Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Epidemiol Biomarkers Prev. Author manuscript; available in PMC 2010 April 1.
Published in final edited form as:
PMCID: PMC2667567

Gender is an age-specific effect modifier for papillary cancers of the thyroid gland



Thyroid cancer incidence rates have increased worldwide for decades, although more for papillary carcinomas than other types and more for females than males. There are few known thyroid cancer risk factors except female gender, and the reasons for the increasing incidence and gender differences are unknown.


We used the National Cancer Institute’s Surveillance, Epidemiology, and End Results 9 Registries Database for cases diagnosed during 1976–2005 to develop etiological clues regarding gender-related differences in papillary thyroid cancer incidence. Standard descriptive epidemiology was supplemented with age-period-cohort (APC) models, simultaneously adjusted for age, calendar-period and birth-cohort effects.


The papillary thyroid cancer incidence rate among females was 2.6 times that among males (9.2 versus 3.6 per 100,000 person-years, respectively), with a widening gender gap over time. Age-specific rates were higher among women than men across all age groups, and the female-to-male rate ratio declined quite consistently from more than five at ages 20–24 to 3.4 at ages 35–44 and approached one at ages 80+. APC models for papillary thyroid cancers confirmed statistically different age-specific effects among women and men (p < .001 for the null hypothesis of no difference by gender), adjusted for calendar-period and birth-cohort effects.


Gender was an age-specific effect modifier for papillary thyroid cancer incidence. Future analytic studies attempting to identify the risk factors responsible for rising papillary thyroid cancer incidence should be designed with adequate power to assess this age-specific interaction among females and males.

Keywords: Thyroid cancer incidence, age-period-cohort analysis, gender


Thyroid cancer is the most common malignant disease of the endocrine system and the sixth most common cancer among females (1). The age-adjusted incidence rates for thyroid cancers have increased worldwide for decades (25), though more among females than males and more for papillary than other histological types (3). There are few known thyroid cancer risk factors except female gender and radiation, and the reasons for the increasing incidence or gender differences are unclear. Understanding the differences in papillary thyroid cancer incidence by gender could inform future analytic studies and intervention strategies. We used the National Cancer Institute’s Surveillance, Epidemiology, and End Results program to examine thyroid cancer incidence in the United States, stratified by gender and with a focus upon papillary lesions. We supplemented standard descriptive techniques (age- standardized temporal trends and age-specific incidence rates) with a comparative age-period-cohort analysis (APC) analysis.

The application of APC methods may facilitate the assessment of descriptive data, which can be confounded by age, calendar-period, and/or birth-cohort effects. Age effects reflect age-associated genetic events and/or carcinogenic exposures (6). Calendar-period effects may be observed due to changes in screening practices, diagnostic techniques, or classification definitions. Calendar-effects are “cross-sectional” since they span or cut across all age groups and birth-cohorts within a given study period. Birth-cohort effects are longitudinal because they reflect the net impact of risk factors on the incidence rates within one generation or birth cohort (7, 8). Comparison of rates by gender and age across study periods and across birth cohorts may help discern the relative roles of these factors and reveal changes in risk that provide clues for further etiologic research.

Three groups have recently examined time trends in thyroid cancer incidence using SEER data (3, 5, 9, 10). Only one used APC models (9), and they found significant age, period, and birth cohort effects for papillary thyroid cancers in both females and males. In this study, we were mainly interested the differential APC effects for papillary thyroid cancer between females and males, using a comparative approach. “Comparative” APC models link standard APC analyses with statistical analysis of failure time data for an assessment of the null hypothesis of no difference across levels of a variable, i.e., gender in this case.


We used the National Cancer Institute’s Surveillance, Epidemiology, and End Results 9 Registries Database, November 2007 Submission (11) to analyze male and female thyroid carcinoma incidence rates from 1976 through 2005. The SEER 9 Registries Database included registries in Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah, and included approximately 10% of the US population.

Invasive thyroid cancer cases were coded using the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) (12) and were stratified into the four major histological types: papillary carcinoma (ICD-O-3 codes 8050, 8260, 8340–8341, 8343–8344, 8350), follicular carcinoma (ICD-O-3 codes 8290, 8330–8332, 8335), medullary carcinoma (ICD-O-3 codes 8345–8346, 8510), and anaplastic carcinoma (ICD-O-3 codes 8012, 8020–8021, and 8030–8032); all other ICD-O-3 codes were categorized as other or unknown.

