Invasive cervical cancer audit: why cancers developed in a high-risk population with an organised screening programme


Dr A Herbert, Guy’s and St Thomas’ NHS Foundation Trust, Histopathology Department, Second Floor, North Wing, St Thomas’ Hospital, London SE1 7EH, UK. Email


Please cite this paper as: Herbert A, Anshu, Culora G, Dunsmore H, Gupta S, Holdsworth G, Kubba A, McLean E, Sim J, Raju K. Invasive cervical cancer audit: why cancers developed in a high-risk population with an organised screening programme. BJOG 2010;

Objectives  To investigate why invasive cervical cancers developed in a high-risk urban population with an established screening programme and to place cancers in the context of high-grade cervical intraepithelial neoplasia (CIN) and cervical glandular intraepithelial neoplasia (CGIN) diagnosed during the same period of time.

Study design  Observational study of CIN2+ (CGIN, CIN3 and CIN2) and invasive cervical cancer diagnosed at Guy’s and St Thomas’ NHS Foundation Trust in 1999–01, 2002–04 and 2005–07 and audit of screening histories of women with invasive cancer analysed according to route to diagnosis, histological type and International Federation of Obstetrics and Gynecology (FIGO) stage.

Results  There were 133 invasive cancers, 53 CGIN, 1502 CIN3 and 1472 CIN2. Screen-detected cancers in asymptomatic women comprised 48.9% of cancers and were successively more likely to be in younger age groups (P = 0.03); all except one were stage IA or IB1. Screen-detected IA cancers were more likely (< 0.001) to be in women screened within 0.5–5.0 years (80.5%) than screen-detected fully invasive (58.3%) or symptomatic cancers (35.3%). Seventy-one (53.4%) women had been screened within 0.5–5.0 years; 11 had negative cytology within 0.5–3.5 years and two tests within 10 years. The other 60 had negative tests less frequently or had previous abnormal cytology, colposcopy or treatment. Potentially avoidable factors were often multiple, including false-negative cytology, high-grade cytology reported as low-grade and lapses in attendance either for routine or repeat screening, or for colposcopy or treatment.

Conclusion  While often potentially avoidable, cancers in previously screened women tended to be early stage, detected by cytology and rare when compared with high-grade CIN.


The National Health Service Cervical Screening Programme (NHSCSP) requires audit of screening histories of all women diagnosed with invasive cervical cancer1 and such audits are also recommended in the European Guidelines for Quality Assurance in Cervical Cancer Screening.2 Audits require a baseline for comparison. Several recent national audits are available.3–6 However, the UK national audit excluded microinvasive cancers and, as it was based on central screening records, could not include data on whether the cancers were diagnosed as a result of the investigation of symptoms or as the result of investigation of an abnormal cytology test in asymptomatic women.3 The Swedish audit assumed that cancers with cytology within 1 month of diagnosis or no cytology within 6 months were symptomatic while those with cytology tests 1–6 months before diagnosis were screen-detected.6 We have used clinical criteria to decide whether the cancers were screen-detected or symptomatic and have used a previous 12-year audit of women in Southampton as a baseline.7,8 As in that study, we categorised the screening histories as recommended in the first edition of Achievable Standards, Benchmarks for Reporting and Criteria for Reporting Cervical Cytopathology.9 The categories in that report were based on previously reported reasons that cancers may not be prevented in screened populations.10

We allowed the term ‘interval cancers’ to include screen-detected cancers because, unlike colorectal, breast or prostate cancer screening, cervical screening aims to prevent rather than detect invasive cancer. It is therefore important to consider the reasons why cancers identified by the test in asymptomatic women have not been prevented.7,8 Conventional definitions of ‘interval cancers’ exclude screen-detected cancers when the test is taken at the recommended time. We chose the time period of 5 years to define interval cancers; first, because the screening interval has varied historically from place to place and still varies according to age group and second, because the NHSCSP calculates ‘screening coverage’ in terms of women aged 25–64 years who have been screened within 5 years. This provides a useful baseline for calculating incidence in ‘previously screened’ and ‘previously unscreened’ women when data for a defined population are available, as it was in the Southampton audit.8

