales cervical cancer ranked as the seventh most common female cancer accounting for 2.5% of cancers in women. By 2001 there were 2,418 newly registered cases of cervical cancer.

These guidelines do not deal with screening. Guidelines on referral for colonoscopy on the basis of cervical cytology results has been published recently by the NHS Cancer Screening Programme (NHSCSP, 2004).

Ovarian cancer

Classification of ovarian cancer is complicated as it is not a single disease but rather a group of cancers which arise from different cell types.

There were 5,817 new registrations of ovarian cancer in England and Wales in 2001. The disease occurs mainly in post menopausal women peaking in the 70–74 years age group.

Over 90% of primary ovarian malignancies are epithelial adenocarcinomas arising from the surface epithelium. Subgroups of the epithelial malignancies include serous adenocarcinoma, and endometroid, mucinoid and clear call cancers.

Cancer of the uterus

There were 5,490 cases of cancer in the uterus (endometrial cancer) in England and Wales in 2001. Incidence is low below the age of 25 years and peaks at approximately 60 years.

Endometrial cancer is almost always a disease of postmenopausal women, and is associated with obesity, low parity and late menopause. Approximately 60–70% are adenocarcinomas.

Figure 1. Newly diagnosed cases of gynaecological cancers in 2001 in England and Wales. (77).

Figure 1Newly diagnosed cases of gynaecological cancers in 2001 in England and Wales. (77)

Mortality

Cancer of the cervix

In 2002 there were 995 deaths caused by cancer of the cervix. Mortality from cervical cancer increases steadily from age 30 years and peaking at age 80 years.

Ovarian cancer

In 1997 ovarian cancer accounted for 6% of all cancer deaths in women. By 2002 deaths caused by cancer of the ovary totalled 4,097. Mortality is steady and low until ages 40–44 when it begins to rise with age. Epithelial ovarian cancer in the second and third decade are uncommon but are treatable if diagnosed without delay.

Cancer of the uterus

There were 919 deaths caused by uterine cancer in 2002. Mortality rates are low in those aged 60 years or less.

Vulval cancer

There are around 1000 new cases of vulval cancer each year in England and Wales, accounting for less than 5% of all gynaecological cancers. In 2002, there were 332 deaths from vulval cancer. It usually presents in the elderly, most commonly aged 70 or over. The aetiology is unclear, although there is an association with vulval dystrophy and vulval intraepithelial neoplasia (VIN). Around 90% of vulval cancers are squamous cell in origin, and 5% are melanomas.

Figure 2. Mortality figures from gynaecological cancers for 2002 in England and Wales. (78).

Figure 2Mortality figures from gynaecological cancers for 2002 in England and Wales. (78)

Review of referral audits

The review(13) identified 45 relevant audits. The proportion of two week referrals found to be in accordance with the symptoms listed in the guidelines ranged from 42% to 100% (19 audits). The proportion of patients referred under the two week system who were found to have cancer ranged from 0% to 25% (17 audits). The proportion of patients with cancer who had been referred under the two week system ranged from 0% to 34% (six audits). 64% to 94% of two week referrals had been considered to be clinically appropriate (six audits).

13.1. Symptoms and Signs

13.1.1. Key Clinical Question

Which symptoms, signs and other features raise a suspicion of gynaecological cancers (cervix, ovary, endometrium, and vulva), and those that make cancer less likely as a diagnosis?

13.1.2. Evidence Question

In people attending primary care services with gynaecological symptoms, which symptoms and signs and other features including family history when compared with the ‘gold standard’ are predictive of a diagnosis of cancer; and which symptoms and signs are not?

13.1.3. Evidence Statements

Expert opinion suggests that abnormal bleeding (postcoital, postmenopausal and intermenstrual bleeding) can be presenting features of cancer(IV)

Cancer of the cervix

The incidence of cervical cancer rises rapidly in the years of 25–35 has peaks at around age 35–39 years (20/100,000 population) and continues at a similar rate peak at 75–79 years (20/100,000). (III)

Cervical cancer is rare in women under 20 years of age after this the incidence rises, reaching a peak at 35–39 years of age (20/100,000) and this rate is maintained at a similar level. (III)

Ovarian

The incidence of ovarian cancer is below 10/100,000 in women aged 40 and below, and 60/100,000 by age 70 years. (III)

Ovarian cancer may present with a variety of non-specific symptoms including tiredness, abdominal discomfort, gastrointestinal and urinary symptoms, and back pain. (III)

Signs associated with ovarian cancer include increased abdominal size and a palpable mass. (III)

Cancer of the uterus

Endometrial cancer is rare below the age of 35 years, but is more common after the age of 50 (50/100,000 population). (III)

Postmenopausal bleeding is regarded in expert opinion as an indication for investigation for endometrial cancer. (III)

Vulval cancer

The presenting symptoms of vulval cancer included vulval bleeding, pruritis and pain. (III)

The presenting signs of vulval cancer include a lump or other lesion. (III)

Introduction

Evidence about the diagnostic value of symptoms and signs of gynaecological cancer among patients presenting in primary care is limited. There were no secondary studies, and the primary studies largely consisted of case series and case control studies and surveys. The studies only included women after referral and diagnosis, and involved retrospective collection of information about the symptoms and signs before or at diagnosis. No study was identified that had been undertaken in primary care to investigate the presenting symptoms or signs that may, or may not, be explained by cancer. One article was identified for inclusion on the signs and symptoms of cervical cancer.

Studies of the presenting features of endometrial cancer were generally concerned with postmenopausal bleeding and consequently are discussed in the context of investigations.

Six studies addressed the signs and symptoms of ovarian cancer. However, these compared the signs and symptoms of women presenting with borderline or ovarian cancer and early and late stage diagnosis.

Guidelines
(SIGN, 2002)(239)

The SIGN guidelines on investigation of post-menopausal bleeding did not cite primary studies on indications for referral, the only reference being to the Department of Health (2000) guidelines. The guideline stated:

General practitioners should take into account the pattern of bleeding, its relationship to the use of HRT and patient preferences when considering referral. Concern by either general practitioner or patient about the possibility of PMB signalling endometrial cancer constitutes sufficient grounds for referral(2)

This statement was followed by a level D recommendation ‘The risk of endometrial cancer in non-HRT users complaining of post-menopausal bleeding and in HRT users experiencing abnormal bleeding is sufficient to recommend referring all patients for investigation.’