In addition to gender and histologic type, we evaluated demographic and tumor characteristics including age at diagnosis, race, SEER historic stage A, and tumor size. Racial groups included Whites, Blacks, and other (American Indian, Alaskan Natives, Asians, and Pacific Islanders combined) or unknown race. SEER historic stage A was used to classify thyroid cancers as localized (limited to the thyroid gland), regional (limited to surrounding tissues), and distant or systemic disease (12). SEER collected all demographic and tumor variables from 1973 forward, except for tumor size. Beginning in 1988, SEER systematically collected information on tumor size, that is the cancer’s greatest diameter as recorded on surgical pathology reports. We categorized tumor size as ≤1 cm, >1 cm and ≤2 cm, >2 cm and ≤4 cm, and greater than 4 cm based on the Extent of Disease- 10 (EOD-10) codes for 1988–2003 and the Collaborative Staging (CS) codes for 2004 and 2005 (12).

Data Analysis

Incidence rates (IR) were calculated using SEER*Stat 6.4.4 (13) and expressed per 100,000 person-years, man-years, or woman-years. Age-adjusted rates were standardized to the 2000 US population. Relative risks were expressed as incidence rate ratios (IRR), where a given characteristic was compared to a referent rate with an assigned IRR of 1.0. Statistical significance of rates and IRR were assessed at the p <0.05 alpha level; all hypothesis tests were two-sided.

We divided the 1976 to 2005 study period into either six 5-year calendar-periods (1976–80, 1981–85, 1986–90, 1991–95, 1996–2000, and 2001–05) or three 10-year calendar-periods (1976–85, 1986–95, and 1996–2005). Temporal trends were expressed as the percentage change (%CH = 100 X [final IR-initial IR]/initial IR) from 1976–85 to 1996–2005 (or 1988–95 to 1996–2005 for tumor size); 95% confidence limits were calculated with the delta method for men and women separately (14). Age-adjusted temporal trends and age-specific incidence rates were plotted on a log y and linear x scale, such that a slope of 10 degrees approximated a change in rates of 1% per year (15).

APC models simultaneously assessed the effects of age at diagnosis, calendar period of diagnosis, and year of birth (78, 1618). Given the relationship [birth cohort] = [calendar-period] minus [age at diagnosis], we had eighteen 5-year birth cohorts (1896 to 1980, referred to by the mid-year of birth) calculated from six 5-year calendar periods (1976–80, 1981–85, …2001–05) and thirteen 5-year age groups (ages 20–24, 25–29, … 80–84 years). The assumption of APC analysis is that the population-based rates have a log-linear relationship and Poisson distribution for age, period, and cohort effects:


Here, pij is the expected log rate over the age groups αi, calendar-periods [var phi]j, and birth-cohorts γk.

Owing to the linear APC relationship, the separate or independent effects of age, period, and cohort cannot be estimated or determined; this is called the non-identifiability problem of APC analysis. Notwithstanding the well-established non-identifiability issue, certain APC parameters can be estimated if the age, period, and cohort effects are orthogonally decomposed into their linear and non-linear components, following the method of Holford (17). The estimable APC parameters included “drifts” (7), “deviations” (17), “curvatures” (8), and “slope contrasts” (18).

APC deviations reflect the non-linear departures from their respective linear trends. Net drift is the linear trend in the logarithm of the age-specific rates over time and is equal to the summation of the period and cohort slopes, i.e., ([var phi]L + γL), where [var phi]L and γL are the linear trends for calendar-period and birth-cohort effects, respectively. The net drift is estimable (identifiable) even though the constituent components [var phi]L and γL are not. Differences in the net drifts were used to test for differential temporal trends. Another type of drift parameter is the longitudinal age trend (LAT). The longitudinal age trend is the linear trend in the logarithm of the age-specific rates across age and is equal to the summation of the age and period slopes, i.e., (αL + [var phi]L), where αL and [var phi]L are the linear trends for age and period, respectively. Differences in the net drifts were used to test for differential temporal trends. Differences in the longitudinal age trends were used to test for differential age-related effects for the study period.

