Although it is widely agreed that regular cytologic screening reduces both the incidence and mortality of cervical carcinoma in women of all ages,1 the effectiveness of screening women younger than 30 years has not been well described. Given this lack of evidence of effectiveness and the transient nature of cervical dysplasia, the Advisory Committee on Cancer Prevention of the European Union concluded that screening should be concentrated in the 30–60-year age group and definitely not include women younger than 20.2 Guidelines widely promulgated in the United States since 1988 promote screening in all women who have reached the age of 18 or who have ever been sexually active. As a result, a large number of US women younger than 30 have been screened for cervical cancer.3 Recently, the American Cancer Society published new guidelines stating that cervical cancer screening should begin at the age of 21 years or 3 years after the initiation of sexual intercourse.4 The impact of this delay in initiating screening on cervical cancer incidence is presumed to be small but has not been quantified.
Several prior studies have evaluated the effect of cervical cancer screening in younger age groups by analyzing the trends in cervical cancer incidence since national screening programs were initiated. Despite overall declines in the incidence of invasive cervical cancer, several studies have demonstrated increasing rates in younger women (generally categorized as less than 35 years old) in Japan,5 Canada,6,7 the United Kingdom,8,9 Sweden,10 France,11 Australia,12 and Norway.13 In addition, Sweden,14 Canada,15 and Taiwan16 have documented increases in the incidence of adenocarcinoma among younger women, despite coincident decreases in squamous cell carcinoma.
In the United States, evidence regarding incidence trends has been contradictory. Unlike those in Europe, most US studies analyzing data from the 1970s and 1980s have not documented increases but have found decreases in the incidence rates of overall cervical carcinoma and squamous cell carcinoma in younger women.17–19 Some studies, however, reported a plateauing of rates among women younger than 35 years.20 With respect to adenocarcinoma, increasing incidence rates particular to younger women have also been shown.15,16
Most of these previous US studies of cervical cancer incidence in younger women were based on regional or subpopulations and might not be generalizable to the rest of the country. Other analyses in which national statistics were used found trends through the mid-1980s that might not represent conditions over the last ten years. The objective of our study was to update and examine the trends in the incidence of cervical carcinoma by histologic type in US women less than 30 years old over the last three decades.
MATERIALS AND METHODS
Incident cases of invasive cervical carcinoma, diagnosed in the United States from 1973 to 1999, were obtained from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program nine-registry, publicuse database.21 The population from the nine SEER registries (in Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah) represents approximately 10% of the US population.22 This population is comparable to the general US population with regard to poverty and education, though it tends to be more urban and have a higher proportion of foreign-born persons and non-black minorities than the general US population. Incidence rates (per 100,000 women per year) were calculated with SEER*Stat software,23 and the denominators used were the county-level female populations for the nine geographic areas. The incidence rates were age-adjusted to the 2000 US Standard Population provided by the US Census Bureau.
Cases of cervical carcinoma were identified when the cancer site was designated as “cervix uteri” and then stratified by single year-at-diagnosis time period intervals and 15-year patient age groups: less than 30, 30–44, 45–59, 60–75, and more than 75 years. To take a more detailed look at incidence rates in younger women, we further stratified our less-than-30 age group into three subcategories: less than 20, 20–24, and 25–29 years. In previous studies analyzing cervical cancer incidence by age in the United States, data up to the mid-1980s was used.18,19,24 To determine whether the trends found in these prior studies have continued, we separately evaluated incidence rates from 1985 to 1999.
Tumors were categorized by histologic type according to their assigned first four digits of the morphology codes of the International Classification of Diseases for Oncology, second edition (ICD-O-2).25 For our analysis, cases were grouped as follows: squamous cell carcinoma (ICD-O-2 codes 8050–8130); adenocarcinoma (ICD-O-2 codes 8140–8147, 8160–8162, 8180–8221, 8250–8506, 8520–8550, 8570–8573, and 8940–8941); and other and unspecified carcinomas (all other ICD-O codes for carcinoma of the uterine cervix). Carcinoma in situ cases were not included. Incidence rates were calculated for three entities: invasive cervical squamous cell carcinoma, invasive cervical adenocarcinoma, and all histologic types of invasive cervical carcinoma (a sum of squamous cell carcinoma, adenocarcinoma, and other and unspecified carcinomas).
To measure the incidence trends, we calculated an estimated annual percent change for each age group and histologic type for both the entire evaluation period (1973–1999) and the subset time period (1985–1999). The estimated annual percent change was calculated by fitting a straight-line regression to the natural logarithm of the rates, with calendar year used as a regressor variable. The estimated annual percent change is equal to 100 × (exp(b) − 1), where “b” is the slope of the regression.26,27 If the estimated annual percent change is statistically greater (smaller) than zero, then the incidence rate has an increasing (decreasing) trend. Statistical significance was assessed with a two-sided P value of .05.