Statements of recommended best practice based on the clinical experience of the guideline development group were then listed:

The following questions should be asked in the assessment of patients with abnormal bleeding on HRT:

  • When does bleeding occur with respect to the oestrogen and progestogen phase? (Women on sequential regimens should ideally not experience withdrawal bleeding before completion of the progestogen component of the preparation).
  • How long does the bleeding last and how heavy is it?
  • Was there a period of amenorrhoea before HRT was started?
  • Is there a problem that suggests poor compliance?
  • Is there a reason to suspect poor gastrointestinal absorption?
  • Is the patient taking any other drugs?
  • Women presenting with post-menopausal bleeding should receive a pelvic examination at some stage during the course of clinical assessment.
  • Whether or not to continue HRT use prior to investigation may depend on the patient’s wishes and how long she has to wait for investigations, but there is no specific reason for discontinuing it.

Cancer of the cervix

(Viikki et al, 1998)(240)

This study investigated the predictive value of bleeding for detecting subsequent gynaecological or urinary cancers among women screened negative for cervical cancer in Finland. Data from the Finnish Mass Screening Registry and National Cancer Registry were used to investigate the long term significance of bleeding symptoms. The mean length of follow-up was 7.0 years. A total of 37,596 screened negative women in the national population-based mass screening programme for cervical cancer were classified as having reported bleeding symptoms when screened. These were categorised into bloody discharge, coital bleeding, irregular bleeding, postmenopausal bleeding, and were followed up (1985–1994) to monitor the subsequent risk of cancer.

The prevalence of postmenopausal bleeding among the 37,596 women (all ages) was 0.2%, bloody discharge was 1.1%, coital bleeding 0.7%, and irregular bleeding 3.9%. During follow up 753 cancers were observed among women with bleeding symptoms; 197 (26%) of these were gynaecological. The relative risk of uterine cancer was 3.6 in women with postmenopausal bleeding. The RR of cervical cancer was 1.1, (95% CI 0.8–1.4), not significantly increased during follow up for a maximum period of ten years. Women with bloody discharge had an elevated risk of gynaecological cancers, attributable to uterine cancer (SIR 2.2, CI 1.3–3.4). Coital bleeding was rare and not associated with gynaecological cancer (SIR 1.0). Irregular bleeding was associated with an increased risk of cancer of corpus uteri (SIR 1.8, 1.3–2.5). Risk of uterine cancer increased with any bleeding symptom (SIR 2.1, 95% CI 1.6–2.6) but the RR associated with postmenopausal bleeding was 3.6 (95% CI 2.0–6.0).

It should be noted that the symptoms were reported at the time of screening and it is not clear whether the findings can be related to people consulting with these symptoms.

Ovarian cancer

No secondary studies were identified. No studies of patients presenting in primary care were identified. The primary studies included patients in secondary care after diagnosis. Consequently, they only provided limited evidence to aid differentiation between symptoms and signs that would or would not indicate the need for referral.

Primary studies
(Flam et al, 1988)(241)

The symptomatology of ovarian cancer was retrospectively reviewed in a case series of 362 patients who had been referred to a single specialist Swedish centre. The disease was classified at early (stages IA–IIA) or advanced (stages IIB–IV) at diagnosis and no follow up was reported. Patients were asked to give an account of their initial symptoms and those leading to their decision to seek a consultation. The main abdominal symptoms could be divided into three groups: 1) abdominal swelling with increased abdominal girth, 2) pain and 3) gastro-intestinal symptoms such as heartburn and eructation.

The presentation of early and late stage cancer was influenced by the Swedish health care system, patients being able to choose whether to consult a primary care practitioner or gynaecologist.

The most common initial symptoms were abdominal swelling and/or palpable tumour, pain and gastro-intestinal symptoms. The initial symptoms, however, were not necessarily those that prompted patients to seek medical advice. The most common reason for seeking advice was pain in the early group, but abdominal swelling in the advanced group (27.9%). Gynaecological disease was suspected by 55.2% of the early group and 37.9% of the advanced group.

(Vine et al, 2001)(242)

The study investigated the types and duration of symptoms among women with invasive versus borderline ovarian tumours. The people included were women 20–69 years of age, diagnosed histologically as having primary epithelial invasive or borderline (i.e. having histological features of benign as well as malignant disease) ovarian cancer between 1994 and 1998. They were identified from 39 hospitals in the US. Information about symptoms was obtained by interviews conducted in the homes of study participants. To reduce the risk of recall bias, only cases interviewed within six months of diagnosis were included in the study. A total of 1278 cases met age and location of residence criteria. Cases were excluded if English was not spoken or patients were not mentally competent (25), diagnosis greater than 6 months prior to interview (296), critically ill or dead (69), or untraceable (15), physician did not give consent to contact (14) and refusal to participate (92), resulting in 767 (60.0%) completed interviews.

The lack of information about tumour stage and size prevented assessment of the effect of these factors on the reporting and duration of symptoms as well as delay in diagnosis. The percentage of women with symptoms was significantly higher in invasive versus borderline disease (any symptoms 92% versus 84%, P=0.001; pelvic discomfort 71% vs. 66%, P<0.05; bowel irregularity 47% versus 35%, P<0.001). Women with borderline disease had symptoms for longer periods of time than those with invasive disease or pelvic discomfort (P<0.001), bowel irregularity (P=0.002) and urinary frequency/urgency (P=0.011). Pre-diagnostic symptom duration was longer among borderline versus invasive cases (median six vs. four months, P<0.001). Women with invasive cancers were significantly older (mean age: invasive, 53.0 years; borderline 44.7 years, P<0.0001), but there was no difference between women with invasive and borderline tumours with respect to race, education, or household income. Borderline and invasive cases reported similar types of symptoms. However, borderline cases were twice as likely as invasive cases to report not having had symptoms (16 vs. 8%, P=0.005), and twice as likely as invasive cases to be diagnosed through routine examination (28 vs. 16%, P=0.001). Invasive cases were more likely to be diagnosed because of symptoms (62 vs. 48%, P=0.002).

(Goff et al, 2000)(243)

This Canadian study focused on symptoms and other factors that may contribute to the delayed diagnosis of ovarian carcinoma. It was a questionnaire survey of women in cancer support groups, and women were invited to copy the questionnaire to other patients. A total of 1725 questionnaires were returned from women in 46 states and four Canadian provinces (1327 originals and 398 copies). The response rate for the initial survey was 88%. The median age of the surveyed patients was 52 years (range 18–84) and 13% were older than 65 years and 70% had Stage III or IV disease classified according to the International Federation of Gynaecology and Obstetrics (FIGO). The symptoms asked about were selected from published reports and small focus groups. In the survey selection bias may have occurred because the women who participated in this study were those who chose to subscribe to a newsletter or those active in support groups.