We have not been able to calculate invasive cancers in proportion to numbers of screened and unscreened women in the female population because the catchment areas of Guy’s and St Thomas’ NHS Foundation Trust (GSTFT) and King’s College Hospital Trust cut through both Lambeth and Southwark, where more than 90% of our populations reside. We are aware that the Thames Cancer Registry had reported cervical cancer mortality and incidence in Lambeth and Southwark to be the highest in any London borough: the age standardised incidence per 100,000 population (European standard) was 10.8 in Lambeth, 11.0 in Southwark and 8.4 in England (2001–05).11 The increased cervical cancer rates correlated with high deprivation indices, cigarette smoking rates, teenage conception rates and HIV prevalence in the same boroughs but not with screening coverage rates.12

Unlike our previous study, we have placed invasive cancers in the context of high-grade cervical intraepithelial neoplasia (CIN) and cervical glandular intraepithelial neoplasia (CGIN) detected during the same period of time because detection and treatment of those conditions forms the mechanism for the effectiveness of cervical screening; and provides the appropriate background in which invasive cancers may be monitored in a local programme.

The Southampton audit showed that cancers in women routinely screened during the 5 years preceding diagnosis (described as interval cancers) were more likely to be early stage, screen-detected cancers in younger women.7,8 One of the aims of the current audit was to investigate those associations in an entirely different population at a later stage of a centrally organised screening programme. We aimed to find out why cancers developed in a high-risk urban population and to look for causes that might be preventable.



The population screened at GSTFT mainly resides in the south London boroughs of Lambeth and Southwark, which had an eligible female population aged 25–64 years of 178 764 at 31 March 2006 with a screening coverage within 5 years of 71.9%. Approximately two-thirds of those women’s cytology tests were reported at GSTFT where the average annual cervical cytology workloads during the audit were relatively constant (Table 1). The local population compared with England as a whole is biased towards younger age bands (Figure 1).

Table 1.   Age band, route to diagnosis and screening history of cancer and cervical intraepithelial neoplasia (CIN) 2+ according to time period of the study
  1. *CIN2, n = 1472; CIN3, n = 1502; and cervical glandular intraepithelial neoplasia, n = 53.

Invasive cancer
Age band (years)
All age groups363859133
% age 20–34 years27.821.137.330.1
Route to diagnosis
All age groups363859133
% screen-detected38.955.350.848.9
Screening history
No cytology record in UK12111841
Screened >5 years before diagnosis651021
Screened within 5 years18223171
All age groups363859133
% screened within 5 years50.057.952.553.4
Total cytology tests 1999–2007102 963108 741104 829 
Total referrals (% of tests)   4115 (4.0)3811 (3.5)3658 (3.5) 
All age groups887101511253027
% age 20–34 years77.272.977.976.0
Figure 1.

 Female population by age band (KC53 data); Lambeth and Southwark compared with England.

Case selection

Women diagnosed with invasive cancer were found from department computer records including a fail-safe database of women referred for colposcopy, from gynaecology oncology and colposcopy multidisciplinary meetings. Their details were reviewed, presented and discussed at those meetings. Since 2007 a letter summarising the screening history is put in the women’s notes and is available for her to see. Data for women diagnosed since 2000 were transmitted electronically to the London Regional Quality Assurance Reference Centre (QARC) access database as part of the NHSCSP regional and national audit. The London database allows an anonymous spreadsheet to be exported to Excel, which formed the basis of the audit. At the time of diagnosis, 90.4% of the women with cancer were resident in Lambeth or Southwark and all were diagnosed and managed at GSTFT. Tertiary referrals from elsewhere in southeast London and beyond were excluded.

