Delays in diagnosis of young females with symptomatic cervical cancer in England: an interview-based study

Background Diagnosis may be delayed in young females with cervical cancer because of a failure to recognise symptoms. Aim To examine the extent and determinants of delays in diagnosis of young females with symptomatic cervical cancer. Design and setting A national descriptive study of time from symptoms to diagnosis of cervical cancer and risk factors for delay in diagnosis at all hospitals diagnosing cervical cancer in England. Method One-hundred and twenty-eight patients <30 years with a recent diagnosis of cervical cancer were interviewed. Patient delay was defined as ≥3 months from symptom onset to first presentation and provider delay as ≥ 3 months from first presentation to diagnosis. Results Forty (31%) patients had presented symptomatically: 11 (28%) delayed presentation. Patient delay was more common in patients <25 than patients aged 25–29 (40% versus 15%, P = 0.16). Vaginal discharge was more common among patients who delayed presentation than those who did not; many reported not recognising this as a possible cancer symptom. Provider delay was reported by 24/40 (60%); in some no report was found in primary care records of a visual inspection of the cervix and some did not re-attend after the first presentation for several months. Gynaecological symptoms were common (84%) among patients who presented via screening. Conclusions Young females with cervical cancer frequently delay presentation, and not recognising symptoms as serious may increase the risk of delay. Delay in diagnosis after first presentation is also common. There is some evidence that UK guidelines for managing young females with abnormal bleeding are not being followed.


Abstract Background
Diagnosis may be delayed in young females with cervical cancer because of a failure to recognise symptoms.

Aim
To examine the extent and determinants of delays in diagnosis of young females with symptomatic cervical cancer.

Design and setting
A national descriptive study of time from symptoms to diagnosis of cervical cancer and risk factors for delay in diagnosis at all hospitals diagnosing cervical cancer in England.

Method
One-hundred and twenty-eight patients <30 years with a recent diagnosis of cervical cancer were interviewed. Patient delay was defined as ≥3 months from symptom onset to first presentation and provider delay as ≥3 months from first presentation to diagnosis.

Results
Forty (31%) patients had presented symptomatically: 11 (28%) delayed presentation. Patient delay was more common in patients <25 than patients aged 25-29 (40% versus 15%, P = 0.16). Vaginal discharge was more common among patients who delayed presentation than those who did not; many reported not recognising this as a possible cancer symptom. Provider delay was reported by 24/40 (60%); in some no report was found in primary care records of a visual inspection of the cervix and some did not re-attend after the first presentation for several months. Gynaecological symptoms were common (84%) among patients who presented via screening.
oncology clinical teams in all NHS hospitals diagnosing cervical cancer in England to notify the study investigators of eligible cases. The study was also publicised using cancer charity websites, so that patients could volunteer to take part directly. Diagnosis was confirmed for all patients by the pathology laboratory or clinical team. Each patient's clinical team was asked to assess eligibility and invite her to take part. For patients providing specific consent, primary care records were requested for the 2 years before diagnosis, plus results of cervical screening tests from the National Cervical Screening Database.

Measures
Data were collected using a semi-structured interview-based measure (the interview schedule is available from the authors on request). 15 This approach has advantages over collecting data from medical records or using self-completion questionnaires, because it allows the interviewer to clarify and check the sequence of events, and to probe for information that patients might not consider relevant. 16 The measure collects data about dates of symptoms and healthcare attendances from the first possible cancer symptom through to diagnosis and risk factors for delay in presentation.

Data collection
The same trained interviewer conducted all the interviews either face-to-face or on the telephone (according to the patients' preference) within 2-6 months of diagnosis. Interviews were audiorecorded for quality assurance and data checking.

Analysis
Characteristics (age, stage [using the International Federation of Gynecology and Obstetrics {FIGO} classification], histology, and geographical region) of patients who were interviewed were compared with those who were not, and of notified cases and patients aged 18-29 years with cervical cancer recorded in the National Cancer Registration Dataset in 2010 (Appendix 1).
Patients were categorised according to whether they had been diagnosed via: 1. Symptomatic presentation: if they had presented to a health professional with one or more symptoms from a checklist of possible symptoms of cervical cancer 5-9,17-19 (Box 1) and that presentation led to diagnosis (including patients diagnosed via abnormal cytology if the test was done for diagnostic purposes).