In response to whether they had symptoms before the diagnosis of ovarian carcinoma, 5% of patients reported they had none, 61% reported increased abdominal size, 57% abdominal bloating, 47% fatigue, 36% abdominal pain, 31% indigestion, 27% urinary frequency, 26% pelvic pain, 25% constipation, 24% urinary incontinence, 23% back pain, 17% pain with intercourse, 16% unable to eat normally, 14% had a palpable mass, 13% vaginal bleeding, 11% weight loss, 9% nausea, 3% bleeding with intercourse, 1% deep venous thrombosis and 1% diarrhoea. When responses were grouped according to symptom categories, 77% reported abdominal symptoms, 70% gastrointestinal, 58% pain, 50% constitutional, 34% urinary and 26% pelvic. Only 11% of women with Stage I/II and 3% with Stage III/IV reported they were completely asymptomatic before their diagnosis. Thirteen percent of participants reported being told by their provider that nothing was wrong, 6% were diagnosed with depression, 12% stress, 6% constipation, 15% irritable bowel syndrome, 9% gastritis and 47% were given other diagnoses. Only 20% of patients were told initially they might have ovarian carcinoma.

Women who had the most symptoms were significantly younger. Women with advanced disease were significantly more likely to have symptoms than those with early stage disease. The types of symptoms between both groups were similar however. Those who ignored their symptoms were significantly more likely to have more total symptoms and advanced stage disease compared with those who did not (85% vs. 74%; P=0.002).

(Olson et al, 2001)(244)

A retrospective case control study was conducted in two US hospitals to assess the presence and duration of various symptoms of ovarian cancer and the use of medications in comparison with healthy women. Symptoms of ovarian cancer in recently diagnosed patients (cases N=168) were compared with those experienced by healthy women in the community using a case-control design (N=251). Between 1994 and 1997, women diagnosed with ovarian cancer, aged 18 or above, resident in the US and English or Spanish speaking were approached while awaiting surgery. The mean time from diagnosis to interview was 4.7 months, and 73% were interviewed within nine months. Women were asked whether they had experienced one or more of eight symptoms or used three types of medication in the six to 12 months before diagnosis. Affected patients were grouped into those with earlier and later stage disease.

Recruitment difficulties in an urban area led to a variety of controls being used. One group of controls was randomly selected from the community (N=81) matched by five year age groups, and another through commercial mailing lists (N=78) matched by demographic, socioeconomic and lifestyle factors. Convenience controls consisting of friends of cases and other women (N=92) were also used. A lengthy questionnaire and the collection of biologic specimens led to a low response rate. Although the controls included a convenience sample and a randomly chosen community group, the results did not differ in meaningful ways when the convenience controls were excluded.

The symptoms were selected based on reviews of earlier reports in the literature and in consultation with clinicians. The most common symptoms among cases were: unusual bloating, fullness and pressure in the abdomen (71%); unusual abdominal pain or lower back pain (52%); and lack of energy (43%). The proportions of controls reporting these symptoms were 9%, 15% and 16% respectively, resulting in odds ratios and 95% CIs of 25.3 (15.6, 40.9), 6.2 (4.0, 9.6), and 3.9 (2.5, 6.1), respectively, for these symptoms. Bloating, fullness and pressure was of more recent onset among cases than controls (4.9 months compared with 7.6 months, P=.01). Lack of energy was noted by 43% of the cases and 16% of the controls (odds ratio 3.9, 95% CI 2.5, 6.1). Patients who experienced bloating, fullness and pressure were more likely than controls to report that the symptoms were constant. Most of the symptoms were experienced for a longer period of time by women with early rather than late stage disease.

The exact response rates were not mentioned. The study was reported to be limited by relatively small numbers of cases, especially women with early disease, and 35% of affected patients mentioned other symptoms that were not listed on the questionnaire. The most common additional symptom was pain in the side or ribs, mentioned by seven.

(Smith et al, 1985)(245)

This US case series evaluated characteristics of ovarian cancer symptoms, their perceived cause, and delay in seeking a diagnosis associated with stage, grade and histologic features of disease at diagnosis among patients with cancer of the ovary (N=83) identified in the Iowa National Cancer Institute-Surveillance, Epidemiology, and End Results (NCI-SEER) population-based cancer registry. Cases had been diagnosed 1980 to 1982. Of a total of 107 eligible cases, data about 82 were obtained. Twenty-five were eliminated either because they could not be interviewed due to severe illness, refusal or physician’s refusal of permission, or death, or lack of staging at the time of study.

Patients were asked if they had experienced any symptoms that prompted them to seek a diagnosis. Those who said ‘No’ (N=26) were asked how their cancer had been discovered; those who said ‘Yes’ (N=56) listed symptoms they noticed before diagnosis and those that convinced them to seek medical attention. The sample included a large number of asymptomatic patients more likely to have later stage disease. The cohort did not include older patients for whom the results may be less applicable. Only pain was significantly more likely to differentiate those with extensive disease. In contrast, localised disease was associated with a greater probability of reporting urination problems, irregular menstrual cycles and fatigue.

A majority (56, 68.3%), regardless of stage, had experienced symptoms that prompted a consultation. Asymptomatic women were more likely to be identified with later stage disease (P<0.10). Among those (N=26) believing they had no symptoms before diagnosis, in 34.6% the tumour was detected during a yearly health check. The most common number of symptoms occurring together was two (72.2%), with abdominal swelling most likely to be identified with other conditions: fatigue (23.5%), urination problems (17.6%), and pain (17.6%). Swelling, pain and fatigue were commonly seen together (29.4%). Only abdominal pain and swelling were significantly associated (P<0.05) with later stage disease. Pain was likely to convince women to seek a diagnosis. Those aged 40–49 years were more likely to report symptoms than patients in other age groups (P<0.05). No relationship between age and type or number of symptoms was found, nor associations with other sociodemographic factors. Less frequently noticed symptoms were irregular vaginal bleeding, metrorrhagia, indigestion and urination problems (frequency or difficulty). Symptoms were viewed less seriously if they were believed to be related to indigestion or menopausal conditions. Irregular menstrual cycles often convinced patients with early-stage cancers to seek a diagnosis.

(Wikborn et al, 1993)(246)

This case series investigated symptoms by reviewing the clinical records of patients with ovarian cancer to identify information from first consultation to operation and diagnosis. A total of 160 patients diagnosed in a Swedish specialist centre between 1981 and 1986 with epithelial ovarian cancer that could be staged constituted the study population.

No specific group of symptoms could be linked to type or stage of ovarian cancer. Gastrointestinal symptoms were more common in patients with histological class IC tumours. Only 21% complained of gynaecological symptoms. The majority of women did not experience symptoms in the genital organs. Women with class IC cancer had significantly more advanced disease than those with 2C–5C cancer as 77% had a stage III–IV tumour compared with 40% of class 2C–5C patients. The mean age was 62.6 years (range 25–87 years). Several women had more than one type of symptom, pain and abdominal swelling being the most common combinations. Irrespective of stage, 37% had symptoms related to the bladder; approximately 65% had pain and 60% had abdominal swelling. Gastrointestinal and general symptoms were less common in stage 1 than in higher stages. This was not the case with tumour classes 2C–5C disease.