Diagnoses of CIN2, CIN3 and CGIN were taken from departmental fail-safe records. Since April 1998 records of all women recommended for colposcopy have been transferred from the laboratory computer system to an Access database, which records the cytology diagnosis and outcome. Biopsy results obtained from other hospitals are included if women are treated elsewhere (the latter account for <5% of high-grade CIN biopsy results). The patient base for the CIN records is not exactly the same as for the cancers because the women were taken from those whose screening tests were reported at GSTFT, which would have been several months before the biopsy. Although the postcodes of the women with cancer were known, those of women with CIN were not. The only change in catchment area took place in 2001, when one general practice (about 1000 women) transferred back to sending their cytology tests to GSTFT. Liquid-based cytology was introduced in April 2004.

Route to diagnosis

Symptomatic cancers were defined as those diagnosed as a result of referral for the investigation of symptoms whether or not a cytology test was taken during that process. Screen-detected cancers were defined as those diagnosed as a result of investigation of an abnormal cytology test in women who had not been referred with or complained of symptoms. Information on route to diagnosis was collected from pathology request cards, multidisciplinary meetings or by extracting the clinical notes and referral letters. Clinical notes and referral letters were reviewed by A.H. and/or E.M. working with K.S.R. or J.S.

Screening history

Screening history was categorised after Open Exeter records had been searched as well as laboratory and hospital records. Biomedical scientists and consultants reviewed negative, inadequate and low-grade cytology slides reported at GSTFT in the 5 years before diagnosis of cancer. Positive and negative control slides were included with the study slides. Although cytology slides reported within 10 years of diagnosis are now reviewed for the NHSCSP cancer audit, those more than 5 years before diagnosis were not reviewed for this study because slides would not have been available for women in the first few years of the study. Details of colposcopy examinations and treatment during the preceding 5 years were reviewed at colposcopy meetings during the latter years of the study. Some women with earlier diagnoses were managed in general gynaecology clinics rather than the colposcopy clinic and records of those visits were not always available.


The chi-square test was used as a test of inference to compare difference between proportions. The chi-square for trend was used to test significance of trend of proportions across study years. The Statcalc calculator in the epi info 6 software (Epidemiology Program Office, Center for Disease Control and Prevention, Atlanta, Georgia, USA in collaboration with Global Program on AIDS, World Health Organization, Geneva, Switzerland) was used for calculation of these tests.


Invasive cervical cancers 1999–01, 2002–04 and 2005–07

There were 133 cancers diagnosed between 1999 and 2007 (Table 1). Of these, 113 (85.0%) were squamous cell carcinomas, 14 (10.5%) were adenocarcinomas, four (3.0%) were adenosquamous carcinomas, one (0.7%) was a small-cell neuroendocrine carcinoma and one (0.7%) was an undifferentiated (lymphoepithelioma-like) carcinoma. One hundred and eighteen (88.7%) were in women aged 20–64 years; 40 (30.1%) were aged 20–34 years (four were aged 20–24 years). The apparent increase in cancers in women aged 20–34 years in the third period of the study (Table 1) was not significant (P = 0.25).

CIN2+ and CGIN in 1999–01, 2002–04 and 2005–07

Between 1999 and 2007 there were 1472 biopsy diagnoses of CIN2, 1502 diagnoses of CIN3 and 53 diagnoses of CGIN (Table 1). Of these, 76.0% were in women aged 20–34 years of age among whom 320 instances of CIN2, 224 of CIN3 and six of CGIN were in women aged 20–24 years (18 women under 20 years of age have been excluded from the study). The age range in 5-year age bands of women with CGIN and CIN2+ is shown alongside that of invasive cancer on Figure 2. The peak age band for detection of CIN2 and CIN3 was 25–29 years.

Figure 2.

 Cancer, cervical glandular intraepithelial neoplasia (CGIN), and cervical intraepithelial neoplasia 3 (CIN3) and and 2 (CIN2) according to age band; Guy’s and St Thomas’ NHS Foundation Trust 1999–2007.

Route to diagnosis of invasive cancers

Screen-detected cancers were diagnosed in 65 asymptomatic women on investigation of cytology reported as at least high-grade dyskaryosis in 87.7% of women. Sixty-four (98.5%) were diagnosed at stage IA or IB and all except three were IA1 or IB1 (Table 2). There were two stage IA, not otherwise specified adenocarcinomas and one woman with IIIB cancer was symptom-free when she attended for her first screening test.

