Pathways to diagnosis of cervical cancer: screening history, delay in follow up, and smear reading

Transcription

1 DOI: /j x Gynaecological oncology Pathways to diagnosis of cervical cancer: screening history, delay in follow up, and smear reading P Priest, a, * L Sadler, a, * J Peters, b S Crengle, c P Bethwaite, d G Medley, e R Jackson a a Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland, New Zealand b Auckland Regional Public Health Service, Auckland District Health Board, Auckland, New Zealand c Te Kupenga Hauora Maori Department of Maori Health, School of Population Health, University of Auckland, Auckland, New Zealand d Department of Pathology and Molecular Medicine, Wellington School of Medicine, Wellington South, New Zealand e Melbourne Pathology, Collingwood, Victoria, Australia Correspondence: Dr P Priest, Department of Preventive and Social Medicine, University of Otago, PO Box 913, Dunedin, New Zealand. patricia.priest@otago.ac.nz Accepted 29 October Published OnlineEarly 12 December Background The aim of this study was to determine the most important ways to reduce incidence of and mortality from cervical cancer by a nationally co-ordinated screening programme. Design Descriptive study. Setting The New Zealand National Cervical Screening Programme: a nationally organised and co-ordinated programme. Sample Women aged younger than 80 years with histologically proven primary invasive cervical cancer, including microinvasive disease, diagnosed between 1 January 2000 and 30 September Consent for access to medical records was gained for 371 of 445 eligible women (83%). A total of 359 (81%) of eligible women or their next of kin consented to interview. Methods Data on events diagnosis were obtained from routine sources, interview, medical record review and slide reread. Main outcome measures Frequency of screening in the 7 years time from abnormal smear or symptoms to appropriate diagnostic confirmation, proportion of negative smears upgraded to high grade on reread. Results Half of the 371 participants (83% of 445 eligible women) had not had a screening smear in the 3 years and 80% were defined as inadequately screened. A maximum of 17% of women overall or within any defined subgroup experienced delays in follow up of abnormal smears or bleeding. Only 11% of women overall had had a high-grade smear, which was originally read as negative. Conclusions The most important factor in women s pathways to a diagnosis of cervical cancer was inadequate screening. While delays in diagnosis could be reduced and laboratory performance improved, priority must be given to improving uptake and frequency of screening. Keywords Cervical cancer, cervical screening, Pap smear, screening programmes. Please cite this paper as: Priest P, Sadler L, Peters J, Crengle S, Bethwaite P, Medley G, Jackson R. Pathways to diagnosis of cervical cancer: screening history, delay in follow up, and smear reading. BJOG 2007;114: Introduction *These two authors contributed equally to this work. No screening programme prevents all cases of disease, and the relatively low sensitivity of the cervical smear 1 means that some cases of cervical cancer will occur even among women who are screened regularly. Nonetheless, cervical cancer screening has prevented the deaths of many women in countries with systematic national programmes 2,3 including New Zealand, where the overall incidence of invasive cervical cancer fell by 45% between 1991 and The reputation of cervical screening programmes has, however, been marred by a number of scandals and enquiries, particularly into the quality of cervical smear reading. 5 7 Routine audit of the cases of women who have developed invasive cervical cancer is an accepted and important part of providing a high-quality screening programme, 2,8 and can include consideration of the adequacy of service along any 398 ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

2 Pathways to cervical cancer of the varied pathways which an individual woman may follow to diagnosis. Without exception, the major finding from these exercises previously is that a significant proportion of women with invasive cervical cancer are either not screened or underscreened The New Zealand National Cervical Screening Programme (NCSP) was established in 1990 and recommends routine three yearly screening of all women aged years. The NCSP is nationally organised and co-ordinated, but there is no national register of eligible women to ensure that all are invited and recalled at appropriate intervals for cervical smears. This study 14 was undertaken in , partly in response to concerns raised at an enquiry into the quality of smear reading in a small New Zealand laboratory. 