Another useful APC function is the “fitted” age-at-onset curve, as recently introduced (19):


The parameter [rho with circumflex]i reflects the age-specific incidence rate curve after adjustment for calendar-period and birth-cohort deviations; and approximates the true age-specific incidence rate curve devoid of period and/or cohort factors. In this model μ is the intercept term, (αL + [var phi]L) is the longitudinal age trend, and [alpha]i is the age deviation over interval i.


Thyroid cancer incidence rates by histologic type and papillary thyroid cancer incidence rates by demographic and tumor characteristics and female-to-male incidence rate ratios are shown in table 1. Approximately 80% of all thyroid cancers were the papillary type, 13% were follicular, <3% were medullary, <2% were anaplastic, and 3% were the other/unknown type. Over the entire 30-year calendar period 1976–2005, the rate among females was 2.6 times that among males (female-to-male IRRFM = 2.58; 2.52–2.63); the IRRFM for papillary thyroid cancer was higher (2.85 overall). Over the entire period, the IRRFM for papillary thyroid cancer decreased from more than 5.0 among those aged 10–19 and 20–29 years to 1.1 at ages 80+ years. Rates were higher among Whites than Blacks and highest among other races, which included Asians and/or Pacific Islanders and Native Americans. Among females and males, most thyroid cancers were small (<2.0 centimeters) and were usually diagnosed at the localized stage. Rates for all races combined were highest in Hawaii, followed by Utah and New Mexico, and lowest in Atlanta.

Table 1
Thyroid Cancer Incidence in SEER’s 9-Registry Database (1976–2005) by gender

Thyroid cancer incidence rates rose over the time period 1976–2005 among men and women for the papillary type in particular (tables 2A and 2B). Papillary carcinoma rates doubled, and the small decline in other/unknown types could have played only a very minor role. Papillary cancer rates rose over time in every age group except ages 0–9 years for both men and women; and rates more than doubled among those aged 40–79 years. The increases were more rapid for females than males, for Whites than Blacks or other races, and for the smallest and largest tumor sizes compared to those size >1 cm and ≤4 cm.

From 1976–80 to 2001–05, age-adjusted papillary thyroid cancer rates rose 158% from 4.7 to 12.1 per 100,000 among females and rose 106% from 1.9 to 3.9 per 100,000 among males (figure 1A). However, rates changed little until after 1985 among females and 1990 among males. From 1990–94 to 2000–04, rates rose 80% among females and 61% among males. The age-specific incidence rate patterns during 1976–2005 differed among females and males, with an early age-at-onset predominance for females (figure 1B). That is, incidence rates rose rapidly with increasing age among females, peaking at ages 40–49 years with a rate of 12.8 per 100,000 woman-years then declined to 4.4 per 100,000 at ages 80+ years. In contrast, age-specific rates among males rose more slowly, peaking at ages 60–69 years with a rate of 5.5 per 100,000 man-years then declined to 3.9 per 100,000 at ages above 80 years.

Figure 1
Papillary thyroid cancer incidence rates in the SEER 9 database by gender

To further evaluate differences in papillary carcinoma rates over time and by age among men and women (figure 1A–B), we stratified age-specific incidence rates by gender and calendar period of diagnosis (figure 2A–B) as well as by gender and birth cohort (figure 2C–D). Among females, period-specific age-specific rates rose rapidly until approximately ages 45–49 years and then fell (figure 2A). Among males in all time periods, age-specific rates rose among males until ages 60–65 years before declining (figure 2B). In contrast to the difference age-specific patterns by gender, the incidence rates increased rapidly with age for all but the oldest age groups among all cohorts for both females and males (figures 2C and 2D).

Figure 2
Age-specific incidence rates for females and males by calendar period and birth cohort (SEER 9: 1976–2005)