Table 2.   Cytology results for screen-detected and symptomatic cancers
 Screen-detectedSymptomatic, cytologySymptomatic, no cytology
Cytology report
Severe, ? invasive1119 
? Glandular neoplasia40 
Severe dyskaryosis283 
Mild dyskaryosis51 
Borderline, glandular/?high-grade32 
Borderline, not otherwise specified00 
% high-grade dyskaryosis/cancer87.777.4 
FIGO stage of cancer
IB, not otherwise specified011
All stages of cancer653137
% stage IB1 or less98.541.916.2

Of 68 symptomatic cancers, 19 (27.9%) were diagnosed at IA or IB1, which was significantly less than for screen-detected cancers (P < 0.001). Cytology tests were carried out during the investigation of symptoms in 31 (45.8%); 77.4% of those tests were reported as at least high-grade dyskaryosis while three (9.7%) were negative (Table 3). Cytology may have prompted diagnosis in some of these symptomatic women if clinical cancer was not seen or recognised; 13 (41.9%) of 31 with cytology were stage IA or IB1, which was significantly less (P < 0.001) than for screen-detected cancers (98.5%) and more (P = 0.019) than for those without cytology tests (16.2%) (Tables 2 and 3).

Table 3.   Stage of cancer and histological type according to route to diagnosis
 Screen-detectedSymptomaticAll cancers% Screen-detected
  1. (n) = stage IA.

FIGO stage
All stages of cancer656813348.9
% stage IB1 or less98.527.962.4 
Histological type
Squamous cell carcinoma58 (39)55 (5)11351.3
Adenocarcinoma7 (2)71450.0
All types of cancer65 (41)68 (5)11348.9

Half (50.0%) of adenocarcinomas and 51.3% of squamous cell carcinomas were screen-detected (Table 2). Ten stage IB1 and IA carcinomas were treated by trachelectomy of which seven were screen-detected, two had severe dyskaryosis on symptomatic smears and one had an inadequate smear; eight were in women aged 20–34 years and three were adenocarcinomas. Excluding women over 65 years, the likelihood of cancer being screen-detected was greater with decreasing age (P = 0.03; Table 4); 15/18 (83.3%) cancers in women aged 20–29 years were screen-detected (Figure 2).

Table 4.   Presentation/stage of cancer and screening history according to age band
 Age band (years)
20–3435–4950–6465 +All ages
Screen-detected IA21164041
Screen-detected IB+6135024
All cancers40552315133
% screen-detected67.552.739.10.048.9
Screening history
No cytology record71491141
Cytology more than 5 years before693321
Cytology within 0.5–5.0 years273211171
All cancers40552315133
% screened within 5 years67.558.247.86.753.4

Screening histories of women with cancer

Excluding tests taken within 6 months of diagnosis, which were either abnormal tests resulting in referral or taken from women with symptoms, 41 women (30.8% of all cancers) had no record of previous cytology on the Open Exeter system. These cancers comprise category 1 (Table 5). One or more cytology tests were recorded more than 5 years before diagnosis (category 2) in 21 women (15.8% of all cancers). Routine cytology tests had been recorded during the 5 years before diagnosis (categories 3–6) in 71 women (53.4% of all cancers). These 71 diagnoses are defined as interval cancers. Screen-detected microinvasive cancers were significantly more likely (P < 0.001) to be interval cancers (80.5%) than screen-detected fully invasive cancers (58.3%) or symptomatic cancers (35.3%) (Table 5). When women over 65 years were excluded, there was no significant association between interval cancers and younger age bands (P = 0.12; Table 4).

Table 5.   Screening history according to presentation/stage
Screening historyPresentation/Stage
Total (% SD)
133 (48.9)
  1. SD, screen-detected; sym, symptomatic.