The straight-line regression presumes a constant estimated annual percent change over the entire study period. However, sometimes the trends of incidence rates changed during a study period. To identify apparent changes in the incidence trends in our study, we applied a joinpoint regression analysis, a statistical method to detect and describe changes in yearly incidence trends over successive segments of time.26 Briefly, the joinpoint regression analysis first assumes that a straight line best describes the trend. It then selects up to three points in time (called joinpoints) at which the rate of the increasing or deceasing trend changes significantly. Permutation tests are performed to determine the number of join-points that best fit the data. We conducted the joinpoint regression analysis by using the National Cancer Institute's Joinpoint Regression software.28 Assuming heteroscedastic random errors, the estimated annual percent changes were estimated by the weighted least-squares method, with the weight assigned to each observation equal to the square of the incidence rate divided by the square of the standard error of the rate. The 95% confidence intervals (CIs) were produced for each estimated annual percent change calculated. If these intervals excluded zero, the estimated annual percent changes are statistically significant (ie, two-sided P value ≤ .05). Estimated annual percent change analyses could not be performed when an observation contained a zero incidence rate.
For women less than 30 years old in the United States, the following observations were made. First, incidence rates of cervical carcinoma generally and squamous cell carcinoma particularly decreased from 1973 to 1999. Second, the rate of decline in the incidence of cervical carcinoma and squamous cell carcinoma in this population was less than that for older women. Third, rates of adenocarcinoma increased overall but have stabilized over the last decade. Finally, the absolute numbers of annual cases of cervical cancer of all types were relatively small.
Table 1 is a summary of the annual cases (as reported in the SEER catchment areas), denominator population, incidence rates, and estimated annual percent changes by histologic type and age in the SEER regions, in women less than 30 years old from 1973 to 1999. For this population as a group, incidence rates of cervical carcinoma (all histologic types) and squamous cell carcinoma declined. The estimated annual percent changes for cervical carcinoma and squamous cell carcinoma were −0.94% (95% CI −1.47%, −0.41%) and −1.10% (95% CI −1.59%, −0.62%), respectively. For women in the substratified age groups of 20–24 and 25–29, estimated annual percent changes also statistically significantly decreased over this period. Conversely, rates of adenocarcinoma increased in women younger than 30, with an estimated annual percent change of +2.90% (95% CI 1.34%, 4.49%). All types of cervical carcinoma were relatively rare in this age group. The number of annual cases of adenocarcinoma ranged from three to 17 cases per year, corresponding to annual incidence rates from 0.7 to 2.7 per million women. For squamous cell carcinoma, the annual observed cases ranged from 46 to 90, and incidence rates ranged from 8.0 to 14.3 per million women.
Given the periodic occurrence of zero annual cases in the youngest age groups, estimated annual percent changes for women younger than 20 (and age 20–24 for adenocarcinoma only) could not be calculated. Of note, for women less than 20 years old, the absolute number of cases of cervical cancer per year reported to the SEER catchment area, from which incidence rates are derived, was ten or less.
Figures 1 and 2 compare the incidence rate trends of women less than 30 years old with that of women in older age groups. The overall incidence of cervical carcinoma and squamous cell carcinoma declined in all age groups from 1973 to 1999 (Figures 1A and 1B). The rate of decline in the incidence of cervical carcinoma and squamous cell carcinoma increased with age (Figure 2). The estimated annual percent changes in cervical carcinoma and squamous cell carcinoma incidence for women less than 30 years old were statistically significantly lower than that for women aged 45 years or more (P ≤ .05).
The incidence rates of adenocarcinoma increased over the last three decades in women less than 30, 30–44, and 45–59 years old (Figures 1C and 2). However, the rates of increase in the incidence of adenocarcinoma among these three age groups did not statistically significantly differ from each other.
With 10 years of additional incidence data generated since prior studies were published, we used a joinpoint regression model to update trends from 1985 to 1999 in women less than 30 years old (Table 2). With the join-point regression analysis, the best-fit trends might be described by more than one estimated annual percent change. In Table 2, if the best-fit model requires only one estimated annual percent change (ie, no joinpoints), then it is reported under “Trend 1.” If two estimated annual percent changes are used to represent the best-fit model, these values are reported under “Trend 1” and “Trend 2” beside the appropriate time range. Similarly, if three estimated annual percent changes are used, they are reported under “Trend 1,” “Trend 2,” and “Trend 3.” Only the best-fit models are shown.