7 The focus of the study was on assessing the potential contribution of service deficits in screening, follow up of abnormal smears and abnormal bleeding and cytology services. Because at the time of the study, women could opt-off individual smears from the NCSP Register, it was necessary to locate women s clinical records to ensure that a complete smear history was obtained. The study had a specific focus on Maori (indigenous) women who have twice the incidence of, and four times the mortality from, cervical cancer compared with non-maori women. 15 Methods Participants and data collection Women were eligible for inclusion if they were younger than 80 years at diagnosis of histologically proven primary invasive cervical cancer (including microinvasive), were diagnosed between 1 January 2000 and 30 September 2002, and had lived in New Zealand for at least four consecutive years of the 7 years diagnosis. Cancer registration details were obtained from the National Cancer Registry. All the histology reports were reviewed to confirm the diagnosis of invasive cervical cancer. Where uncertainty persisted, further information was sought from healthcare providers. Data were collected relating to participants screening history and pathway to diagnosis for the 7 years diagnosis. These were gathered from the NCSP Register, structured interviews with women or next of kin, primary care and hospital gynaecological records, and a reread of cervical cytology specimens taken in the 6 weeks to 4 years diagnosis. Women were classified as Maori if they were identified as Maori by self-report or by their next of kin at interview or, if no interview was performed, if at least one routine national data source had identified Maori as one of their ethnic groups. Socio-economic status was measured for interviewed women whose address at diagnosis could be assigned to a New Zealand Index of Deprivation 2001 decile, an areabased measure of deprivation calculated from census data. 16 Cancer staging The International Federation of Gynaecology and Obstetrics clinical stage 17 was collected from clinical notes, except for stage 1A, which was defined from histology reports. Pathways to diagnosis Smears in the 6 months immediately diagnosis were excluded from calculations of the frequency of screening, as it was assumed that most were part of the diagnostic process either the positive test that led to detection of early cancer or tests as part of the investigation of symptoms. Inadequate screening was determined among women aged years (those eligible for the NCSP) and categorised in three ways: no screening smears in the 6 84 months diagnosis; no screening smears in the 3-year period from 6 42 months diagnosis and unscreened or screened less than three yearly in the 6 84 months diagnosis. In comparing screening between Maori and non-maori, the non-maori proportion unscreened was age standardised to the Maori population to account for the younger age structure of the Maori population. Expected times for follow up of abnormal smears, bleeding, and abnormal histology were based on targets and standards for the NCSP. 18 Smear reread A smear reread was undertaken to assess the laboratories that initially read the smears against the gold standard of a known high-quality screening laboratory. The reread methodology replicated a normal screening environment, rather than being an expert review, and was conducted in a high-quality Australian laboratory that was commercially independent from New Zealand laboratories. Smears were categorised into four groups: unsatisfactory; negative; low-grade (atypical squamous cells of uncertain significance [ASCUS], low-grade squamous intraepithelial lesion, atypical glandular cells of uncertain significance); and high grade (ASCUS favour high grade, dysplastic cells cannot grade as either high or low, high-grade squamous intraepithelial lesion, adenocarcinoma in situ, and malignant). Smears originally read as unsatisfactory, negative and low grade in the 6 weeks to 4 years diagnosis (index smears) were collected from laboratories and reread in sets of 50 containing up to five index smears, up to five reference smears (made up of known low-grade, high-grade, and unsatisfactory smears), and negative control smears. All the smears were read by three discrete teams and reported as upgraded if all three teams independently reported a diagnosis that would be accompanied by a recommendation for more urgent management than the original diagnosis. ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology 399