Category 1 (no cytology record in UK)41 (24.4)3731
No record of previous cytology test353527
Previously screened outside UK6024
Category 2 (cytology more than 5 years before)21 (38.1)5313
Negative cytology185211
Abnormal cytology and/or CIN treatment3012
Categories 3–6 interval cancers (screened within 5 years)71 (66.2)331424
% screened within 5 years53.480.558.335.5
Category 3 (negative cytology only)27 (51.9)10413
Latest test within 0.5–3.5 years17 [6] 7 [2] 2 [1]  8 [3]
Latest test within 3.5–5.0 years10 [2]3 2 [1]  5 [1]
[n] = one negative test within 10 years    
Category 4 (early repeat(s) advised) 8 (62.5)503
Repeat cytology as recommended3201
Delayed or no follow-up cytology5302
Category 5 (previous referral advised)22 (86.4)1273
Delay diagnosis as the only factor8602
Other factors (negative and/or repeats)14671
Category 6 (previous CIN treatment)14 (64.3)635
Negative follow-up as recommended1001
Insufficient follow-up ± delays6213
Incomplete excision, CIN persists7421

Categories of screening history

Category 1: no cytology record (n = 41)

There were 41 women with no cytology record on Open Exeter of whom two were below (aged 20–24 years at diagnosis) and eight were above (aged 70 years or over) the age groups eligible for routine screening in the 5 years before diagnosis. Six women (aged between 40 and 55 years) were known to have had previous cytology tests outside the UK (Australia, New Zealand, Pakistan and Lithuania).

Category 2: cytology more than 5 years before diagnosis (n = 21)

Negative cytology tests were recorded in 18 of 21 women whose latest test was more than 5 years before diagnosis. One woman had a repeat recommended for mild dyskaryosis and no follow-up cytology. Two had previous treatment for CIN; one had no follow-up cytology and one (aged 70 years at diagnosis) had two negative follow-up smears after a cone biopsy for CIN3 at age 63 years.

Category 3: negative cytology (no abnormal tests) within 5 years of diagnosis (n = 27)

Negative cytology had been reported within 5 years of diagnosis in 27 women (20.3% of all cancers) of whom 11 had been screened within 3.5 years and had two negative tests within 10 years (Table 5). Four women had only one previous test (two were aged 23 and 27 years at time of diagnosis) and four had one test within 5 years and an interval of at least 5 years since previous tests. Of the 27 women, 14 (51.7%) had screen-detected cancers revealed by a subsequent abnormal cytology test and 10 (37.0%) were aged 20–34 years. Negative cytology was reported within 5 years in addition to abnormal cytology in 21 women in categories 4–6 and will be discussed in the section below on slide review.

Category 4 (n = 8): repeats advised for low-grade or persistent inadequate cytology

Of eight women in this category (5.9% of all cancers) three had been followed up with repeat cytology as recommended (one followed two consecutive inadequate smears) while five (of which four had been reported elsewhere) either had delayed or no follow-up cytology (Table 5). Five of these eight women had screen-detected microinvasive cancers diagnosed within 6 months of subsequent abnormal cytology tests. Low-grade or inadequate cytology was reported in addition to tests recommending colposcopy in 13 women in categories 5 and 6, which will be discussed in the section below on slide review.

Category 5 (n = 22): colposcopy advised more than 6 months before diagnosis (no treatment)

Colposcopy was recommended more than 6 months before diagnosis in 22 women (16.3% of all cancers) who did not have subsequent CIN treatment. The test recommending colposcopy was the first and only test in eight of 22 while 14 also had low-grade or negative cytology. Of these 22 women, 19 (86.4%) had screen-detected cancers (Table 5).

In seven of these 22 women there was an intervening colposcopic biopsy for CIN before the excision biopsy on which cancer was found so that 6–8 months (and in one woman 10 months) elapsed between cytology test and cancer diagnosis, with no record of nonattendance. Six of 22 did not attend colposcopy appointments (one of these was pregnant). Four of 22 had previous colposcopy without treatment; one of these refused treatment over a period of 7 years, one was pregnant and treatment was deferred, one had low-grade changes at initial colposcopy and one at first had low-grade changes and a subsequent negative endometrial biopsy for atypical glandular cells. Of the remaining five women, one was delayed because referral was thought to be nonurgent for low-grade cytology, one was not referred when recommended, in one woman there was delay in reporting as well as referral and the woman had previously defaulted from cytology follow-up and in two the reason for delay was unknown.

