International Journal of Epidemiology O International Epldemtologlcal Association 1996 Vol. 25, No. 1 Printed In Great Britain Cervical Neoplasia in Pap Smears: Risk of Cervical Intra-Epithelial Neoplasia (CIN) after Negative or No Prior Smears in a Population without a Mass Screening Programme SIRI FSMO,* BJARNE K JACOBSEN* AND HELGE STALSBERG t Forsmo S (Institute of Community Medicine, University of Tromso, N-9037 Tromse, Norway), Jacobsen B K and Stalsberg H. Cervical neoplasia in Pap. Risk of cervical intra-epithelial neoplasia (CIN) after negative or no prior in a population without a mass screening programme International Journal of Epidemiology 1996; 25: 53-58. Background. The aim of this study was to examine, In a population not submitted to mass screening, the risk of cancer and cervical intra-epithelial neoplasia (CIN) in Pap from women without previously reported positive. Methods. In a logistic regression model consisting of 58 271 from 40 536 Norwegian women the risk of cytologically indicated CIN was studied according to age and time elapsed. Results. The risk of CIN was highest In from women with no previously registered and in taken after an interval of *5 years. Odds ratio for CIN Ml adjusted for age was highest in first time and In taken after an interval of 5 years. Odds ratio for CIN III was highest In first registered. No difference in risk of CIN III was found in taken within one year or 2-3 years after the last smear. The increased risk of CIN III In first was most pronounced in postmenopausal women (>50 years). Nine of 16 with cytological indication of cancer were found in women having a smear taken for the first time. All with malignancy In were >50 years. Conclusions. The risk of cytologically diagnosed premalignant cervical conditions increases with time since the previous smear, but more than 5 years have to elapse before the risk is comparable with that in first taken. Postmenopausal women without previous run the highest risk of serious cervical premallgnancies and cancer. Keywords: cervical neoplasia, Pap smear, mass screening, Norway Detection and treatment of cervical intra-epithelial neoplasia (CIN) is an established method of secondary prevention of squamous cell carcinoma of the uterine cervix. Many countries have systematized this in mass screening programmes, with vaginal (Pap ) as the screening test. Until 1992, Norway had no such screening programme. Nevertheless, there has been a widespread and increasing use of Pap over the last 20-25 years. The majority of are, however, taken from young women. Older women are less frequently tested. 1 Institute of Community Medicine, University of TronKO, N-9037 Tromse, Norway. f Department of Pathology, University Hospital of Tromsf), N-9036 Trornso, Norway. 53 Several risk factors have been noted for cervical premalignancies and cervical cancer. 213 Some of them are human papillomavirus (HPV) infections, 4ti previous dyskaryotic 6 and the diagnosis of CIN. Women with these risk factors should be followed up routinely. As the main objectives of a screening programme for cervical cancer are to detect precursors for disease and early cancer in women without known major risk factors, we believe it is important to evaluate the risk of having indicating CIN and cancer in these women. Two variables of obvious interest are the age of the woman and the period elapsed since the last negative smear. It is the aim of the present study to evaluate this risk. The study population includes the results of more than 58 000 from 40 000 women without previously noted positive in a population with frequent Pap smear testing, but no organized screening programme.