3 Priest et al. Analyses Data were analysed using SAS versions 6.12 and 8.2 (SAS Institute Inc., Cary, NC, USA). Where appropriate, differences between proportions have been presented as risk differences with 95% CI calculated using the Wilson method. Chi-square tests have been used in some instances for comparisons of frequency data, with statistical significance defined as P < Results Eligible women and consent Of the 562 women whose information was supplied by the National Cancer Registry, 445 were eligible for inclusion in the study (Figure 1), of whom 62 had died data collection. Consent for access to medical records was gained for 371 (83%) of the eligible women. Seventeen of these women did not consent to interview. In total, 359 (81%) of the eligible women or their next of kin consented to interview. Ninety-five eligible women were Maori; their consent rates were not significantly different from those of non-maori women. Participating women This article reports the results for the 371 eligible women (77 Maori) who gave consent to access to their medical records (participating women). Participating women were more likely to have stage 1 disease and to have had smears in the 6 84 months diagnosis than nonparticipants (Table 1). However, the high consent rate meant that participants were representative of all eligible women in terms of age and stage at diagnosis and the number of smears recorded on the NCSP Register. Table 2 shows information on the demographic and disease characteristics of the participating women. Few women Information from National Cancer Registry 562 Confirmed invasive* cervical cancer, diagnosed in New Zealand, histological date of diagnosis 1/1/00 to 30/9/ n = 7: date of diagnosis out of range n = 9: no cervical histology available n = 14: primary cancer not confirmed as cervical carcinoma n = 18: histology available but not invasive n = 6: histology initially reported as invasive, but final diagnosis noninvasive n = 7: diagnosed overseas Excluded not fulfilling residency criterion 18 over 80 years at diagnosis Table 1. Comparison of women with consent for collection of medical record data with nonconsenting and with all eligible women Consent n 5 371, Nonconsent n 5 74, All eligible n 5 445, Ethnicity* New Zealand European 292 (79) 50 (68) 342 (77) Maori 77 (21) 18 (24) 95 (21) Pacific Island 7 (2) 6 (8) 13 (3) Asian 13 (4) 3 (4) 16 (4) Stage** 1A 58 (16) 9 (12) 67 (15) 1B 131 (35) 17 (23) 148 (33) 2 and higher 86 (23) 28 (38) 114 (26) Unknown 96 (26) 20 (27) 116 (26) Age at diagnosis (10) 7 (9) 43 (10) (30) 20 (27) 130 (29) (22) 21 (28) 104 (33) (17) 8 (11) 72 (16) (12) 10 (14) 53 (12) (9) 8 (11) 43 (10) Smears on NCSP-Register*** None 145 (39) 45 (61) 190 (43) (35) 19 (26) 149 (33) (26) 10 (14) 106 (24) Sample meeting inclusion criteria 445 Figure 1. Confirmation of eligible sample. *Including microinvasive. *From interview where available, otherwise recorded on at least one routine data source. Some data sources allow the recording of several ethnic groups, so numbers in this section of the table add to more than 100%. Asian ethnic group includes women identifying as Chinese or Indian or other Asian. **International Federation of Gynaecology and Obstetrics clinical stage as recorded on the National Cancer Registry. ***Smears recorded on the NCSP Register in the 6 84 months diagnosis of cancer. 400 ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

4 Pathways to cervical cancer Table 2. Characteristics of women who gave consent for access to medical records Total n SCC stage 1A, SCC stage 1B1, All SCC, Adenocarcinoma, Other types, Total women (28) 172 (46) 287 (77) 57 (15) 27 (7) Age* (61) 7 (19) 29 (81) 7 (19) (38) 45 (41) 89 (81) 14 (13) 7 (6) (22) 44 (53) 65 (78) 14 (17) 4 (5) (23) 32 (50) 48 (75) 11 (17) 5 (8) (16) 21 (49) 30 (70) 7 (16) 6 (14) (3) 23 (66) 26 (74) 4 (11) 5 (14) Ethnicity Maori (34) 35 (45) 63 (82) 8 (10) 6 (8) Non-Maori (27) 137 (47) 224 (76) 49 (17) 21 (7) Years of secondary education** (58) 6 (32) 18 (95) 0 1 (5) (30) 67 (49) 111 (80) 16 (12) 11 (8) (33) 57 (39) 107 (73) 32 (22) 8 (5) Deprivation index 338 decile at diagnosis*** (30) 97 (44) 167 (76) 41 (19) 12 (5) (27) 54 (46) 91 (77) 15 (13) 12 (10) Total household income in year 284 diagnosis** Less than $20 000**** (33) 28 (46) 50 (82) 5 (8) 6 (10) More than $ (32) 96 (43) 170 (76) 40 (18) 13 (6) SCC, squamous cell carcinoma. *P, for distribution of age by stage in women with squamous cell carcinoma and known stage at diagnosis. **Questions on education and income were asked only of the 313 women who were alive at interview. ***This score assigns a decile rating to each census area unit, with 10 being the most deprived decile. ****This is equivalent to about GBP were from non-maori, non-european ethnic groups, so ethnic-specific analyses could be performed only for Maori. Average age at diagnosis was 47 years. Ten women with squamous cell carcinoma could not be assigned a stage at diagnosis. As expected from their higher incidence of cervical cancer, Maori women made up a higher proportion of the sample (21%) than New Zealand women aged years (12%). 