Vulval cancer

Secondary studies
(Ghurani and Penalver, 2001)(247)

This paper is a recent authoritative review. The most common symptoms were reported to be pruritus, a visible or palpable mass, pain, bleeding, ulceration, dysuria, and vaginal discharge. The authors recommended that any vulval lesion discovered on physical examination should be biopsied to rule out neoplasm.

Primary studies
(Rosen and Malmstrom, 1997)(248)

This study was a retrospective review of the hospital records of 328 patients with histologically confirmed primary invasive vulval cancer (Sweden). It was performed to evaluate the survival after treatment of vulval cancer in relation to various prognostic factors (FIGO staging, tumour grading, age at diagnosis, heredity for any cancer, childbirth, and prior history of any cancer).

Mean and median age at diagnosis was 69 (range, 26–105) years. The most common presenting symptoms were pruritus (24%), smarting pain (15%), and a vulval lesion (15%). Patient’s median delay was six months, and the average was 16 (range, 0–360) months. Squamous cell carcinoma was the most common histological form of vulval cancer, constituting 91.4% of the cases (N= 300). Melanoma constituted 3% (N = 10), Paget’s disease 2.4% (N = 8), cancer of the Bartholin’s glands 1.8% (N = 6), adenocarcinoma 0.6% (N = 2), and basal cell carcinoma 0.6% (N = 2). Survival analyses were limited to the 300 patients with squamous cell vulval cancer.

The majority of patients with squamous cell vulval cancer were stages I (35%) or II (37%) at diagnosis, 36% were well-differentiated tumours, 43% moderately differentiated tumours, and 15% poorly differentiated tumours. There were significant differences in survival when comparing patients older than mean age at presentation (69 years) with the patients who were younger than mean age (P < 0.01). There were significant (P < 0.00001) differences in corrected survival times between different FIGO stages: five year survival rate was 93% for stage I, 60% for stage II, 40% for stage III, and 13% for stage IV. Histologic grade was also shown to be a significant prognostic marker for survival (P = 0.02): well-differentiated tumours had a five year survival rate of approximately 70% while moderately or poorly differentiated tumours had a five year survival rate of approximately 55%. Both parity and previous history of cancer did not influence survival times significantly.

(Messing and Gallup, 1995)(249)

The authors undertook a retrospective review of the hospital medical records (USA) of 78 women treated for squamous cell carcinoma of the vulva over a period of 15 years. They compared women younger than 45 years with those 45 years and over for historic risk factors, treatment modality, and outcome.

The mean age was 60.7 years (median 63, range 29–91). Two age peaks were noted at ages 50 and 70. Over the study interval, the average presenting age of these patients decreased from 69 to 55 years. Eighteen (23%) of the cases were in women younger than 45 years of age. The median duration of symptoms before seeking medical care was six months. Patients presented with complaints of a lesion, lump, or pain in 70% of cases. There was no significant difference in the duration of symptoms for younger versus older women.

Women under 45 were found to have a stronger history of condyloma (P<0.001, 95% confidence interval 3.69–87.96). There was no significant difference by age in smoking history, alcohol consumption, or tumour size. Older women were more likely to have advanced stage disease (P=0.03, 95% CI 0.43–0.91) but no metastatic disease. The median tumour size at presentation was 4cm (range 0–27). Lesion size over 2cm was significantly associated with the presence of metastatic disease (P < 0.001). The following were associated with decreased survival: FIGO stage IV (P <0.001, 95% CI 1.6–5.1), presence of metastases (P < 0.001, 95% CI 1.5–3.6), and tumour size greater than 2cm (P=0.002, CI 0.09–0.34). There was no detected difference in survival for women in either group.

(Jones et al, 1997)(250)

The objective of this study was to determine trends in the clinicopathology of vulval squamous cell carcinoma over the past two decades, with particular reference to vulval intraepithelial neoplasia (VIN) during this time. The authors reviewed retrospectively the clinical records of two groups of women presenting with squamous cell carcinoma of the vulva to a New Zealand gynaecological oncology unit. One group involved 56 cases presenting between 1965 and 1974, and the other involved 57 cases presenting between 1990 and 1994.

The mean age at presentation was 68.4 (44–92) years (median 72 years) in the 1965–1974 cohort, and 69.2 (22–93) years (median 71 years) in the 1990–1994 cohort. In the 1965–1974 cohort, only one patient was younger than 50 years of age, whereas in the 1990–1994 cohort, 12 women (21%) were younger than 50 years (P = 0.001). There were no statistical differences in FIGO stage between the two cohorts.

When stratified according to age, 11 of 13 women younger than 50 years, compared with 10 of 100 women older than 50 years of age, smoked cigarettes (P < 0.001). Ten of the 13 women younger than 50 years of age, compared with 13 of 100 women 50 years of age or older, had warty and/or basaloid VIN III associated with their invasive carcinoma (P < 0.001). Multiple lower genital tract neoplasia was also more common in women younger than 50 years of age (P < 0.001).

Warty and basaloid VIN was associated with 16 of 19 (84%) warty or basaloid carcinomas and with seven of 94 (7.4%) typical squamous cell carcinomas (P< 0.001). In contrast, non-neoplastic epithelial disorders were associated with 55 of 94 (58.5%) typical squamous cell carcinomas and with none of the 19 basaloid or warty carcinomas.

(Sturgeon et al, 1991)(251)

The authors identified 2,948 cases of in situ and 2,346 invasive squamous cell tumours of the vulva diagnosed between 1973 and 1987 from population-based cancer registries (USA), in order to examine recent trends in the incidence of vulval cancer.

The annual incidence of in situ vulval carcinoma for all races combined nearly doubled from 1.1 to 2.1 per 100,000 woman years, during the period from 1973 to 1976 and 1985 to 1987. The largest proportional increase occurred among white women <35 years old, for whom the rate nearly tripled. Increases were more modest among black women than among white women, with the rate not quite doubling among black women <35 years old. In situ rates among blacks of all ages were higher than those among whites before 1977, but the black-white differential had diminished in more recent years. The peak in situ rate has shifted over time from women > 54 years to women aged 35 to 54.

The invasive squamous cell carcinoma incidence for all races combined was relatively stable over 1973 to 1976 and 1985 to 1987. Rates in each age group were also relatively steady, although among white women they tended to decline among those aged >55. Little racial difference was evident under age 35; rates were higher at ages 35 to 54 among black women and at ages >54 among white women. In contrast to in situ cancers, invasive rates increased steadily with age.

Risk Factors

Cancer of the cervix
(Parikh et al, 2003)(252)

The authors conducted a meta-analysis after pooling the data from previously reported case-control studies (N= 57) of cervical cancer or dysplasia, which contained individual-level information on socio-economic characteristics, to investigate the relationship between cervical cancer, social class, stage of disease, geographical region, age and histological type.