Category 6 (n = 14): previous treatment of CIN

Eleven of these 14 women had previous treatment of CIN3, which was recorded as incompletely excised in six. One had CIN2 on a large loop excision of the transformation zone (LLETZ) for moderate dyskaryosis. Two women investigated for severe dyskaryosis had no more than CIN1 on excision biopsy. Seven of the 14 women had persistent high-grade CIN after treatment and were managed appropriately, invasion being diagnosed on a subsequent excision. Six of the 14 had persistent high-grade cytology or biopsies after their initial treatment; two of these had subsequent LLETZ for CIN1 or less, two women refused hysterectomy and in three of the six women there were delays before initial treatment or subsequent cancer diagnosis when the women changed address or failed to attend appointments. One of the 14 women had six negative follow-up cytology tests after a cone biopsy more than 20 years before developing symptomatic cancer.

Including two who had not been screened within 5 years (described in category 2 above), 16 women had CIN treatment at some time before diagnosis of cancer. Of these 16, five were aged 20–34 years, nine were aged 35–39 years and two were aged 50–64 years when the initial treatment was carried out. The percentage aged 20–34 years (31.3%) compares with 76.0% of women aged 20–34 years in routine colposcopy referrals (P < 0.001).

Slide reviews of negative, inadequate and low-grade cytology

In addition to the 27 women with previous negative cytology and no abnormal tests (category 3), there were 21 women with negative cytology within 5 years who also had abnormal cytology (categories, 4, 5 or 6). Of these 48 women, 20 had tests reported elsewhere or had tests recorded on Open Exeter as ‘unknown source’. Of slides reported at GSTFT, 33 were reviewed for the study; 20 (60.6%) were confirmed as negative and seven (21.2%) were reassessed as high-grade dyskaryosis (Table 6). All except one of the slides reviewed as high-grade dyskaryosis contained occasional clusters of dyskaryotic cells or hyperchromatic crowded groups of cells.

Table 6.   Slide review
 Original cytology
NegativeInadequateLow-grade (5 borderline; 4 mild)
Total slides reviewed3349
Review diagnosis
Borderline, not otherwise specified001
Borderline, koilocytosis100
Borderline chages in glandular cells102
Borderline, high-grade dyskaryosis not excluded202
Mild dyskaryosis001
Moderate or severe dyskaryosis623
? Glandular neoplasia100
% original result confirmed60.625.066.7
% abnormal on review33.350.0100.0
% high-grade on review21.250.033.3
% borderline, high-grade dyskaryosis not excluded or borderline changes in glandular cells9.10.044.4

In addition to eight women with early repeats recommended and no recommendation for colposcopy (category 4), 13 also had recommendations for colposcopy 6 months or more before diagnosis or had previous treatment for CIN (categories 5 and 6). Among these 21 women, 11 were either reported more than 5 years before diagnosis (having had later negative or high-grade cytology) or at other laboratories. There were 13 slides with reports at GSTFT recommending early repeat available for review of which five were reassessed as high-grade dyskaryosis. Cytology confirmed as low-grade was either borderline, high-grade not excluded or in glandular cells in four of six women (Table 6).