19 Of the 354 women or next of kin interviewed, 350 (99%) reported that an identified GP cared for the woman or her family during the 7 years diagnosis. Of them, 202 (58%) had only one GP during that time. Screening history Table 3 gives proportions of women aged years with inadequate screening by disease type and socio-demographic category. Seventy-nine percent of women aged years had at least one 3-year period with no smear in the 6 84 months diagnosis. By this definition, approximately three-quarters or more of all subgroups of women, other than women with adenocarcinoma were poorly screened. In general, for each definition of screening used, older women were less well screened, as were women with lower levels of education, lower incomes, and those living in high deprivation areas. A comparison of the proportion of Maori unscreened with the age-standardised proportion of non-maori unscreened shows that Maori were more likely to be unscreened in the 6 84 months diagnosis (41 versus 26%; risk difference 15% [95% CI 4 26]) and in the 6 42 months prior to diagnosis (58 versus 46%; risk difference 12% [95% CI 0 25]). Delays to diagnosis Table 4 provides data on clinical history and the proportions for whom follow up of high-grade smears and abnormal bleeding was prolonged. Maori women were more likely to experience delays than non-maori. However, a maximum of 17% of women overall or within any defined subgroup by ethnicity, stage, or histological type experienced delays in follow up of abnormal smears or bleeding. ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology 401

5 Priest et al. Table 3. Screening among 20- to 69-year-old women by histology, stage, and demographic characteristics All women Maori Non-Maori Total n 6 84 months 6 42 months Screened less than 3-yearly*, Total n 6 84 months 6 42 months Screened less than 3-yearly*, Total n 6 84 months 6 42 months Screened less than three yearly*, Total women (33) 171 (51) 265 (79) (41) 43 (58) 59 (80) (31) 128 (49) 206 (79) Stage SCC stage 1A (19) 48 (46) 81 (78) 26 5 (19) 12 (46) 22 (85) (19) 36 (46) 59 (76) SCC stage 1B (48) 93 (62) 128 (86) (62) 27 (79) 30 (88) (44) 66 (57) 98 (85) Adenocarcinoma 53 6 (11) 14 (26) 36 (68) 8 2 (25) 2 (25) 4 (50) 45 4 (9) 12 (27) 32 (71) Age (11) 16 (44) 28 (78) 12 2 (17) 8 (67) 10 (83) 24 2 (8) 8 (33) 18 (75) (17) 42 (38) 78 (71) 27 5 (19) 10 (37) 18 (67) (17) 32 (39) 60 (72) (36) 47 (57) 66 (80) (55) 13 (65) 16 (80) (30) 34 (54) 50 (79) (47) 36 (56) 57 (89) 11 8 (73) 8 (73) 11 (100) (42) 28 (53) 46 (87) (25) 30 (70) 36 (84) 4 4 (100) 4 (100) 4 (100) (59) 26 (67) 32 (82) Deprivation index (28) 89 (44) 155 (77) 23 6 (26) 9 (39) 16 (70) (28) 80 (45) 139 (78) (40) 63 (61) 85 (82) (48) 29 (69) 37 (88) (35) 34 (55) 48 (77) Education (years at secondary school) (60) 11 (73) 14 (93) 8 5 (62) 6 (75) 7 (87) 7 4 (57) 5 (71) 7 (100) (31) 71 (54) 96 (73) (33) 20 (61) 25 (76) (31) 51 (52) 71 (72) (21) 53 (38) 108 (77) 17 3 (18) 4 (24) 12 (71) (33) 49 (40) 96 (78) Income ,$ (42) 33 (62) 49 (75) 17 6 (35) 9 (53) 12 (71) (44) 24 (67) 28 (78) $ (23) 91 (42) 161 (75) (36) 20 (51) 30 (77) (20) 71 (40) 131 (74) SCC, squamous cell carcinoma. *Unscreened or screened less than three yearly in 6 84 months diagnosis. 402 ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

6 Pathways to cervical cancer Table 4. Pathways to diagnosis among all participants Total n 5 371, SCC stage 1A n 5 105, SCC stage 1B1 n 5 172, All SCC n 5 287, Adenocarcinoma n 5 57, Maori n 5 77, Non-Maori n 5 294, Pathways for women with high-grade smears Total women having a high-grade 271 (73) 100 (95) 107 (62) 214 (75) 41 (72) 57 (74) 214 (73) smear diagnosis of cancer Time from first high-grade smear to 14 (4) 9 (9) 2 (1) 11 (4) 2 (4) 6 (8) 8 (3) colposcopy.12 weeks Time from first high-grade smear to 25 (7) 14 (13) 7 (4) 21 (7) 3 (5) 7 (9) 18 (6) histology.12 weeks Time from first high grade smear to diagnosis.6 months 34 (9) 16 (15)* 10 (6)* 26 (9) 6 (11) 12 (16)** 22 (7)** Pathway for women with abnormal bleeding alone Total women with persistent PCB or 48 (13) 0 32 (19) 33 (11) 9 (16) 9 (12) 39 (13) PMB without high grade smear diagnosis of cancer Time from first clinical record of PCB or PMB to diagnosis.2 months (in women with no high-grade smear) 17 (5) 0 11 (6) 11 (4) 4 (7) 3 (4) 14 (5) Pathway from colposcopy to diagnosis Total women having a colposcopy 134 (36) 68 (65) 47 (27) 116 (40) 13 (23) 33 (43) 101 (34) and CIN21/AIS