Overall, an increased relative risk of dysplasia and cervical cancer with decreasing social class was observed. Women in the middle social class group were at approximately a 26% increased risk of cervical disease (95% CI 17–36%, whereas women in the lower social class tertile were at approximately 80% increased risk when compared to women in the upper tertile (95% CI 69–92%). These elevated risks persisted after analysis was restricted to those studies which included only women aged <50 years (97% increase in risk of invasive cancer for the low socio-economic group; 95% CI 80–115%, and 58% increase in risk of dysplasia for the low socio-economic group; 95% CI 41–78%). When stratified by geographical region, the increased risk identified in studies that originated from Western Europe appeared to be only moderate, with a 45% increased risk of cervical disease in the low social class group as opposed to the high social class group (95% CI 29–62%). When the analysis was restricted to studies that only included cases of cervical cancer, the increase in risk between social class and invasive cervical cancer was reduced to 28% (95% CI 10–49%) for Western European studies.

There was significant unexplained heterogeneity in most of the pooled odds ratios, which might have been possible because of the inability to control for variables such as background HPV prevalence.

(Paley, 2001)(253)

This authoritative review reported that risk factors for cervical cancer are well defined and those with the most impact on risk of developing cervical cancer include early sexual activity, smoking, multiple sexual partners and an immunocompromised state. Due to the link between sexual activity and cervical cancer, infection with human papillomavirus (HPV - the most common sexually transmitted viral disease) is believed to be a strong predictor of cervical cancer.

Ovarian cancer
(Bell et al, 1998)(254)

Although a wide variety of risk factors have been proposed for ovarian cancer, this review of the use of CA125 for screening identified four main risk factors of epithelial ovarian cancer which are summarised in Table 1.

Table 1. Major risk factors for epithelial ovarian cancer(254).

Table 1

Major risk factors for epithelial ovarian cancer(254).

Information from pooled data from 12 US case control studies indicated that pregnancy and oral contraceptives had a protective effect, with risk reducing for each term of pregnancy or length that the contraceptive is used. The findings indicate that the strongest risk factor for epithelial cancer is a first or second degree relative with ovarian cancer although they concede that only 7% of women diagnosed with ovarian cancer report a family history of the disease.

(Paley, 2001)(253)

This authoritative review concluded that the majority of women who are diagnosed with ovarian cancer have no identifiable risk factors. Nevertheless there are some risk factors that have been identified in some cases, the most significant of which being genetic predisposition, with up to 10% of epithelial ovarian cancer being familial. Additionally this review suggests that 90% of these familial ovarian cancers are attributable to mutations in BRCA1 or BRCA2 genes, resulting in a life time risk for ovarian cancer of 40–60% in women with BRCA1 mutation.

(Stratton et al, 1998)(255)

A UK systematic review of published case-control and cohort studies sought to estimate the relative and lifetime risk of ovarian cancer in women with various categories of family history. The outcome measures of relative and lifetime risks of developing ovarian cancer were calculated for women with 1) an unaffected first degree relative, 2) an affected mother, 3) an affected sister, and 4) women with more than one affected relative. Published articles were identified using the MEDLINE databases from 1966 to 1998. Data from electronic databases were supplemented by hand searches. Sixteen studies were identified. Three of these were retrospective cohort studies and 13 were case-control studies. One study was excluded because of probable selection bias.

The 15 studies were assessed for homogeneity. Although there was heterogeneity in the studies used to estimate risk in first-degree relatives, this did not alter the estimate of the pooled relative risk. Two studies reported the relative risks to first-degree relatives according to age at diagnosis or death of the index case. The pooled estimate of RR was 1.7 (95% CI 1.2–2.5) where the index case was diagnosed or dies from ovarian cancer before the age of 40, compared with 3.8 (95% CI 2.6–5.5) if the index case was diagnosed or died at an older age. Four studies reported RRs according to the ages of first-degree relatives. For women younger than 50 with an affected first degree relative the RR was 2.9 (95% CI 1.9–4.3), while for women older than 50 with an affected first degree relative the risk was two (95% CI 1.5–2.5). The risk to daughters of an affected mother was given in three case-control studies which provided a pooled estimated RR of six (95% CI 3.0–11.9). The risk to mothers with an affected daughter was given by two cohort studies and one case-control study. The estimated RR was 1.1 (95% CI 0.8–1.6). Four studies reported risks associated with having an affected sister. The pooled estimate from these studies gave an RR of 3.8 (95% CI 2.9–5.1). Only two case-control studies and no cohort study examined the risks associated with having a second degree relative with ovarian cancer. The pooled relative risk estimated from these studies was 2.5 (95% CI 1.5–4.3). Two studies examined the risks involved in having more than one affected relative (either first or second degree) with ovarian cancer. The pooled risk estimate was 11.7 (95% CI 5.3–25.9).

Cancer of the uterus
(Paley, 2001)(253)

This recent authoratitive review of endometrial cancer summarised risk factors and considered the case for screening. The review reported that principal risk factors for endometrial cancer are conditions that result in high oestrogen levels. Therefore, the risk factors include diabetes, obesity, chronic anovulation, oestrogen secreting tumours, tamoxifen use and unopposed exogenous oestrogen administration.

Vulval cancer
(Ghurani and Penalver, 2001){979]

In this authoritative review, factors associated with the development of vulval cancer were reported as including granulomatous infection, herpes simplex virus, and human papillomavirus. The DNA of human papillomavirus has been identified in invasive carcinomas and preinvasive lesions of the vulva. Other factors associated with vulval cancer include chronic immunosuppression, hypertension, diabetes and obesity.

13.2. Investigations

13.2.1. Key Clinical Question

Should any investigations be undertaken in primary care, before referral in women with suspected gynaecological cancer?

13.2.2. Evidence search question

In women attending primary care services with gynaecological symptoms, which investigations when compared with the “gold standard” are predictive of a diagnosis of cancer, and which are not?

13.2.3. Evidence Statements

Despite an effective screening program cervical cancer occurs among women who have had, and who have not had, regular cervical screening tests, and even among women who have had two or more smears in the past 5 years (III).

There is no evidence about the value of investigations that may be used routinely in primary care to investigate women suspected of having cervical, endometrial, ovarian or vulval cancer (III).

Cancer of the cervix

Primary studies
Janerich et al, 1995{581}

Patients with invasive cervical cancer diagnosed in Connecticut from March 1985 to February 1990 were identified for the study. Cases were selected from new cases as soon as possible after diagnosis in 35 hospitals. Information was obtained through interviews for 481 (72%) of the 664 eligible patients. Verification of invasion was based on a biopsy or hysterectomy report, or both. In cases where invasion had not definitely been established, pathology slides from biopsies or hysterectomies or both were requested and reviewed to establish the diagnosis.