54 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY MATERIALS AND METHODS A computerized registration system was established in 1977 at the Department of Pathology at the University Hospital of Troms0. All cytological and histological samples ever examined in the laboratory are registered. The activity goes back to 1973 and covers the pathology service in the two northernmost counties of Norway, Finnmark and Troms. Since 1981-1982, virtually all histological and cytological specimens from these counties are examined by this laboratory. The system has been operating with an automatic call for follow-up in the case of positive or suspicious cytological diagnoses. No data are registered which give the indication for the smear, whether this is an investigation of gynaecological symptoms or a smear taken as part of a routine gynaecological examination. For our purpose, all registered in the period 1979-1990 were successively categorized as either first smear, follow-up smear or smear from repeated opportunistic testing, respectively. Smears from women with no previously registered smear since 1973 were defined as first smear (smear of entry into the registry). Cytological diagnosis of cervical intraepithelial neoplasia, malignant cells, herpes simplex or HPV infections and benign diagnosis accompanied by a recommendation of follow-up, were classified as positive and subsequent testing as follow-up. Smears from women with no previously registered positive or suspicious were regarded as the result of opportunistic testing. During the 3-year period 1988-1990 a total of 80 191 from 47 754 women were investigated at the laboratory. Of these, 7209 women (15%) were eliminated due to previously registered positive in the registry, leaving 58 271 from 40 536 women aged 13-100 years (mean age 36 years) for this study. Smears with premalignant cytological diagnoses were classified in two levels, CIN I II and C1N III, as we were not able to fully differentiate between CIN I and CIN II due to the diagnostic nomenclature used. Time was calculated for women with a previous negative smear. In the analysis, this interval is given in four categories: «12 months, 13-35 months, 36-59 months and s=60 months (5 years) since the last smear. Age was defined as age at the first smear taken in the period and the analyses were performed in 10-year age groups. Statistics Differences between proportions were tested using % 2 test. Trends were tested with the x 2 test for linear trend. Odds ratios () and 95% confidence intervals (CI) for the risk of CIN I II or CIN III by age and time since last TABLE 1 in women without previous positive by age at first registered smear during the 3 years 1988-1990. Percentage first and median time (months) between two consecutive by age in women with two or more Age >19 20-29 30-39 40-^»9 50-59 60-69 >70 Total Women examined 3847 12 428 9428 6928 3734 2606 1565 40 536 N 4962 18531 13 761 9955 5370 3568 2124 58 271 First smear (%) 66.0 21 2 7.4 5.6 7.5 16.0 30.2 17.9 Median time between two (months) 14.2 20.3 24.3 25.4 26.3 30.4 28.4 23.3 smear were calculated by logistic regression in a model with age in 10-year age groups and time since last smear in four levels as the two independent variables. The analyses were performed with one model consisting of all 58 271 and one model based on the result of the first smear registered in the period 1988-1990 for the 40 536 women. All computations were done after the exclusion of women with a more serious diagnosis, i.e. elimination of women with CIN III and cancer, for calculations of for CIN I II. Odds ratios for cancer in were only computed for women aged s»50 since this diagnosis was not found in from younger women. Data were analysed with SAS. 7 RESULTS Table 1 gives the number of by age of the women and the relation between age and time since last smear. There was a U-shaped relationship between age and the proportion of from women registered for the first time, with the highest proportion in from young women. Among women with two or more registered since 1973, there was a direct relationship between age and time since the last negative smear. About 27% of the from women aged 50-69 years were taken for the first time or after a period of S5 years since last negative smear, compared to 17% from women 30-49 years of age (P < 0.001). In women with two or more, about 60% of all were taken within a 3-year period. In women aged 20-29 years CIN I II was found in 1.1% of the 18 581 and CIN III was found in 0.1% of from the women in this age group. However, CIN III was most frequent in from women >70 years (0.4%). In 10 412 from