Abnormality detected on NHS Cervical
Screening Programme: if they were diagnosed as a result of a finding during a routine cervical screening test after an invitation on the NHS Cervical Screening Programme.
3. Incidental finding: if they were diagnosed during management of an unrelated condition.
Date of diagnosis was defined as the date the patient said she was told she had cervical cancer. 20 For patients diagnosed via symptomatic presentation, the 'trigger' symptom was identified, defined as the earliest symptom from the checklist in Box 1 that led the patient to present to a health professional and led to diagnosis. For all patients, the symptom that she believed to be the first symptom of her cancer was identified at the time of interview (the 'first attributed' symptom) and the earliest symptom from the checklist that the patient reported, but did not attribute to cervical cancer at the time of the interview (the 'initial' symptom).
Patient duration of symptoms from onset to first attendance with the trigger symptom was calculated, and provider duration of symptoms from first attendance with trigger symptom to diagnosis, rounding to the nearest month. Delayed presentation was defined as patient duration of symptoms of ≥3 months (a cut-off point that is convention in studies of this nature [21][22][23], and provider delay as provider duration of symptoms of ≥3 months. Primary care delay was defined as referral ≥6 weeks after first attendance and secondary care delay as diagnosis ≥6 weeks after referral. For patients who presented via symptomatic presentation and reported provider delay, the general practice records were examined (including free text). Whenever possible, the period in the British Journal of General Practice, October 2014 e603

How this fits in
There have been reports that young females with cervical cancer experience delays in diagnosis because of a failure to recognise symptoms. This study has found that a high proportion of patients aged <30 years with symptomatic cervical cancer delay presentation. Delay in referral after presentation is also common. There is some evidence that the UK guidelines for managing young females with abnormal bleeding are not being followed.

Box 1. Cervical cancer symptom checklist
• Bleeding after sex [5][6][7][8][9][17][18][19] • Bleeding between periods 7,9,17,19 • Change in periods 5 • Bleeding during pregnancy • Painful sex • Persistent vaginal discharge 5,6,9,18,19 • Abdominal or pelvic pain or discomfort 7,9,19 • Unusual fatigue primary care records which matched the period described at interview, in terms of symptoms and date of attendance (within 3 months of the date reported by the patient), was identified. Records were examined for documented evidence of visualisation of the cervix, tests for genital infection, or cervical cytology. Dates of referral were extracted and used to calculate time from first attendance to referral and from referral to diagnosis.
For patients aged 25-29 years diagnosed via symptomatic presentation, cervical screening history was examined in the national database to find out whether they had been screened and, if so, how long before the presentation, and the result(s).
Odds ratios and 95% confidence intervals (CIs) were calculated to evaluate the association between possible risk factors for delay (Box 2), treated as binary categorical variables, and delay status.
The χ 2 test or Fisher's exact test (when the expected cell frequency was <5) were used to assess differences between delays and symptom characteristics between age groups and pathways to diagnosis.
All statistical analyses were performed using Stata Statistical Software (Release 12). A P-value of <0.05 was considered statistically significant. All statistical tests were two-sided.

RESULTS
One-hundred and sixteen hospitals identified 333 patients (237 from pathology laboratories and 96 from clinical teams) over an average recruitment period of 8 months (range 1-11 months). One additional patient volunteered to take part after seeing the study on a charity website. Within the study period, 286 (86%) patients were invited; 164 agreed, and 128 (38% of eligible patients identified) were interviewed. Mean time from diagnosis to interview was 4.3 months (range 1.0-8.5 months).
Characteristics of patients identified who were interviewed versus not interviewed were comparable, although interviewed patients were younger and more had adenocarcinoma (Appendix 1). Patients identified were broadly representative of patients identified nationally by the cancer registry for age, stage, histology, and geographical region.
Of the 22 patients <25 years, one was diagnosed after a cervical lesion was seen during intrauterine device removal and another via screening (just before her 25th  20 (19%) were diagnosed via symptomatic presentation, and one was diagnosed after a cervical lesion was seen during insertion of an intrauterine device.