A total of 137 cases (28.5%) occurred among women who had never had a Pap test, and another 113 cases (23.5%) in women whose last Pap test was more than five years before diagnosis of cervical cancer. The average age of women who were never screened was 64.5 years compared with 46.5 years for the remainder of the 481 case patients. Of the 481 patients, slides could be obtained for 137, and of these 6.9% were classified on re-evaluation as misread i.e. had originally been incorrectly classified as normal. Delay in the follow-up of suspicious smears occurred in 52 of the 481 cases (10.8%).

(Carmichael et al, 1984)(256)

A Canadian retrospective review was conducted of the cytologic history of 245 patients who developed invasive carcinoma of the cervix and were registered with the Ontario Cancer Foundation Clinic between January 1973 and October 1982. The aim of the study was to delineate causes for failure of cervical cytologic screening in a group of patients who eventually developed invasive cervical carcinoma. Three groups of patients were identified. Group 1 included 149 patients (60.8%) who had never had a cervical cytology examination, group 2 included 26 patients (10.6%) whose cytology history in terms of frequency of examination and or timing was considered to be unsatisfactory and group 3 included 70 patients (28.6%) whose cytology history was satisfactory. A satisfactory cytology history was defined as two or more smears within five years, and three or more smears within ten years. Smears within three months of the patient’s anniversary date were excluded. The original smears that were obtained from the identified laboratories were requested and reviewed by a senior pathologist who was not aware of the original cytology diagnosis.

Fifty three (35.6%) of the patients in group 1 had stage 1 disease. Stage 1 disease was present in 16 patients (61.5%) of group 2 and in 55 patients (78.6%) of group 3. There was no significant difference between the three groups with respect to site of residence or access to the health care system. Of the patients in group 3, 20 (28.6%) had normal findings and 50 (71.4%) had abnormal cytology findings. A review of 229 original cervical smears revealed that 52 (17.4%) had been significantly undercalled (i.e. the severity of abnormalities had been adequately identified), but only 21 (7.0%) had been undercalled as normal. In these patients, staging was unrelated to screening.

(Woodman et al, 1997(257)

A questionnaire survey of all general practices (N=111) and family planning doctors (N=62) in Manchester Health Authority was undertaken to determine why more smears were taken in primary care than were scheduled by the screening programme. An 82% response rate was obtained. Questionnaires were addressed to the senior partner in the practice with a request that they determine the most appropriate person within the practice to complete it.

Ninety-one general practices (82%) and 50 family planning doctors (81%) eventually responded; 22 (24%) of the questionnaires returned by general practices had been completed by the practice nurse. The indications for additional smear tests most frequently cited by responders were postcoital (88%), postmenopausal (84%), or intermenstrual bleeding (55%), genital warts (87%) and multiple sexual partners (52%). Forty-six percent maintained that a woman should have a repeat test within one year of her first ever test. Family planning doctors were less likely than general practices to take an extra smear if a woman was starting the oral contraceptive pill, having an intra-uterine contraceptive device (ICD) inserted, or attending for a postnatal check; or if she had a history of multiple sexual partners.

Ovarian cancer

Secondary studies
(Royal College of Obstetricians and Gynaecologists, 2003)(254)

This guideline was developed following a review based on a search of relevant databases for publications from 1966 onwards. It was recommended that ovarian cysts in postmenopausal women should be assessed using CA125 and transvaginal grey scale sonography, and that a ‘risk of malignancy index’ should be used to select those women who require primary surgery in a cancer centre by a gynaecological oncologist. The guidelines did not deal with primary care assessment.

(SIGN, 2003)(254)

SIGN has published guidelines on epithelial ovarian cancer. Although the guidelines were principally focused on managed of diagnosed cases, it was recommended that GPs should include ovarian cancer in the differential diagnosis when women present with recent onset persistent non-specific abdominal symptoms (including women whose abdominal and pelvic clinical examinations appear normal). No studies were identified that assessed the usefulness of the measurement of CA125 in women with vague abdominal symptoms and the guideline did not recommend the routine measurement of CA125. It was recommended that women with a pelvic mass should be referred to a gynaecologist irrespective of CA125 test results.

Primary studies
(Andolf et al, 1986)(258)

Ultrasound scan for detection of ovarian enlargements was performed in a target group of out-patients attending a specialist Swedish outpatient clinic for various reasons in the 40–70 years range. Overall 805 women were examined, in 99% of whom the ovaries and/or their vessels could be identified. The findings at the ultrasound examination were compared with those at pelvic examination, surgery and with subsequent histological examination (gold standard). All patients had a manual pelvic examination before the ultrasound scan. Pelvic examination was performed by one of the 30 or so experienced gynaecologists available at the clinic.

The findings of the manual examination were not available to the ultrasonographer. Pelvic examinations were performed by gynaecologists whilst all ultrasound examinations were made by one technician, a specially trained midwife. The ultrasound examination was performed with the full-bladder technique using a ‘real’ time sector scanner. The uterus and ovaries were measured in three planes and the volume of the ovaries was calculated using a simplified version of the ellipsoid formula: length × width depth × 0.52. Pathological findings were suspected in 83 of the 805 women at the first scan and were confirmed in 50 after a repeat scan, of whom 39 subsequently underwent surgery. None of the borderline or malignant ovarian lesions were found by manual pelvic examination.

Cancer of the uterus

Secondary studies
(Tabor, 2002)(259)

The review was limited to original research reports written in English concerning symptomatic women having vaginal ultrasonography before a diagnostic test and not receiving tamoxifen. Blinding did occur as those studies reporting only Doppler indices or abdominal scans, or those in which the ultrasound examination was done after the diagnostic test were excluded. A literature search was conducted using the MEDLINE database from January 1 1991 to September 30, 1997. A total of nine UK and Danish studies were identified from 48 in the meta analysis. The nine studies yielded data on 3483 symptomatic women. 710 women were premenopausal, 2407 postmenopausal who were not taking HRT and 366 taking HRT) and 330 symptomatic women with endometrial cancer (seven premenopausal and 293 postmenopausal not on HRT and 30 on HRT). In six studies a dilation and curettage was done, in two studies an endometrial biopsy was taken, and in one multicentre study a dilation and curettage or an endometrial biopsy was taken.

The value of endometrial thickness was used as a test for endometrial cancer in postmenopausal women with vaginal bleeding (symptomatic women). In all studies, a histologic examination was performed to determine whether or not the women had endometrial cancer at the time of screening. Studies were included in the analysis only if the authors’ original data could be obtained to calculate each centre’s median endometrial thickness in unaffected symptomatic women. A questionnaire was sent to the corresponding author of each paper requesting supplementary information. The authors were asked to give the mean, standard deviation, median and 10th and 90th centiles of endometrial thickness.