CERVICAL NEOPLASIA IN PAP SMEARS 55 TABLE 2 Odds ratio () and 95% confidence intervals (CI) for cervical intra-epilhelial neoplasia (CIN) I-ll and CIN III in 58 271 from w omen without previously registered positive by age at first registered smear and time Smear diagnosis CIN I II CIN III Age 95* CI 3R «19 20-29 30-39 40-49 50-59 60-69 >70 4962 18 531 13 761 9955 5370 3568 2124 43 203 96 43 22 10 10 0.79 0.63 0.39 0.37 0.25 0.43 (0.57-1.10) (0.50-0.81) (0.28-0.54) (0 24-0.58) (0.13-0.48) (0.23-0.81) 0.62 0.66 0.40 0.37 0.23 0.36 (0.44-0.88) (0.51-0.85) (0.28-0.56) (0.24-0.58) (0.12-0.43) (0.19-0.68) 1 ().18 21.00 27.72 13.14 12.96 7.72 9!.73 (0.02-1.32) (0.97-3.06) (0.57-2.30) (0.97-4.01) (0.74-4 09) (1.72-8.22) 0.09 2.20 1.45 2.31 1.56 2.72 (0.01-0.71) (1.22-3.97) (0.71-2.97) (0.66-3.72) (1.24-5.99) Time since last smear First smear <12 months 13-35 months 36-59 months»60 months 10412 7192 26 697 8881 5089 123 35 157 65 47 0.41 0.49 0.61 0.78 (0.28-0.59) (0.38-0.62) (0.45-0.83) (0.55-1.09) 0.40 0.53 0 72 1.03 women with no previously registered smear, CIN I II and CIN HI were found in 1.1% and 0.3% of the, respectively. for CIN and adjusted for age or time are presented in Table 2. Odds ratio for CIN I II decreased with age for women aged &20 years. Controlling for age, the for CIN I II was highest in taken after an interval of 3*5 years, similar to the risk in first registered. Odds ratio for the diagnosis of CIN III increased with age and was highest in women aged s»70 (/"-value for trend < 0.001). A very low risk was found in the youngest women., the for CIN III was lower overall in from women with previously registered than in from women registered for the first time. However, the increased with time. Only in taken after an interval of 3=5 years was the not statistically significantly lower than for CIN III in first registered. The same analysis in a model consisting of only women 3=50 years, and presumably postmenopausal, revealed very low estimates for women who ever had a smear taken and increased with time elapsed (Table 3). A total of 16 women had a malignant cytological diagnosis. Mean age at detection was 66.2 years; the (0.27-0.59) (0.41-0.68) (0.52-) (0.72-1.48) 32 4.00 ().18 22 ().27 14 0.51 18 15 (0.06-0.51) (0.15-0.46) (0.27-0.96) (0.64-2.05) 0.13 0.17 0.31 0.67 (0.04-0.37) (0.10-0.31) (0.16-0.60) (0.36-1.22) youngest woman was 55 years old. for smear indicating cancer increased with age. When adjusting for time since last negative smear, however, no statistically significant difference was found (Table 3). For nine of the 16 women, the cancer indication was given in their first registered smear. Five women had their last negative smear S*5 years prior to the smear indicating malignancy and no smear registered in the interval. Odds ratio for the cytological diagnosis of cancer according to time is given in Table 3. Due to only one case, the intervals of =S35 months were pooled together. When restricting the analyses to the result of the first smear registered during 1988-1990 from each of the 40 536 women tested, the same trends in risk of CIN I II and CIN III by age and time since the last smear were found. In women with no positive and one or more negative previously, 825 and 6800 had to be taken to find one case of cytologically diagnosed CIN III or cancer, respectively. If not tested previously, the numbers of to be taken were 325 and 1200 for CIN III and cancer, respectively. These figures are, however, dependent on the age group and are lowest in the oldest age groups.
56 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 3 Odds ratio () and 95% confidence intervals (CI) for cervical intra-epithelial neoplasia (CIN) III and cancer of the cervix in 11 062 from 7905 women aged a>50 without previously registered positive by age and time since last negative smear Smear diagnosis CIN III Cancer Age 95* CI 50-59 60-69 70-5370 12 1 00 3568 7 0.88 (0.35-2.23) 0.56 (0.22-1.43) 2124 9 1.90 (0.80-4.52) 0.79 (0.32-1.97) 3 8 4.02 (1.06-15.16) 2.55 (0.67-9.75) 5 4.22 (1.01-17.68) 1.87 (0.43-8.15) Time since last smear First smear <12 months 13-35 months 36-59 months»60 months 1618 946 4616 2105 1777 17 1 2 3 5 0.10 (0.01-0.75) 0.04 (0.01-0.17) 0.13 (0 04-0.46) 0.26 (0.10-0.72) 1 00 0.09 0.04 0 12 0.26 (0 01-0.69) (0.01-0.16) (0 03-O.44) (0.09-0.70) 9 1 0 1 5 1 00 0 03 0.08 0 50 (0.00-0.25)* (0.01-0.67) (0.17-1.51) 0.04 0 10 0.54 (0.00-0 31)' (0.01-0.78) (0 18-1.63) 1 Computation of odds ratio of indicating cancer in intervals of < 12 months and 13-35 months were