Age, period, and cohort effects by gender for papillary thyroid cancer incidence in the period 1976–2005 were further assessed with APC models. Comparing women and men, the age deviations (p for the null hypothesis of no difference = 0.39), period deviations (p for difference = 0.86), and cohort deviations (p for difference = 0.80) were similar. However, irrespective of similar gender-related deviations, the APC longitudinal age trends (LAT) and net drifts were very different among women and men, as were the “fitted” age-at-onset curves and “fitted” age-specific temporal trends (figure 3). For example, the LAT rose at a rate of 3.56% (95% CI; 3.36%, 3.76%) per year of attained age among women, whereas the LAT was 5.02% (95% CI; 4.66%, 5.44%) among men. Consequently, the APC fitted age-at-onset curve (that was adjusted for period and cohort factors) also rose more steadily with advancing age among men than women, which is consistent with the cross-sectional age-specific incidence rates that were not adjusted for period and cohort effects (figure 1B). Consequently, the female to male age-specific incidence rate ratio (IRRFM) declined steadily with age in all time periods. Indeed, IRRFM fell more than five at ages 20–24 to 3.4 at ages 35–44 and approached one at ages 80+ (figure 4). Of note, advancing age appeared to impact the cross-sectional (Figure 1B) more than the fitted age-specific rates (Figure 3A), likely due to period and/or cohort effects. This phenomenon is observed when there is progressive increase in cancer risk from one period and/or one generation to the next (20). By removing these period and/or cohort trends, the fitted curves provided estimates of the cross-sectional age-specific incidence rate curves free of secular tends.

Figure 3
APC fitted age-at-onset curves and age-specific trends for females and males
Figure 4
Age-specific female-to-male incidence rate ratios (IRRFM)

Whereas the LATs were greater for men than women, the net drifts were greater for women and men (figure 3). For example, the net drift rose at a rate of 4.2% (95% CI; 4.0%, 4.4%) per year of diagnosis among women, whereas the net drift was 3.0% (95% CI; 2.1%, 3.4%) per year of diagnosis among men. Consequently, the age-specific temporal trends rose more steadily for all age groups among women than men, which is consistent with the age-adjusted temporal trend (figure 1A). In a sensitivity analysis, we observed similar age-specific effects and secular trends among women and men with small (≤ 2.0 centimeters) and large (> 2.0 centimeters) papillary thyroid cancers, data not shown. That is, irrespective of tumor size, LATs were greater for men than women while net drifts were greater for women than men.


In the United States, thyroid cancer incidence has risen steadily for decades, with an increase of 74% from 1976–1985 to 1996–2005. The rise has been largely limited to the papillary histopathological subtype. Papillary carcinoma rates rose in all age groups except the very young, among both blacks and whites, and for all tumor sizes and all stages of disease, and these increases were more rapid among females than males. Indeed, we found statistically significant temporal and age-related differences for females and males. Gender-related differences (age-adjusted and age-specific incidence rates) were robust in age-period-cohort (APC) models that were simultaneously adjusted for age, period, and cohort effects. For example, at the end of our study period in 2005, age-adjusted incidence rates among females were approximately 3 times those among males, and the APC net drifts over the period 1976–2005 were 40% higher for females (4.2% per year) than for males (3.0% per year). Age-related gender differences were manifested as early-onset predominance among females. In sum, secular trends were greater for females (summarized in the net drifts), whereas the age-related trends were greater for males (summarized in the longitudinal age trends). All in all, these results suggest that gender is an effect modifier of papillary thyroid cancer over time and across age.

Diagnostic sensitivity and opportunities for detection have improved over the past decades with the introduction of thyroid ultrasound in the early 1980s and final-needle aspiration technology in the late 1980s; these technologies could have potentially impacted the secular trends more in one gender than the other. Compared to males, papillary thyroid cancer incidence rates started higher among younger females and rose faster, culminating in a larger net drift among females. A potential explanation for the rapid increase in incidence observed among females occurring early in life may be greater detection during annual obstetrical and gynecological examinations during the reproductive years, whereas the slower rise in incidence among males might reflect more frequent medical visits later in life. Although consensus guidelines do not advocate universal thyroid function screening during pregnancy (21, 22), thyroid testing has become routine in nearly half of prenatal care practices in some areas of the United States (23). Furthermore, while thyroid ultrasound has traditionally been performed in the radiology department, it is increasingly performed in the physician’s office. Increases in office-based use of ultrasound may increase the number of thyroid cancers identified, which may also explain some of the interaction with gender.