The median endometrial thickness in women with endometrial cancer was 3.7 times that in unaffected women with the same menopausal status and same hormone replacement therapy use category. The detection rate was 63% (95% CI 58, 69) for a 10% false-positive rate, or 96% (95% CI 94, 98) for a 50% false-positive rate. It was concluded that 4% of the endometrial cancers would still be missed with a false-positive rate as high as 50%. It underlined the importance of determining the median and distribution of endometrial thickness in each centre, and not using a fixed cut off. In the two centres which reported medians for premenopausal women who did not take HRT, the median endometrial thickness was 2–3mm higher than in postmenopausal women who did not take HRT (P<0.01 for each).

There were statistically significant differences in endometrial thickness between centres presumed to reflect differences in measurement techniques or the populations studied. The multiples of the median (MoM) endometrial thickness in different studies of women with endometrial cancer ranged from 2.1 to 5.9 multiples of the median. The overall median endometrial thickness was 3.7 multiples of the median.

Primary studies
(Gredmark, 1995)(260)

A Swedish prospective cohort study designed to investigate endometrial histopathology in a geographically defined population of 457 postmenopausal patients presenting with uterine bleeding. The main outcome measures involved the frequency of bleeding and its correlation to endometrial histopathology and in relevant cases to pathological conditions in cervix and ovaries. Dilation and curettage using general anaesthesia was performed on the 457 postmenopausal women. All women referred to the county gynaecological departments because of uterine bleeding, appearing one or more year after menopause were eligible for inclusion. The study covered an 18 month period between September 1986 and March 1988.

Menstrual status, obstetric and medical history as well as pharmacological therapy were recorded and a general physical and gynaecological examination was performed. Women using HRT (N=19) for vasomotor symptoms were excluded from the study. Two women who had undergone subtotal hysterectomy were also excluded from the study.

The incidence of postmenopausal bleeding decreased with increasing age while the probability of cancer as the underlying cause increased. The peak incidence of endometrial carcinoma was found in women between 65 and 69 years of age. The mean age of the women with bleeding was 61.4 years (41–91) and the median age when menopause occurred was 50.6 years. Endometrial histopathology showed: atrophy (50%); proliferation (4%); secretion (1%); polpys (9%); different degrees of hyperplasia (10%); adenocarcinoma (8%); not representative (14%); other disorders (3%). In six women a squamous carcinoma of the cervix was found and eight proved to have ovarian tumours.

A high percentage (14.2%) of the endometrial specimens obtained in this study were unsatisfactory. One fourth of these showed fibromuscular tissue, while the remaining samples did not contain any tissue that could be evaluated.

13.3. Delay and Diagnostic Difficulties

13.3.1. Key Clinical Question

In people attending primary care services with gynaecological symptoms, which psychosocial and socio-demographic factors are associated with delayed presentation? Which factors influence delay by the patient and which delay by the provider?

What diagnostic difficulties do primary care practitioners themselves report in determining whether a women who presents with gynaecological symptoms/signs may or may not need urgent referral with suspected cancer?

13.3.2. Evidence Question

In people attending primary care services with gynaecological symptoms, which psychosocial and socio-demographic factors are associated with delayed presentation? Which factors influence delay by the patient and which delay by the provider?

What diagnostic difficulties do primary care practitioners themselves report in determining whether a woman who presents with gynaecological symptoms/signs may or may not need urgent referral with suspected cancer?

13.3.3. Evidence Statement

Delay in the detection of ovarian cancer may be associated with the non-specific nature of the symptoms (III).

Delay

Introduction

There were few studies of diagnostic delay, and most were limited to a description of the time intervals to diagnosis and its association to disease characteristics.

Secondary studies

No papers were identified.

Primary papers
Cancer of the cervix

No studies of delay in diagnosis among symptomatic patients were identified

Ovarian Cancer
(Kirwan et al, 2002)(261)

The authors of this UK study undertook a retrospective review of the general practice records of 135 patients with epithelial ovarian cancer. The aim of the study was to identify referral pathways from primary care for women with ovarian cancer, and particularly to examine delays between the onset of symptoms and presentation to the general practitioner and delays between presentation and referral to hospital.

105 patients (78%) presented to the general practitioner within one month of developing symptoms and 64 (47%) within two weeks. Only 11 patients (8%) delayed more than three months before seeking medical advice. Primary symptoms in the patients’ notes were abdominal swelling (65), change in bowel habit (34), weight loss (11), backache (3), vaginal bleeding (15), and other (30). General practitioners referred 68 (50%) patients to hospital directly after their first consultation, 82 (60%) within two weeks, and 99 (73%) within one month. 36 patients (27%) experienced delays of over three months, half of whom were misdiagnosed as having irritable bowel syndrome. The mean age of the survivors was less than that of patients who died (63.7 years vs. 69.0 years, P=0.014).

Multivariate analysis with survival as the dependent variable identified age (odds ratio 0.96, 95% confidence interval 0.93 to 0.99), cancer stage III or more (0.15, 0.05 to 0.43), and non-specific symptoms (0.36, 0.14 to 0.89) as significant variables. The study suggests that delays attributable to the patient and general practitioner are roughly equal but that these intervals do not affect survival beyond 18 months in women with ovarian cancer.

(Goff et al, 2000)(243)

The study’s aim was to evaluate preoperative symptoms and factors that may contribute to delayed diagnosis for women with ovarian carcinoma. The authors conducted a postal survey among 1,725 women with ovarian carcinoma who subscribed to a newsletter about ovarian carcinoma. All women who returned the questionnaire were North America residents (USA and Canada).

The median age of the surveyed patients was 52 years (range, 18–84), and 13% were older than 65 years. At the time of diagnosis, 71% of respondents had FIGO stage III/IV disease.

95% of patients had symptoms before the diagnosis of ovarian carcinoma. Duration of symptoms was reported as two months or less by 30% of patients, three to six months by 35%, 7–12 months by 20%, and longer than 12 months by 15% of women. Women who ignored their symptoms were significantly more likely to be diagnosed with advanced disease compared to those who did not (85% vs. 74%; P = 0.002). There was no correlation between specific symptoms and delayed diagnosis (no P value given).

Women with the most symptoms required significantly more time to make the diagnosis (P = 0.001); they were also more likely to be treated for another condition (P = 0.001), were younger (P = 0.001), were less likely to receive a diagnosis at an early stage (P = 0.001), and more likely to perceive that health care provider attitude towards them was a problem (P = 0.001).

The type of health care provider initially seen was a family practitioner in 34% of cases, an obstetrician-gynaecologist in 37%, an internist in 16%, a nurse practitioner in 3%, and other in 10%. The type of insurance did not correlate with delayed diagnosis. The time required by a health care provider to make the diagnosis was reported as less than three months by 55%, but greater than six months by 26%, and greater than one year by 11%. Time required to make the diagnosis was similar for the main three health care provider types (family practitioner, obstetrician-gynaecologist, internist). Significantly more stage I/II tumours were diagnosed by obstetricians-gynaecologists than by other health care providers (P = 0.009).