histology Time from colposcopy to diagnosis 36 (10) 18 (17)*** 12 (7)*** 31 (11) 3 (5) 13 (17)**** 23 (8)**** among women with CIN21/AIS histology.2 months Past history of cervical treatment Total women with history of cervical 30 (8) 7 (7) 17 (10) 24 (8) 5 (9) 9 (12) 21 (7) treatment at least 1 year diagnosis of cancer In 1 7 years diagnosis 10 (3) 4 (4) 4 (2) 8 (3) 1 (2) 4 (5) 6 (2).7 years diagnosis 22 (6) 3 (3) 15 (9) 18 (6) 4 (7) 6 (8) 16 (5) AIS, adenocarcinoma in situ; CIN21, cervical intraepithelial neoplasia grade 2 and above (not cancer); PCB, postcoital bleeding; PMB, postmenopausal bleeding; SCC, squamous cell carcinoma. *Risk difference 9% (95% CI: 2 18). **Risk difference 8% (95% CI: 1 18). ***Risk difference 10% (95% CI: 3 19). ****Risk difference 9% (95% CI: 1 19). Smear reread Forty-eight percent of women had at least one negative, low-grade or unsatisfactory smear reread (250 of the 257 eligible smears were reread). Table 5 shows the proportion of all participating women who had smears upgraded. Women with adenocarcinoma were more likely than women with squamous disease to have a smear upgraded to high grade, because they were more likely to have smears eligible for reread (they were better screened). The most important upgrade is the upgrade of prior negative smears to high grade, a threshold with good interobserver reproducibility and the misread most important to avoid since it represents a lost opportunity for immediate referral for investigation. Twenty-four percent of women, who had a negative smear in the 6 weeks to 4 years or 11% of all women, had a negative smear upgraded to high grade. No factors identified Twenty-seven women (7%) had an adequate screening history, no delays to diagnosis as defined above, no previous treatment, and no smears upgraded to high grade on reread. Eight (30%) of these women had adenocarcinoma, 10 (37%) had stage 1A squamous disease, 4 (15%) squamous 1B, 2 (7%) squamous stages 2 and 3, and (11%) had other epithelial types. ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology 403

7 Priest et al. Table 5. Smear reread results among all participants Total SCC 287, Adenocarcinoma 57, Total 371, Total women with at least one smear reread 120 (42) 43 (75) 178 (48) Findings of reread of prior negative smears Total women with negative smears reread 106 (37) 40 (70) 160 (43) At least one negative smear upgrade 44 (15) 16 (28) 61 (16) Any negative smear upgrade to high grade 27 (9) 11 (19) 39 (11) Any negative smear upgrade to unsatisfactory 6 (2) 2 (4) 9 (2) Any negative smear upgrade to mixed high, low and unsatisfactory 13 (5) 3 (5) 16 (4) Any negative smear upgrade to low grade Findings of reread of prior low-grade smears Total women with low-grade smears reread 39 (14) 10 (18) 53 (14) At least one low-grade smear upgraded by reread 18 (6) 5 (9) 25 (7) Findings of reread of prior unsatisfactory smears Total women with unsatisfactory smears reread 6 (2) 2 (4) 8 (2) At least one unsatisfactory smear upgraded to high grade by reread At least one reread smear (negative, low grade, or unsatisfactory) upgrade to high grade 42 (15) 14 (25) 59 (16) SCC, squamous cell carcinoma. Discussion The pathway to a diagnosis of cervical cancer may include one or more factors such as inadequate screening, missed abnormal smears, delays in investigation of abnormal smears or bleeding, and a history of previous treatment of dysplasia. Because not all the cases of cervical cancer can be prevented by screening, and because a woman may experience more than one of these factors, we did not attempt to place all women in mutually exclusive categories of screening failure, but documented the proportion of women who experienced each factor. Almost all (93%) of the women in this study experienced at least one of these factors diagnosis. While the relatively low sensitivity of cervical cytology 1 means that no cytology screening programme can prevent all cases of cervical cancer, inadequate screening is the predominant factor in the failure of the New Zealand Cervical Screening Programme to protect these women from cervical cancer. Half of all women with cervical cancer had not had a screening smear in the 3 years diagnosis. The proportion screened less than three yearly over the 7 years diagnosis was even higher at 80%. Older women, Maori, and economically and educationally disadvantaged women were more likely to be inadequately screened than other women. The New Zealand Cervical Screening Programme monitoring reports show that the adjusted for hysterectomy proportion of New Zealand women aged years with a smear recorded on the Register in the previous 3 years has been about 73% since 1997, 20,21 significantly below the Programme target of 85%. 