However, enhanced clinical detection (particularly among females) may not be the sole explanation for differential secular trends among females and males. Increased diagnostic scrutiny would enhance the detection of small tumors (3), whereas our results show significant increases for all tumor sizes. Furthermore, there was a differential gender effect among large as well as small tumors. Consequently, although the combination of the ability to detect and aspirate small nodules has clearly facilitated the diagnosis of smaller thyroid cancers, this does not seem to account for the substantial increases of larger tumors. However, a large proportion of tumor sizes were classified as unknown in the SEER 9 database and this subsequently suggests that though intriguing, these results should be reexamined in future studies with comprehensive tumor size ascertainment.

The results of our age-specific analyses also suggest that differential age-related exposures by gender should also be pursued in future analytical studies. Most notable is the female predominance, which is greatest at the younger ages and has been observed worldwide (24). It is also noteworthy that the greatest percent change in incidence over time were observed among females in the older age groups. However, the overall epidemiological evidence of the relationship between reproductive risk factors has been inconsistent and weak (2529). The age-specific female-to-male rate ratios did not show notable changes around the menopausal years that would suggest a role of changing hormonal exposures. That is, there was a constant decline in the rate ratio with age, although the ratios were similar at ages 35–39 and 40–44 (3.4) and at ages 65–69 to 75–79 (1.4 to 1.6). Possibly, a better understanding of gender-related differences in thyroid stimulating hormone, thyroid biology, and thyroid development may confer important etiological clues.

Our study was limited by the usual concerns related to analyses of registry data: non-review of histopathologic diagnoses, potential incomplete data collection, and inconsistencies in tumor classification over time due to changing staging systems. Our descriptive results are consistent with other population-based studies. However, where other studies have relied largely on standard descriptive techniques, we have supplemented secular trends and age-specific incidence patterns with both traditional and comparative APC analysis, adjusting simultaneously for age, period, and cohort effects and/or artifacts.

In sum, gender was an age-related effect modifier for papillary thyroid cancer, with differential secular trends and age-specific incidence rates among females and males. Given similar age, period, and cohort deviations, the differences between male and female thyroid cancer incidence patterns appears to be largely related to linear trends over time (net drifts) and age (longitudinal age trends). This is of special interest since these drift parameters are often considered nuisance factors when interpreting parameter estimates for a single group. The combination of differential secular and age-related linear trends for papillary thyroid cancer suggests a complex interaction between different temporal (period and/or cohort) and age-related biological factors among females and males. Future analytic studies attempting to identify the risk factors responsible for rising papillary thyroid cancer incidence should be designed with adequate power to assess age-specific interactions between females and males.


We thank the reviewers for their helpful comments that have improved the content of this manuscript. We also thank John Lahey of IMS, Inc. for figure development. Briseis Kilfoy is supported by NIH training grant TU2CA105666.


None of the co-authors has a financial conflict of interest that would have affected this research. This research was supported in part by the Intramural Research Program of the National Institutes of Health, National Cancer Institute. The authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.