pooled together due to only one case. DISCUSSION Our study shows the risk of CIN and cancer in from a selected part of a population not submitted to organized mass screening. All are from women without registered positive previously, and the results do not represent the final diagnosis after histological investigation, nor do they reflect the prevalence of CIN or cancer in all taken from the North Norwegian population. However, they reflect the prevalence of CIN and cancer in vaginal in women who have either tested negatively previously or have no earlier registered. A small number of women have probably had investigated by other laboratories before moving into the geographical region of Troms and Finnmark. This is mostly a problem for the age group 20-29 years (e.g. students) which is empirically more mobile than older people. The proportion of first time registered may therefore have been slightly overestimated, but we find it unlikely that this has significantly biased the main results of our study. Previous investigation of reasons for cervical testing has revealed that the majority of Pap represent opportunistic testing, 8 i.e. taken outside an organized programme in the absence of genital symptoms. However, clinical symptoms are probably more often the reason for vaginal in older age groups than in the younger. Women aged 50-69 years had a relatively low proportion of first time registered, but the highest risk of having CIN III or cancer diagnosed in of that category. A surprisingly high prevalence of CIN was found in from women aged ^70 in our study. A high proportion, about 30% of all taken in this age group, were registered for the first time. The prevalence of CIN and cancer in from postmenopausal women in organized screening should therefore be expected to be lower than the prevalence found in this study. In our study, many women contribute more than one smear (mean 1.44 ), and a bias may have been introduced as the results from two are not independent. However, our results were essentially the same when only the first smear in the time period 1988-1990 from each woman was included in the analyses. The cytological diagnosis of CIN was given in all age groups. The risk varied with age and interval since the last negative smear. Consistent with knowledge of the natural history of the disease, CIN I II is most frequently diagnosed in young women. 9-10 In our study, the risk was highest in women 20-29 years, whereafter the risk decreased. The risk for CIN III was very low in young women («19 years) and, compared to women 20-29 years old, statistically significantly increased in women aged 2*30.
CERVICAL NEOPLASIA IN PAP SMEARS 57 The risk of CIN III was significantly higher in women with no previous cervical smear than in women with <5 years since last negative smear. This was particularly true for women aged 5=50 and could indicate that CIN III only rarely develops de novo in postmenopausal women." Several authors emphasize the importance of including postmenopausal women not tested for the last 10-12 years in a screening programme until the age of 70-75 years. 12 " 15 However, continuous screening after the age of 50 of women with a prior screening history of regular negative contributes only marginally to the reduction of cervical cancer incidence," with a high cost-benefit ratio. 15 As expected, we found an increasing risk of CIN with increasing intervals since the last negative smear. This has been documented before, 1617 although to our knowledge not in a population without previously noted abnormal. The estimates for the cytological diagnosis of cancer are based on few. Thus, the CI are very wide and should be interpreted with caution. However, the majority of the with a malignant cytological diagnosis were detected in women without previously registered, and, although the final and histologically verified diagnoses for these women are not included in our study, the results indicate that women in this category are at the highest risk for developing cancer of the cervix. Knowledge about the risk of cancer and CIN at different intervals after a negative smear is important when planning a mass screening programme. l8-19 Frequent screening at low risk leads to a high cost-benefit ratio, and undesirable side effects, such as an increased risk of false-positive tests. However, long screening intervals imply an increased risk for rapidly progressing cancers (interval cancers) which would surface clinically at unfavourable stages. The decision about the screening interval has to take into account these factors. The Norwegian