18 Figure 2 shows the overall population agespecific proportions of Maori and non-maori, non-pacific women without a smear on the NCSP Register in the 3 years 2001, the most recent year for which these data are available. These data are not strictly comparable with the data presented here on women with cancer because participants in the study were more likely to have smears on the NCSP Register than those who did not consent to take part (so the true proportion of women with cancer who are unscreened would be higher than that found in this study) and because the population figures use a denominator unadjusted for women who have had hysterectomy (so the true proportion of the eligible population who are unscreened would be lower than that shown in the figure). However, they do illustrate the fact that, in general, women with cancer were less well screened than the population. That 50% of women with cervical cancer had not had a screening smear in the 3 years diagnosis is consistent with recent published studies, 9,11,13 as is the finding that onethird of women were effectively never screened. 8,10 12 Ethnic, age, and socio-economic differences in screening found in this study are also broadly consistent with the few studies that report screening in ethnic minority and deprived populations. 9,13,22 Poor smear sensitivity means that the success of cervical screening relies upon smears being performed regularly. Previous studies have reported rates only for never screened and for a single smear within a 3- to 5-year period. 8 11,13,23,24 The inclusion of two complete screening cycles in this study unmasked the true extent to which women were underscreened when regular screening was broadly measured according to the New Zealand guidelines, women were much more poorly screened than would appear to be the case when only one smear cycle was used to determine adequacy of screening. 404 ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

8 Pathways to cervical cancer Māori Non-Māori, non- Pacific Age group (years) Figure 2. Proportion of women aged years with no record of a smear on the NCSP Register in the 36 months to 31 December 2001, by 5-year age group and ethnicity. From data in NCSP Annual monitoring Report No more than 17% of women overall or within any defined subgroup by ethnicity, stage, or histological type experienced delays in follow up of abnormal smears or bleeding. Previously reported studies have defined delay to diagnosis in a number of different ways, 8 13 but none has reported as comprehensive data on delay in follow up of abnormal smears or symptoms as this study. Sasieni et al. found that 8% of cancer cases were not diagnosed within 6 months of a highgrade smear, 8 and reported delays by other definitions are also consistent with the audit finding of 9% women experiencing a delay from high-grade smear to diagnosis. Delay was more common in early stage disease, and also consistent with previous studies. 8,9,13 Our data show that as well as being more likely to be inadequately screened, Maori women were generally more poorly served than non-maori by the parts of the screening programme that follow-up symptoms and abnormal smears. No other studies have reported delay in follow up by ethnicity, although there is increasing evidence in the literature that indigenous and minority groups are more poorly served by health systems. 25,26 Leyden et al. 13 found that failure in follow up (a smear read as ASCUS or more severe in the 6 36 months diagnosis) was not more common in any particular ethnic group within women who had been members of a comprehensive prepaid health plan for at least 33 of the 36 months diagnosis. However, the populations from which their cases were drawn were well screened, with relatively low rates of invasive cervical cancer. Only 11% of women with cancer had had a high-grade smear, which was originally misread as negative. There is a paucity of published data with which to compare these results, but the proportion of smears upgraded did not uphold the concern that laboratories performance at a national level was unacceptable. 