Other factors significantly associated (by univariate analysis) with delay in diagnosis were omission of pelvic examination at first visit (P = 0.016), and not initially organising an ultrasound, computed tomography, or CA125 (P = 0.001).

Multivariate analysis was performed with linear regression to evaluate factors that were associated with the number of months to make a diagnosis. Only 20% of the delay was explained by the factors evaluated by the authors. The factors most significantly associated with delay in diagnosis were amount of time patients had symptoms (P = 0.001), the number of health care providers seen (P = 0.06), symptoms being ignored (P = 0.07), and initial incorrect diagnosis (P = 0.08).

The findings of this study should be interpreted with caution. Responses from women to the questionnaire were not verified, as the authors had no access to medical records, or indeed population cancer registries. The study identified associations, and these must not be assumed to indicate causation.

(Smith and Anderson, 1995)(245)

The study aimed to evaluate characteristics of symptoms, their perceived cause, and delay in seeking a diagnosis associated with stage, grade, and histologic features of disease at diagnosis among incident cancers of the ovary. Women included in the study (N= 82) had a histologically confirmed primary of the ovary within the last three months, were all white, and identified from a US population-based cancer registry.

The authors interviewed all women to ascertain reasons for, and extent of, delay in diagnosis. Women were also asked to provide information on individual socio-demographic factors (income, education, occupation, age, and marital status). Disease related information was extracted from the cancer registry and medical records.

56 (68.3%) women noticed symptoms before diagnosis. Women who were 40 years of age or older were significantly (P < 0.05) more likely to report having symptoms that convinced them to see a physician for diagnosis. Overall, fewer than 10% thought that they had cancer, and most women believed that their problems were due to either menstrual conditions or to unknown causes. There was a trend (P < 0.10) for earlier-stage disease and the perception that symptoms were due to cancer. There was no association between perceptions of the causes of symptoms and socio-demographic factors.

The median number of weeks delay in seeking medical attention was four. More than half (52.5%) saw a physician in one month or less, but about one fourth (22.5%) waited three months or longer. Nonetheless, there was no association between stage and delay, regardless of symptoms, nor was there an association between delay and perceived cause or seriousness of symptoms. The most frequent reasons given for delay were: “fear” (22.7%), repeat appearance of a previous benign condition (22.7%), and symptoms interpreted as “not serious” (18.2%). Fear showed a weak association with greater delay (P < 0.10).

(Wikborn et al, 1996)(246)

The study attempted to investigate the process from first recognition of symptoms to final diagnosis at operation in patients with epithelial ovarian cancer. The authors studied the medical records of 160 women diagnosed with epithelial ovarian cancer at a Swedish hospital between 1981 and 1986 in order to obtain information on patient- and doctor- related delay. Data were collected on age, symptomatology, diagnostic process time span, tumour histopathological class, and tumour stage.

The patients’ mean age was 62.4 years with a range of 25–85 years. The mean symptom duration before consulting a doctor was 12 weeks for serous cancers (SD 16.1) compared with seven weeks for the others (SD 11.2)(P < 0.05). Of all the women, 56% were diagnosed within four weeks; no significant differences were found between different histopathological groups. In the eight weeks following recorded first consultation, as many as 30% of women had not been correctly diagnosed.

Cancer of the uterus
(Crawford et al, 2002)(262)

The authors investigated links between delays in treatment and survival by collecting data from the case notes of all women resident in Scotland who were diagnosed in the two year period 1996–1997 as having endometrial carcinoma. 703 cases that involved operative treatment were analysed (out of a total of 781 cases identified).

The study exclusively examined secondary care provider delay, as the authors looked at the time interval from referral to definitive operation and not the delay.

The median interval from referral to definitive operation was 62 days (90th centile 150 days), with large variations between health board areas. Delay and survival were inversely related: women with the shortest delay had more advanced disease and survival was least likely for these patients (P values not provided by the authors).

(Aziz et al, 1993)(263)

The purpose of the study was to compare the prognostic factors - including grade, stage, depth of myometrial invasion, status of lymph nodes, and peritoneal cytology - and survival of black and white patients with endometrial carcinoma. The authors undertook a retrospective study of 290 patients with endometrial carcinoma who were treated at two US hospitals between 1975 and 1990.

136 (47.2%) patients were black, 135 (46.9%) were white, 15 (5.2%) were Hispanics, and the racial origin of four patients was not known. The mean age was 63 years in the range of 28–95 years (standard deviation 10.6). Black and white patients had similar treatments.

Black patients had more advanced stage disease than white patients (stage I, 45.9% vs. 54.1%; stage II, 48.4% vs. 51.6%; stage III, 88.9% vs. 11.1%; stage IV, 100% vs. 0%; P = 0.034). Black patients also had more advanced grade disease (P = 0.008), myometrial invasive disease (P = 0.038), and lymph node involvement (P = 0.01).

The corrected ten year survival for white patients was 72% compared to 40% for the black patients (P = 0.0003). The overall survival for blacks vs. whites less than 60 years of age (P = 0.002), and for blacks vs. whites more than 60 years of age (P = 0.003) was significantly lower in black patients as compared to white patients. Survival comparison stratified by both age and race indicated that black patients under 60 years of age had the worst survival rate. Survival comparisons, when stratified by race and each prognostic group, showed statistically significant overall survival differences in favour of white patients.

Vulval cancer
(Jones and Joura, 1999)(264)

The authors examined the preceding clinical events in 102 women presenting to a New Zealand tertiary care gynaecologic oncology unit with squamous cell carcinoma of the vulva between the years 1989 and 1996. History, clinical findings, previous physician contact, investigations and treatment were analysed.

The age range was 36–94 years. Vulval symptoms were present for more than six months in 88% of patients and for more than five years in 28%. No statistical differences were noted in the duration of symptoms when the patients were grouped according to age. A history of intermittent or chronic vulval irritation was elicited in 94% of patients.

In 31% of cases the women had had three or more medical consultations on account of vulval symptoms more than six months before the diagnosis of invasive cancer. The length of the history and the number of consultations were independent of age.

A history of the prior application of topical oestrogen or corticosteroid to the vulva was elicited in 27% of women. Twenty-five percent of patients had previously had a diagnostic biopsy. Seventeen women (68%) with a history of preceding biopsy presented with stage I disease as compared with 26 (34%) in the cohort without a preceding biopsy (P < 0.01).

Diagnostic Difficulties

No articles reporting studies of the difficulties encountered by primary care professionals in identifying patients to be referred for suspected gynaecological cancer were identified.