In this 1985–1999 subset analysis, rates of cervical carcinoma in women younger than 30 were stable until 1993, at which time a steady annual decline of −5.35% (95% CI −10.22%, −0.22%) was noted. When this age group was stratified into smaller age categories, we found no changes in the less-than-20 and 25–29-year age groups. However, for those aged 20–24, there was an increase in rates from 1985–1993, then a decrease. For squamous cell carcinoma, women younger than 30 exhibited a steady decline in incidence rates over the entire 15 years (estimated annual percent change −1.66%; 95% CI −3.11%, −0.18%). When this population was stratified by age, an estimated annual percent change could not be calculated for women less than 20 years old, because of the rarity of occurring cases. Women aged 20–24 exhibited a downward trend after 1993. No change in rates were observed in the 25–29-year age group since 1985. For adenocarcinoma, an estimated annual percent change could not be calculated in women less than 25 years old, whereas no changes for women aged 25–29 were observed. However, when combining all women less than 30 years old as a group, an increase in rates was observed from 1985 to 1990 (estimated annual percent change −19.37%; 95% CI 2.08%, 39.59%). Rates of adenocarcinoma have neither increased nor decreased since 1990.
Prior US studies reported that the incidence of squamous cell carcinoma plateaued during 1979–1985 in women less than 35 years old.20 In our analysis, incidence rates of squamous cell carcinoma in women younger than 30 years declined since 1985. With regard to cervical adenocarcinoma, some researchers have previously documented national increases in rates among women of all ages,24,29 whereas others have found isolated increased rates only in younger age groups, without similar trends in older women.17,18 Our results show that although the incidence of adenocarcinoma in women less than 30 years old has been increasing over the 27-year observation period as a whole, the rates for these young women have been stable since 1990.
Our study has strengths and limitations. One strength involves the SEER Registry, which served as the basis for our analysis. This database is one of the most reliable sources of national cancer incidence statistics available, and careful evaluation is performed routinely to maintain data quality. A primary limitation of our study is based on the relative rarity of invasive cervical carcinoma, and particularly of adenocarcinoma, in the population of women less than 30 years old. Precision around our rate estimates is low because of the small number of observed annual cases from which our rates were calculated. In all histologic groups for women younger than 20 years and in adenocarcinoma for women younger than 30, the observed annual cases in the SEER catchment areas numbered less than 20, which might lead to considerable variation in our rate calculations.
Much concern has been raised over the increasing incidence of adenocarcinoma in younger women. However, it is important to consider the relatively low incidence in this age group. Over the last 27 years, the incidence rate of adenocarcinoma in US women less than 30 years old has been as low as 0.7 per million in 1973, and as high as 2.7 per million in 1990. In the very young (less than 20 years old), 0–2 adenocarcinoma cases were reported annually in the SEER catchment areas. As an illustration of the rarity of the incidence rates stated above, it might be useful to consider breast cancer in men, generally thought to be too rare of a disease to justify screening. From 1995 to 1999, the annual age-adjusted incidence rate for breast cancer in men of all ages was 1.1/100,000.27 This rate is the same order of magnitude as that of all cervical carcinoma and more than five times that of cervical adenocarcinoma in women less than 30 years old during the same period. Therefore, the estimated annual percentage changes we report reflect the relative change in what are, in absolute terms, small numbers. Caution must be exercised in interpreting the clinical significance of these observed trends.
If we assume that our findings are valid, we can only pose hypotheses that might account for these observations. The less dramatic decline in incidence rates for squamous cell carcinoma in younger women compared with that in older women might be because of the following: First, invasive cancer in women less than 30 years old might be a rapidly progressive subset of neoplasias.30 Perhaps the screening interval in this population allows for missed preinvasive lesions, leading to a limited effectiveness in reducing cancer incidence in this age group. On the other hand, aggressive screening of this population might in fact be mitigating an otherwise rising number of cases secondary to the increased prevalence of several risk factors for cervical squamous cell carcinoma, including more risky sexual behavior,31 oral contraceptive use,32 and cigarette smoking.33 This balancing between effective screening and increasing risk factors might result in a stable trend over time. Finally, this is a rare disease in this age group, and perhaps the lack of a more marked decline in incidence is due to the difficulty in accurately assessing trends.
Several authors have suggested possible factors as causing the increased incidence rates of adenocarcinoma in young women. Oral contraceptive use,34–36 more liberal sexual behavior,37,38 and the rising prevalence of smoking37 in this population have been proposed. None of these risk factors explains the plateau in rates of invasive adenocarcinoma observed since 1990.