14 Some questions remain unanswered by this study. Although initially we expected to make a distinction between screen-detected cancers and those detected through the investigation of symptoms we were unable to do so. Even in primary care notes, it was not always clear why a woman had a smear in particular, when she had not had a smear for a long time, it was often unclear why she had come for a smear at this point, and sometimes self-report, primary care and hospital notes differed as to whether there had been any symptoms that had prompted the request for a smear. However, stage 1A cancers, 28% of cancers in this study, are likely to be screen detected and could be considered (limited) successes of screening. It was not possible to identify the reasons for irregular or absent screening, delays in diagnosis and the disparities identified by this study. Reasons for nonparticipation in screening were not sought in this study. It is more appropriate to investigate these among women without cancer who are not participating in screening, as information from these women is less likely to be affected by recall bias. More detailed study of diagnostic services would be necessary to define the reasons for diagnostic delay. ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology 405

9 Priest et al. This study shows that Maori women with cancer, like Maori women in general, are less well screened than non- Maori women. This presumably contributes to their higher incidence of cervical cancer. The higher mortality : incidence ratio in Maori women (mortality four times non-maori, incidence two times) could be partly due to higher stage disease in Maori, but this does not appear to be supported by these data. Further analyses to investigate this are planned. Conclusion A study of women with invasive cancer does not provide a complete picture of a screening programme. Many women in New Zealand have precancerous lesions successfully detected and treated as a result of screening. However, this study shows that the most important attributable factor on the pathway to cervical cancer is insufficient screening. Addressing inadequate levels of screening is necessary to reduce the risk of this largely preventable disease for New Zealand women. At present in New Zealand the smear taker, usually the GP, is primarily responsible for invitation and recall. The low levels of screening among women in this study occurred despite almost all of them having a GP identified as caring for the woman or her family during the 7 years with more than half having only one GP in this time. The report of this study recommended establishment of a population-based register within the New Zealand NCSP to improve invitation and recall for screening. 14 The finding that Maori and other disadvantaged women are more likely to be underscreened is consistent with disparities seen elsewhere in health care in New Zealand, 26 and the poorer screening and follow up of women from less powerful socio-economic groups is also likely to be mirrored in other countries. These high-risk groups who are underscreened must be the focus of efforts to increase screening coverage and improve progress through the diagnostic pathway. This study was given impetus largely by concerns that poor laboratory performance in New Zealand might be widespread, but systemic failings in smear reporting were not detected by this study. 14 Laboratory performance can always be improved, and new technologies, such as liquid-based cytology that might improve the sensitivity of smears attract much attention, despite lack of supportive evidence. 27,28 Liquid-based cytology has been recommended in England and Wales on the basis of a cost-effectiveness study of a completely screened population. 29 However, the findings of this study emphasise that priority must be given to initiatives to increase the uptake and regularity of screening. New screening technologies should have lower priority for New Zealand unless they also facilitate easier, cheaper, more accessible, and effective screening for high-risk populations. 30 Ethical approval Received in May 2002 from all New Zealand Ethics Committees. Acknowledgements Ruth Herbert and William Rainger managed the project. Sandra Coney, Peter Sykes, Peter Bethwaite, Gabriele Medley and Harold Neal provided expert advice to the project. Rachel Brown, Claire Macdonald, Orewa Barrett-Ohia, Charlotte Bergman, and Rhondda Kerins performed the interviews, and Catherine Dick, Stephanie Smith, Kirstine Sutton, and Sarah Sparks abstracted the medical records data. This study was funded by the New Zealand Ministry of Health (MoH) and carried out as a partnership between the MoH and Uni- Services Ltd, University of Auckland. The proposal for the study was initiated by the MoH, and the MoH managed operational aspects of the study. Interviewers and medical records abstractors were employed by the MoH, with oversight of the content of their work by the authors. The authors designed the study and performed analysis and interpretation of results independently. The full report is available online at j References 1 Nanda K, McCrory DC, Myers ER, Bastian LA, Hasselblad V, Hickey JD, et al. Accuracy of the Papanicolaou test in screening for and follow-up of cervical cytologic