1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin. 2008 Mar-Apr;58(2):71–96. Epub 2008 Feb 20. [PubMed]
2. Zheng T, Holford TR, Chen Y, et al. Time trend and age-period-cohort effect on incidence of thyroid cancer in Connecticut, 1935–1992. Int J Cancer. 1996 Aug 7;67(4):504–509. [PubMed]
3. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973–2002. JAMA. 2006 May 10;295(18):2164–2167. [PubMed]
4. Reynolds RM, Weir J, Stockton DL, Brewster DH, Sandeep TC, Strachan MW. Changing trends in incidence and mortality of thyroid cancer in Scotland. Clin Endocrinol (Oxf) 2005 Feb;62(2):156–162. [PubMed]
5. Albores-Saavedra J, Henson DE, Glazer E, Schwartz AM. Changing patterns in the incidence and survival of thyroid cancer with follicular phenotype--papillary, follicular, and anaplastic: a morphological and epidemiological study. Endocr Pathol. 2007 Spring;18(1):1–7. [PubMed]
6. Anderson WF, Chen BE, Brinton LA, Devesa SS. Qualitative age interactions (or effect modification) suggest different cancer pathways for early-onset and late-onset breast cancers. Cancer Causes Control. 2007 Dec;18(10):1187–98. Epub 2007 Sep 6. [PubMed]
7. Clayton D, Schifflers E. Models for temporal variation in cancer rates. I. Age-period and age-cohort models. Stat Med. 1987;6:449–67. [PubMed]
8. Clayton D, Schifflers E. Models for temporal variation in cancer rates. II. Age-period and age-cohort models. Stat Med. 1987;6(4):469–81. [PubMed]
9. Zhu C, Zheng T, Kilfoy BA, et al. Birth Cohort Analyses of the Incidence of Papillary Thyroid Cancer by Gender and Histological Subtypes in the United States, 1973–2004. Submitted.
10. Enewold L, Zhu K, Ron E, et al. Rising thyroid cancer incidence in the United States by demographic and tumor characteristics, 1980–2005. Submitted. [PMC free article] [PubMed]
11. Surveillance, Epidemiology, and End Results (SEER) Program. (November 2007 submission.). Public-Use Database (1973–2005), National Cancer Institute DCCPS, Suveillance Research Program, Cancer Statistics Branch, Released April 2008, based on the November 2007 submission.
12. Johnson CH, Adamo M, editors. SEER Program Coding and Staging Manual 2007. NIH Publication number 07-5581. National Cancer Institute; Bethesda, MD: 2007.
13. Surveillance Research Program, National Cancer Institute SEER*Stat software. version 6.4.4.
14. A note on the delta method by Gary W. Oehlert, The American Statistician, 46: 27–29, 1992.
15. Devesa SS, Donaldson J, Fears T. Graphical presentation of trends in rates. Am J Epidemiol. 1995 Feb 15;141(4):300–4. [PubMed]
16. Holford TR. The Estimation of Age, Period and Cohort Effects for Vital Rates. Biometrics. 1983;39:311–324. [PubMed]
17. Holford TR. Understanding the effects of age, period, and cohort on incidence and mortality rates. Annu Rev Public Health. 1991;12:425–57. Review. [PubMed]
18. Tarone RE, Chu KC. Evaluation of birth cohort patterns in population disease rates. Am J Epidemiol. 1996;143:85–91. [PubMed]
19. Anderson WF, Rosenberg PS, Menashe I, et al. Age-related crossover in breast cancer incidence rates between black and white ethnic groups. J Natl Cancer Inst. 2008;17;100(24):1804–14. [PMC free article] [PubMed]
20. Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000. The global picture. Eur J Cancer. 2001 Oct;37(Suppl 8):S4–66. Review. [PubMed]
21. Vaidya B, Anthony S, Bilous M, et al. Detection of Thyroid Dysfunction in Early Pregnancy: Universal Screening or Targeted High-Risk Case Finding? J Clin Endo Metab. 2007;92:203–207. [PubMed]
22. Surks MI, Ortiz E, Daniels GH, et al. Subclinical thyroid disease: scientific review and guidelines for diagnosis and management. JAMA. 2004;291:228–238. [PubMed]
23. Haddow J, McClain M, Palomaki G, Kloza E, Williams J. Screening for thyroid disorders during pregnancy: Results of a survey in Maine. American Journal of Obstetrics and Gynecology. 194(2):471–474. [PubMed]
24. Kilfoy BA, Zheng T, Han X, et al. International patterns and trends in thyroid cancer incidence, 1973–2002. Cancer Causes and Control. 2009 In press. [PMC free article] [PubMed]
25. Negri E, Dal Maso L, Ron E, et al. A pooled analysis of case-control studies of thyroid cancer. II. Menstrual and reproductive factors. Cancer Causes Control. 1999 Apr;10(2):143–55. [PubMed]
26. La Vecchia C, Ron E, Franceschi S, et al. A pooled analysis of case-control studies of thyroid cancer. III. Oral contraceptives, menopausal replacement therapy and other female hormones. Cancer Causes Control. 1999;10:157–66. [PubMed]
27. Memon A, Darif M, Al-Saleh K, Suresh A. Epidemiology of reproductive and hormonal factors in thyroid cancer: evidence from a case-control study in the Middle East. Int J Cancer. 2002;97:82–9. [PubMed]
28. Zivaljevic V, Vlajinac H, Jankovic R, et al. Case-control study of female thyroid cancer--menstrual, reproductive and hormonal factors. Eur J Cancer Prev. 2003;12:63–6. [PubMed]
29. Goodman MT, Kolonel LN, Wilkens LR. The association of body size, reproductive factors and thyroid cancer. Br J Cancer. 1992;66:1180–4. [PMC free article] [PubMed]