screening programme includes women aged 25-70 years with recall every third year. 20 Our study shows that this interval may be appropriate. Only in taken after an interval of s*5 years, was the risk of CIN not significantly reduced compared to the risk of CIN in taken for the first time. Our material does not include the results of histological investigations following the positive as our primary aim was to study the prevalence of women with cytological ly detectable cervical pathology according to the interval since the last negative smear. Mostly women with pathology in Pap are referred for further investigation and treatment, thus knowledge about the prevalence of cytologically detectable malignancy and premalignancy is important when planning a screening programme. However, data about the test validity are necessary to evaluate the results of spontaneous Pap testing on the incidence of CIN and cervical cancer in the population. In summary, we found that the risk of diagnosing CIN and cancer in increases with time since the last negative smear, and women with no previous smear run the greatest risk. This is most pronounced in postmenopausal women. The absolute risk of CIN I III and cancer is, however, low after previous negative. Our study shows the importance of testing postmenopausal women never or rarely tested. If the women are told not to disregard relevant symptoms in the screening interval, an interval of 3 years seems appropriate. REFERENCES ' Forsmo S, Buhaug H, Stalsberg H. Use of Pap- in a population without a mass screening program. Ada Obst Gynecol Scand 1994; 73: 824-28. 2 Clarke E A, Hatcher J, McKeown-Eyssen G E, Lickrish G M. Cervical dysplasia: Association with sexual behavior, smoking and oral contraceptive use? Am J Obstet Gynecol 1985; 151:612-16. 3 Harris R W C, Bnnton L A, Cowdell RHel a/. Characteristics of women with dysplasia or carcinoma in situ of the cervix uteri. Br] Cancer 1980; 42: 359-69. 4 Gram 1 T, Macaluso M, Churchill J, Stalsberg H. Trichomonas vaginalis (TV) and human papillomavirus (HPV) infection and the incidence of cervical intraepithelial neoplasia (CIN) grade III. Cancer Causes Control 1992; 3: 231-36. 'Mitchell H, Drake M, Medley G. Prospective evaluation of risk of cervical cancer after cytological evidence of human papilloma virus infection. Lancet 1986; i: 573-75. 6 Soutter W P, Fletcher A. Invasive cancer of the cervix in women with mild dyskaryosis followed up cytologically Br McdJ 1994; 308: 1421-23. 7 SAS Institute Inc. SAS/STAT, Version 6, 4th Edn. Cary, NC: SAS Institute Inc., 1989. 8 Haugen O A. Forbruk av vaginalcytologiske prpver i Norge. [Use of vaginal in Norway] (Norwegian, English abstract). Tidsskr Nor Lageforen 1984; 104: 952-56. 'Richart R M. Natural history of cervical intraepilhelial neoplasia. Clin Obstet Gynecol 1968; 10: 748-84. 10 Parkin D M, Leach K, Cobb P, Clayden A D. Cervical cytology screening in two Yorkshire areas: Result of testing. Public Health 1982; 96: 3-14. " Van Wijngaarden W J, Duncan I D. Rationale for stopping cervical screening in women over 50. Br Mcd J 1993; 306: 967-71. 12 Fletcher A. Screening for cancer of the cervix in elderly women. Lancet 1990; 335: 97-99. "Mandelblatt J, Schecter C, Fahs M, Muller C. Clinical implications of screening for cervical cancer under Medicare. The natural history of cervical cancer in the elderly: What do we know? What do we need to know? Am J Obstet Gynecol 1991; 164: 644-51. '* Anonymous. Cancer of the cervix: Death by incompetence. Lancet 1985; ii: 363-64. 15 Eddy D M. Screening for cervical cancer. Ann Intern Med 1990; 113: 214-26.
58 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY 16 La Vecchia C, Decarli A, Gentile A, Franceschi S, Fasoli M, Tognoni G. 'Pap' smear and the risk of cervical neoplasia: Quantitative estimates from a case-control study. Lancet 1984; II: 779-S2. 17 Jones C J, Brinton L A, Hamman. RFrt al. Risk factors for in situ cervical cancer: results from a case-control study. Cancer Res 1990; 50: 3657-62. " IARC Working Group on Evaluation of Cervical Cancer Screening Programmes. Screening for squamous cervical cancer duration of low nsk after negative results of cervical cytology and its implication for screening policies. BrMedJ 1986; 293: 659-64. "Syrjinen K J. Epidemiology of human papillomavirus (HPV) infections and their associations with genital tquamous cell cancer. APM1S 1989; 97: 957-70. 20 Masseunderstkelsc for Kreft i LJvmorhaUcn. (Mass screening for cancer of the uterine cervix.) Oslo: Universitetsforlaget, NOU 1987:8. (Revised version received July 1995)