abnormalities: a systematic review. Ann Intern Med 2000;132: Sasieni P, Adams J, Cuzick J. Benefit of cervical screening at different ages: evidence from the UK audit of screening histories. Br J Cancer 2003;89: Peto J, Gilham C, Fletcher O, Matthews FE. The cervical cancer epidemic that screening has prevented in the UK. Lancet 2004;364: New Zealand Health Information Service. Mortality and Demographic Data Wellington: Ministry of Health, Wells W. Review of Cervical Screening Services at Kent and Canterbury Hospitals NHS Trust. London: NHS Executive, McGoogan E, Day N, Dorrian L, Hudson EA. Report of the Inquiry into Cervical Cytopathology at Inverclyde Royal Hospital, Greenock. Edinburgh: The Stationery Office Books, Duffy AP, Barrett DK, Duggan MA. Report of the Ministerial Inquiry into the Under-Reporting of Cervical Smear Abnormalities in the Gisborne Region. Wellington: Ministry of Health, Sasieni PD, Cuzick J, Lynch-Farmery E. Estimating the efficacy of screening by auditing smear histories of women with and without cervical cancer. Br J Cancer 1996;73: Sung HY, Kearney KA, Miller M, Kinney W, Sawaya GF, Hiatt RA. Papanicolaou smear history and diagnosis of invasive cervical carcinoma among members of a large prepaid health plan. Cancer 2000;88: Janerich DT, Hadjimichael O, Schwartz P, Lowell DM, Meigs J, Merino MJ, et al. The screening histories of women with invasive cervical cancer, Connecticut. Am J Public Health 1995;85: ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology

10 Pathways to cervical cancer 11 Stuart GCE, McGregor E, Duggan MA, Nation JG. Review of the screening history of Alberta women with invasive cervical cancer. CMAJ 1997;157: Baldauf JJ, Dreyfus M, Ritter J, Meyer P, Philippe E. Screening histories of incidence cases of cervical cancer and high grade SIL. A comparison. Acta Cytol 1997;41: Leyden WA, Manos MM, Geiger AM, Weinmann S, Mouchawar J, Bischoff K, et al. Cervical cancer in women with comprehensive health care access: attributable factors in the screening process. J Natl Cancer Inst 2005;97: Sadler L, Priest P, Peters J, Crengle S, Jackson R. The New Zealand Cervical Cancer Audit, Whakamaatau Mate Pukupuku Taiawa o Aotearoa. Wellington: Ministry of Health, New Zealand Health Information Service. Cancer: New Registrations and Deaths Wellington: Ministry of Health, Ministry of Health. NZDep2001 Population Distributions. [Internet] 2004 [ Distributions/$File/NZDep2001PopulationDistributions]. Accessed 30 August Benedet JL, Odicino F, Maisonneuve P, Beller U, Creasman WT, Heintz AP, et al. Carcinoma of the cervix uteri. Int J Gynaecol Obstet 2003; 83: National Cervical Screening Programme. Interim Operational Policy and Quality Standards. Wellington: Health Funding Authority, Statistics New Zealand Census: Ethnic Groups [ Accessed 26 April Independent Monitoring Group. Annual Monitoring Report 2001: National Cervical Screening Programme. Dunedin: Hugh Adam Cancer Epidemiology Unit, Department of Preventive and Social Medicine, University of Otago, National Screening Unit. Independent Monitoring Group Reports. [ Accessed 26 April Baker D, Middleton E. Cervical screening and health inequality in England in the 1990s. J Epidemol Community Health 2003;57: Kenter GG, Schoonderwald EM, Koelma IA, Arentz N, Hermans J, Fleuren G. The cytological screening history of 469 patients with squamous cell carcinoma of the cervix uteri; does interval carcinoma exist? Acta Obstet Gynecol Scand 1996;75: Womack C, Warren AY. Achievable laboratory standards; a review of cytology of 99 women with cervical cancer. Cytopathology 1998; 9: Crengle S, Crampton P, Woodward A. Mäori in Aotearoa/New Zealand. In: Healy J, McKee M, editors. Accessing Health Care: Responding to Diversity. Oxford: Oxford University Press; pp Ajwani S, Blakely T, Robson B, Bonne M, Tobias M. Decades of Disparities: Ethnic Mortality Trends in New Zealand Wellington: Ministry of Health, U.S. Preventive Services Task Force. Recommendations and Rationale: Screening for Cervical Cancer [ 3rduspstf/cervcan/cervcanrr.htm]. Accessed 29 June Randall P, Majoribanks J, Farquhar CM. Liquid-based cytology in cervical screening: a systematic review. A Report for the National Screening Unit of the Ministry of Health. Wellington: Ministry of Health, Karnon J, Peters J, Platt J, Chilcott J, McGoogan E, Brewer N. Liquidbased cytology in cervical screening: an updated rapid and systematic review and economic analysis. Health Technol Assess 2004;8: Sawaya G, Grimes D. New technologies in cervical cytology screening: a word of caution. Obstet Gynecol 1999;94: ª 2006 The Authors Journal compilation ª RCOG 2006 BJOG An International Journal of Obstetrics and Gynaecology 407