Our findings have implications for current screening guidelines. We found no statistically significant changes in cervical carcinoma incidence in women less than 20 years old over many decades of screening in the United States, which suggests a possible lack of screening effectiveness in very young women. This observation supports the 2002 American Cancer Society cervical cancer screening guidelines, which recommend delay of initial screening until age 21 years or 3 years after beginning vaginal intercourse. It must be acknowledged, however, that the modest declines in overall cervical carcinoma and squamous cell carcinoma observed since 1973 in the 20–24-year age group, if valid, might be attributable to screening women less than 20 years old.
Future research should address other possible explanations for the observed trends. More importantly, additional information on cervical cancer screening harms in young women, including long-term outcomes of cryotherapy and cone biopsy in large cohorts, is needed. As benefits of screening become increasingly small, uncommon screening harms must also be quantified. Finally, cost-effectiveness and cost-utility analyses need to be performed to craft rational screening policies concerning an optimal age to begin screening that maximize the benefits and minimize the harms in young women.
1. Gustafsson L, Ponten J, Zack M, Adami HO. International incidence rates of invasive cervical cancer after introduction of cytological screening. Cancer Causes Control 1997;8:755–63.
2. Advisory Committee on Cancer Prevention. Recommendations on cancer screening in the European Union. Eur J Cancer 2000;36:1473–8.
3. Centers for Disease Control and Prevention. Trends in cancer screening—United States, 1987 and 1992. MMWR Morb Mortal Wkly Rep 1996;45:57–61. Available at: http://www.cdc.gov/mmwr/preview/mmwrhtml/00040014.htm#00001467.htm
. Accessed 2003 Apr 1.
4. Saslow D, Runowicz CD, Solomon D, Moscicki AB, Smith RA, Eyre HJ, et al. American Cancer Society guideline for the early detection of cervical neoplasia and cancer. CA Cancer J Clin 2002;52:342–62.
5. Ito T, Ishizuka T, Suzuki K, Ikoma Y, Saito J, Onuma M, et al. Cervical cancer in young Japanese women. Arch Gynecol Obstet 2000;264:68–70.
6. Carmichael JA, Clarke DH, Moher D, Ohlke ID, Karchmar EJ. Cervical carcinoma in women aged 34 and younger. Am J Obstet Gynecol 1986;154:264–9.
7. Gallagher RP, Elwood JM. Recent trends in recorded incidence and mortality from uterine cancer. Natl Cancer Inst Monogr 1982;62:83–7.
8. Smales E, Perry CM, Ashby MA, Baker JW. The influence of age on prognosis in carcinoma of the cervix. Br J Obstet Gynaecol 1987;94:784–7.
9. Chilvers C, Mant D, Pike MC. Cervical adenocarcinoma and oral contraceptives. Br Med J (Clin Res Ed) 1987;295:1446–7.
10. Kjellgren O. Mass screening in Sweden for cancer of the uterine cervix: Effect on incidence and mortality. An overview. Gynecol Obstet Invest 1986;22:57–63.
11. Saint-Paul MT, Bremond A, Rochet Y. Le cancer du col uterin avant 35 ans: aspects epidemiologiques et pronostiques. Etude retrospective de 46 cas de cancers du col uterin avant 35 ans, a partir d'une serie de 449 cas de stades IA2 et IV. J Gynecol Obstet Biol Reprod (Paris) 1993;22:737–42.
12. Elliott PM, Tattersall MH, Coppleson M, Russell P, Wong F, Coates AS, et al. Changing character of cervical cancer in young women. BMJ 1989;298:288–90.
13. Eide TJ. Cancer of the uterine cervix in Norway by histologic type, 1970–84. J Natl Cancer Inst 1987;79:199–205.
14. Hemminki K, Li X, Mutanen P. Age-incidence relationships and time trends in cervical cancer in Sweden. Eur J Epidemiol 2001;17:323–8.
15. Liu S, Semenciw R, Probert A, Mao Y. Cervical cancer in Canada: Changing patterns in incidence and mortality. Int J Gynecol Cancer 2001;11:24–31.
16. Wang PD, Lin RS. Epidemiology of cervical cancer in Taiwan. Gynecol Oncol 1996;62:344–52.
17. Peters RK, Thomas D, Hagan DG, Mack TM, Henderson BE. Risk factors for invasive cervical cancer among Latinas and non-Latinas in Los Angeles County. J Natl Cancer Inst 1986;77:1063–77.