This audit of cancers diagnosed in a high-risk urban population12 with an established screening programme but low coverage compared with England (71.9% compared with 79.2% in 2006) showed the same strong association between cancers in women who had previously been screened (described as interval cancers) and screen-detected cancers as was reported in a previous study of a provincial population in Southampton during the years in which organised screening was introduced.7,8 In that study, there had been significant trends towards increasing proportions of both screen-detected and interval cancers as screening had become established and as incidence of cervical cancer had fallen in line with England. There have been no significant trends during the 9-year period of the present study, during which incidence nationally and locally has been relatively constant. Screen-detected and interval cancers represented 48.9% and 53.4% of cancers, respectively, which is similar to 38.6% and 48.6% in the final 3-year period in Southampton (1994–96). We did not show the same strong association between interval cancers and younger women as was seen in Southampton.8 At the time of the Southampton study, as in England, there had been a marked increase in cervical cancer incidence and mortality in women aged 20–34 years.13 Interval cancers in the early 1990s, when the NHSCSP first became established, probably included women in that age group with examples of ‘false-negative’ cytology that would be less likely to be seen today now that quality control is well established.8 Nevertheless, in the present study 67.5% of cancers in women aged 20–34 years had been screened within 5 years in a population with 71.9% screening coverage, suggesting that the risk of cancer in screened and unscreened women may be similar, at least among young women, which supports the results of national audits.3

In our view, the effectiveness of screening can only partly be assessed by the relative risk of cancer in screened and unscreened women and our audits in London and Southampton focus on the importance of screen-detected cancers, which are not necessarily stage IA. Compared with symptomatic cancers, the International Federation of Obstetrics and Gynecology (FIGO) stage of screen-detected fully invasive cancers was significantly lower; all but one in the present audit were stage IB1. Ten cancers were small enough to allow trachelectomy, which preserves fertility in most women.14

Screen-detected cancers represent a small proportion of abnormalities detected by screening and the graph shown in Figure 2 is not exceptional. A similar age distribution of 711 instances of CIN3/CGIN and 70 invasive cancers was reported in Southampton in 1991–9315 and reflects the relationship between carcinoma in situ (which includes CIN3 since 1984) and invasive carcinoma seen in England.16 In 2005–07 there were 59 355 diagnoses of carcinoma in situ and 6844 of invasive carcinoma of the uterine cervix of which 67.0% and 23.9% respectively were in women aged 20–34 years compared with 74.5% and 30.1% in our relatively young population.16 Although treatment of early-stage screen-detected cancers is effective in terms of reducing mortality, we would prefer to prevent them. Most IB1 cancers require radical treatment and IA1 cancers usually arise in widespread CIN3, which is more likely to require knife cone biopsy. These wider biopsies are more likely than LLETZ to cause severe adverse outcomes in pregnancy.17,18

The apparent upward trend in cancers in women aged 20–34 years could not be analysed in the absence of a population base to our study but data for England show that incidence has increased in women aged 25–29 and 30–34 years between the second and third periods of our audit.16

We aimed to investigate reasons why cancers were not prevented and to look for causes that might be avoided. Despite regular invitations to all women aged 25–64 years (aged 20–64 years until 2004) 41 women had no record of previous cytology, 21 had last been screened more than 5 years before diagnosis and ten had been screened within 5 years but were overdue for 3-yearly recall—a longer interval that has been shown to carry a greater risk, especially in younger women.3,19 Hence, 72 (54.1%) women had not been screened 3-yearly. Our study is similar to a recent audit in New Zealand, which showed that 50% of women with cancer had not been screened within 3 years whereas 80% had been screened less frequently than recommended.4 The message about the benefits of regular screening should be strengthened and appropriately targeted for maximum impact.

The largest category (27 of 71; 38.0%) of interval cancers comprised women who had negative cytology within 5 years, which was also the case in Southampton and a recent national audit in Sweden.6,8 That category represented a slightly higher proportion of interval cancers (49.1%) in the Southampton study and in both studies half the cancers were symptomatic, making it the most clinically significant category of interval cancers.8 Almost half the women with cancer in our series had negative cytology at some time in their screening history (including those with abnormal cytology as well), which is consistent with the view that the sensitivity of cytology remains a challenge. We found abnormalities in 33.3% of slides originally reported as negative, which is similar to other studies20,21 although the proportion with high-grade dyskaryosis was smaller than in Southampton.8 Most abnormalities found on review in the present study were occasional clusters of cells or hyperchromatic cell groups, which were in the categories known to be ‘at risk’ for being false negative.22–24 It is clearly important to avoid false-negative cytology. We were encouraged by recent experience using the ThinPrep imager as a quality control tool, which supported evidence that it is more sensitive than routine screening.25 During a 9-month trial period we detected significantly more high-grade abnormalities using the imager compared with rapid review;26 and found that the abnormalities localised by the imager tended to be hyperchromatic crowded groups of cells, which are easy to overlook on routine screening. It is also likely that primary human papillomavirus screening would have detected most, but not all, of these lesions.27