Patients diagnosed via symptomatic presentation
Of the 40 patients diagnosed via symptomatic presentation, 11 (28%) reported patient delay, and 24 (60%) reported provider delay ( Table 2). The most common trigger symptom (33/40) was abnormal vaginal bleeding, with 19 patients describing postcoital bleeding (Table 3). Twenty-three patients (56%) reported earlier symptoms that had not prompted help-seeking: 10 attributing these symptoms to cancer at the time of interview (six with abnormal vaginal bleeding); and 13 not attributing them to cancer at the time of interview, most commonly vaginal discharge (n = 6).
Five patients had been diagnosed as a result of abnormal cytology on a diagnostic test after symptomatic presentation. None of these patients had ever attended for cervical screening after a routine invitation.
Sixteen (80%) of the 20 patients aged 25-29 years with symptomatic presentation consented to access of their screening data. Nine (56%) had attended routine screening within the 3.5 years before diagnosis (that is, were interval cancers), and the rest had never been screened. Of the screened patients, only one had had abnormal cytology but findings had been normal on a subsequent colposcopy 16 months before diagnosis.
Risk factors for delayed diagnosis among patients diagnosed via symptomatic presentation. Table 4 shows the frequency of possible risk factors for delays. None of the differences were statistically significant. The largest odds ratio for patient delay, however, was for not knowing what the symptoms of cervical cancer were. For provider delay this was using hormonal or intrauterine contraception.

Explaining provider delays among patients who presented symptomatically
Primary care records were received for 21/24 patients who reported a provider   Percentages are calculated based on the total number of patients diagnosed via symptomatic presentation.
The most common trigger symptom is in bold. delay after symptomatic presentation. For three of these, however, the first attendance had been at a sexual health clinic, meaning there were no records of the relevant consultations. For the remaining 18, a consultation could be identified in the GP notes matching the first attendance reported by the patient (in terms of similar symptoms and date) for 10 (56%) patients. For six of these patients, the delay occurred in primary care. For two of these, there was no record of visualising the cervix. Both had presented with intermenstrual bleeding (one also had postcoital bleeding) and both had tested negative for genital infection. In both of these patient's records, the records documented advice to return if symptoms continued; however, for both, the subsequent attendance was 5-6 months later.
For the other four patients with primary care delay, visualisation of the cervix was recorded: normal in two, cervical polyp for one, and 'cervical bleeding on contact' for one. The notes recorded tests for genital infections in all four patients, of whom three had positive results; one chlamydia and two bacterial vaginosis. For three of these patients, 2-6 months elapsed between first presentation and the next attendance. Advice to re-attend was documented in only one of the patient's GP notes. The remaining patient was referred routinely to gynaecology after 3 months of re-presenting with heavy intermenstrual bleeding.
For the remaining four patients, there was secondary care delay. In three of these, no malignancy was found on the initial biopsy in secondary care. The remaining patient waited 8-10 weeks to be seen in gynaecology, then colposcopy.

Patients who were pregnant
Three patients (8%) diagnosed via symptomatic presentation were pregnant at the time of first presentation. All three had reported bleeding during their   pregnancy, presented straight away, and were managed in secondary care. Two of these patients had a colposcopy while pregnant, but no malignancy was identified on biopsy. All three patients continued to have symptoms throughout pregnancy and after giving birth, and were eventually diagnosed 9-10 months after presenting with their trigger symptom (stages 1b1, 2b, and 3b). The remaining 37 patients who reported symptoms said they had not sought help for their symptoms. Their most common symptoms were vaginal discharge (16/37, 43%), postcoital bleeding (13/37, 35%) and abdominal pain and fatigue (both 12/37, 32%). Median duration of symptoms from first symptom to diagnosis was 17.8 months (IQR 7.7-73.4 months).