18. Schwartz SM, Weiss NS. Increased incidence of adenocarcinoma of the cervix in young women in the United States. Am J Epidemiol 1986;124:1045–7.
19. Chu J, White E. Decreasing incidence of invasive cervical cancer in young women. Am J Obstet Gynecol 1987;157:1105–7.
20. Devesa SS, Young JL Jr, Brinton LA, Fraumeni JF Jr. Recent trends in cervix uteri cancer. Cancer 1989;64:2184–90.
21. National Cancer Institute. Surveillance, epidemiology, and end results (SEER) program public-use data (1973–1999). Released April 2002, based on the November 2001 submission. Bethesda, Maryland: National Cancer Institute, Division of Cancer Control and Population Sciences, Surveillance Research Program, Cancer Statistics Branch.
22. Ries LAG, Eisner MP, Kosary CL, Hankey BF, Miller BA, Clegg L, et al, eds. SEER cancer statistics review, 1973–1999. Bethesda, Maryland: National Cancer Institute. Available at: http://seer.cancer.gov/csr/1973_1999
. Accessed 2002 Feb 5.
23. National Cancer Institute. SEER*Stat 4.2, 2002. Silver Spring, Maryland: Information Management Services, 2002. Available at: http://seer.cancer.gov/seerstat/
. Accessed 2002 March 3.
24. Zheng T, Holford TR, Ma Z, Chen Y, Liu W, Ward BA, et al. The continuing increase in adenocarcinoma of the uterine cervix: A birth cohort phenomenon. Int J Epidemiol 1996;25:252–8.
25. Percy C, Van Holten V, Muir C, eds. International classification of diseases for oncology. 2nd ed. Geneva: World Health Organization, 1990.
26. Kim H, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med 2000;19:335–51.
27. Ries LA, Wingo PA, Miller DS, Howe HL, Weir HK, Rosenberg HM, et al. The annual report to the nation on the status of cancer, 1973–1997, with a special section on colorectal cancer. Cancer 2000;88:2398–424.
28. National Cancer Institute. Joinpoint regression program, version 2.6. Bethesda, Maryland: National Cancer Institute, Division of Cancer Control and Population Sciences, Statistical Research and Applications Branch, 2002. Available at: http://srab.cancer.gov/joinpoint
. Accessed 2002 March 3.
29. Smith HO, Tiffany MF, Qualls CR, Key CR. The rising incidence of adenocarcinoma relative to squamous cell carcinoma of the uterine cervix in the United States—a 24-year population-based study. Gynecol Oncol 2000;78:97–105.
30. Sawaya GF, Sung H, Kearney K, Miller M, Kinney W, Mandelblatt J, et al. Advancing age and cervical cancer screening and prognosis. J Am Geriatr Soc 2001;49:1499–504.
31. Warren CW, Santelli JS, Everett SA, Kann L, Collins JL, Cassell C, et al. Sexual behavior among U.S. high school students, 1990–1995. Fam Plann Perspect 1998;30:170–200.
32. Piccinino LJ, Mosher WD. Trends in contraceptive use in the United States: 1982–1995. Fam Plann Perspect 1998; 30:4–10, 46.
33. Fiore MC. Trends in cigarette smoking in the United States. The epidemiology of tobacco use. Med Clin North Am 1992;76:289–303.
34. Moreno V, Bosch FX, Munoz N, Meijer CJ, Shah KV, Walboomers JM, et al. Effect of oral contraceptives on risk of cervical cancer in women with human papillomavirus infection: The IARC multicentric case-control study. Lancet 2002;359:1085–92.
35. Lacey JV Jr, Brinton LA, Abbas FM, Barnes WA, Gravitt PE, Greenberg MD, et al. Oral contraceptives as risk factors for cervical adenocarcinomas and squamous cell carcinomas. Cancer Epidemiol Biomarkers Prev 1999;8:1079–85.
36. Brinton LA, Herrero R, Reeves WC, De Britton RC, Gaitan E, Tenorio F. Risk factors for cervical cancer by histology. Gynecol Oncol 1993;51:301–6.
37. Kjellberg L, Hallmans G, Ahren AM, Johansson R, Bergman F, Wadell G, et al. Smoking, diet, pregnancy and oral contraceptive use as risk factors for cervical intraepithelial neoplasia in relation to human papillomavirus infection. Br J Cancer 2000;82:1332–8.
38. Ursin G, Pike MC, Preston-Martin S, d'Ablaing G 3rd, Peters RK. Sexual, reproductive, and other risk factors for adenocarcinoma of the cervix: Results from a population-based case-control study (California, United States). Cancer Causes Control 1996;7:391–401.