More than half of the interval cancers in our series had other factors in the histories, often in addition to negative cytology. Low-grade cytology had rarely been managed as recommended and none of the cancers were in women with low-grade cytology followed by two or three negative tests. Abnormalities were high-grade on review of a third of the slides, which was similar to a recent Norwegian audit.5 Our slide reviews showed borderline changes, high-grade not excluded or in glandular cells [equivalent to atypical squamous cells of undetermined significance (ASC-H) and atypical glandular cells (AGC) respectively in the Bethesda system] in seven cases; both categories are recognised as special types warranting colposcopy rather than repeat cytology,28 justifying the British Society for Clinical Cytology proposal for the NHSCSP to formalise these categories.29 Poor compliance for follow-up of low-grade cytology was a factor in this audit and is an indication for better counselling and better understanding of the risks of abnormal cytology.

As with the Southampton study, we found specific causes of delay in referral among women with cancers, which were usually diagnosed after the same or a subsequent cytology test; 86.4% were screen-detected. Several causes of delay have been remedied by fail-safe systems in colposcopy and the laboratory, which were not in place in the early years of the study. Delays between initial biopsy and LLETZ no longer occur; and direct referral for colposcopy as well as improved (and mandatory) cytology turnaround times have virtually eliminated some of these problems.

We found that cancers occurring after treatment of CIN were more likely to affect women whose initial treatment was later in life (aged 35 years or over), which was similar to the Southampton study8,30 and is consistent with evidence that recurrence of CIN after treatment is more frequent in women aged 50 years or over.31 CIN3 spreads laterally before invasion develops32 and wider lesions are more difficult to excise completely.

All in all, our audit shows the importance of regular screening, accurate cytology, compliance with recommendations for follow-up, prompt investigation of high-grade or persistent low-grade cytology and complete excision of high-grade CIN. It justifies quality control of all aspects of screening and shows that ‘screen-failures’ are a rarity compared with correctly identified and treated high-grade CIN; and that screening young women detects early cancers and high-grade CIN when they are more amenable to fertility-preserving treatment.

Disclosure of interest

None of the authors have interests that conflict with the study.

Contribution to authorship

AH conceived and led the audit during a sabbatical while employed as lead cytopathologist at Guy's & St Thomas' NHS Foundation Trust (GSTFT). Anshu, while a Commonwealth Fellow at GSTFT helped to collect the data and prepare the spreadsheets. GC reviewed the histopathology when necessary and acted as lead histopathologist in the multidisciplinary team. HD collected colposcopy data and audited cases attending colposcopy. SG carried out the statistical analysis. GH is screening commissioner for Lambeth and Southwark and provided information about follow-up of women and screening coverage. AAK is the lead colposcopist, and led the colposcopy multidisciplinary meeting and audited cases attending colposcopy. EMcL helped to collect the women’s data from the laboratory computer records, reviewed the cytology with AH and reviewed medical records with JS and KSR. JS is the consultant nurse specialist for the multidisciplinary team; she reviewed medical records and provided clinical information about route to diagnosis and stage. KSR is the lead gynaecological oncologist, reviewed medical records and provided clinical information about stage and treatment of women. All authors have contributed to editing the final manuscript.

Details of ethics approval

Approval for this audit was gained from Guy’s and St Thomas’ NHS Foundation Trust when it was registered as a multidisciplinary audit. Ethics approval was not required.


Guy’s and St Thomas’ Charity provided a grant for Amanda Herbert to carry out this audit during a sabbatical.