DISCUSSION Summary
Among 128 young patients diagnosed with cervical cancer aged 18-29 years, 40 were diagnosed via symptomatic presentation. These patients were more likely to have more advanced cancers than patients who were diagnosed via screening, although very few patients had stage 2 or above. Of the patients eligible for screening, one-fifth were diagnosed via symptomatic presentation.
Most patients presenting symptomatically had abnormal vaginal bleeding, most commonly postcoital bleeding. One-quarter of these delayed presentation for ≥3 months after onset of symptoms: this was more common in those aged <25 years. Over half reported gynaecological symptoms some time before the symptom that caused them to present: this may have been an opportunity for earlier diagnosis. Vaginal discharge was the most common of these.
Provider delay was reported frequently. No conclusive evidence was found about risk factors for provider delay, although not all patients with provider delay had had their cervix visualised in primary care and many had co-existing genital infection. There was also some evidence that some delay in primary care may be because of patients not re-attending after first presentation promptly despite persistent symptoms.

Strengths and limitations
To the authors' knowledge, this study has provided the most detailed information to date on the nature and duration of symptoms for young females with cervical cancer. The methods and reporting are consistent with the Aarhus statement on studies of early cancer diagnosis. 20 In the present study, a high identification rate was achieved and the population was broadly representative of patients diagnosed with cervical cancer in England over a similar period. Another key strength was the high quality of the data collected. By using an interview-based measure, which was developed and tested in young patients with cervical cancer, 15 it was possible to record complicated data systematically.
The possibility of recall error cannot be ruled out. This was minimised using calendar anchoring and collaborative completion of a timeline detailing the events that led to diagnosis. 15 In addition, most patients were interviewed within 6 months of diagnosis.
Identification of risk factors for delays in cervical cancer diagnosis was limited by small study numbers, which is inevitable with rare diseases. A study with sufficient power to find statistically significant risk factors for delay in presentation would have to recruit patients for years.

Comparison with existing literature
The present findings are consistent with research showing that cervical cancer symptom awareness among young British

Funding
This study was funded by NHS Cancer Screening Programmes (JP/pat/L155). Julietta Patnick (Director of Cancer Screening Programmes in England) commented on and approved the final version of the manuscript, but was not involved in the development of methods, data collection, analysis, or interpretation of the data. females is low. 24 An online YouGov survey of 2726 females in the UK found that 20% of 18-24-year-olds said that they would not seek help for bleeding during sex (M Durrant, personal communication, August 2013). A systematic review that examined the risk factors for delayed presentation in cancer found that non-recognition of the severity of symptoms, symptom type (for example, vague), infrequent care-seeking, and fear were the predominant risk factors for delayed presentation for gynaecological cancers. 25

Implications for research and practice
The study confirms the findings of case reports suggesting that patients <25 years often delay presentation with cervical cancer and have more advanced stage. This supports the idea that earlier presentation could improve outcomes in this group.
Although this study was small, it provides preliminary data that could be used to inform interventions to promote earlier presentation in this group by ensuring that they set out the possible symptoms of cervical cancer clearly (framing these in particular for young females of low education, who are at higher risk of cervical cancer in any case), 26 perhaps emphasising that other priorities, embarrassment, and worry about wasting the doctor's time should not deter them from presenting, and that they will be able to see a female doctor if they wish.
The challenge is that gynaecological symptoms in young females are very common, and promoting early presentation for these symptoms could lead to a large increase in attendances in primary care and unnecessary anxiety in females who do not have cancer (or cervical abnormalities). For females aged 15-29 years, at least 1.6% presented to primary care with intermenstrual bleeding, 0.5% with postcoital bleeding, and 1.3% with vaginal discharge in any 1-year period. 11 Although most screen-detected patients reported symptoms, a high proportion of these were longstanding yet most had early stage cancer. This implies that some of their symptoms were unlikely to have been related to their cancer.
The present data suggest that pregnancy at the time of presentation may be a risk factor for provider delay; however, these patients appear to be difficult to diagnose given that even at colposcopy (including biopsy), malignancy was not easily identified.
The data from primary care records were too sparse to be able to draw any firm conclusions about the quality of management of gynaecological symptoms in primary care, although patients may not be having their cervix visualised when they present with abnormal vaginal bleeding, despite the UK guidance on this. 17 Furthermore, patients may delay re-attending after the first presentation, and re-attendance should be strongly advised if their symptoms persist.

Ethical approval
This study was approved by the North West London REC2 Ethics Committee (10/ H0720/65) on 27 September 2010 and by the National Information Governance Board for Health and Social Care in October 2011. All participants gave informed consent before taking part.

Provenance
Freely submitted; externally peer reviewed.