A Large Cohort Study of Hypothyroidism and Hyperthyroidism in Relation to Gynecologic Cancers

Table 2

Hazard ratios (95% confidence intervals) of endometrial cancer by self-reported history of hypothyroidism and hyperthyroidism status in the combined Nurses’ Health Study (NHS, 1976–2010) and Nurses’ Health Study II (NHSII, 1989–2011).

Endometrial cancer	History of hypothyroidism			History of hyperthyroidism
	No	Yes			No	Yes
Number of cases	1176	123			1176	15
Person-years	3,857,750	267,130			3,857,750	24,091
Age-adjusted	1 (ref)	1.19 (0.98, 1.43)			1 (ref)	1.46 (0.87, 2.44)
Multivariable-adjusted*	1 (ref)	0.91 (0.75, 1.10)			1 (ref)	1.43 (0.85, 2.39)
Multivariable-adjusted^†	1 (ref)	0.89 (0.74, 1.08)			1 (ref)	1.41 (0.84, 2.37)
Multivariable-adjusted^‡	1 (ref)	0.79 (0.62, 1.00)			—	—
Multivariable-adjusted^∣∣, BMI < 25 kg/m^2	1 (ref)	1.08 (0.76, 1.55)			—	—
Multivariable-adjusted^∣∣, BMI ≥ 25 kg/m^2	1 (ref)	0.87 (0.69, 1.09)			—	—
	No	Yes <8 years^¶	Yes 8+ years^¶	𝑃 trend	No	Yes <6 years^¶	Yes 6+ years^¶	𝑃 trend
Number of cases	1176	56	67		1176	6	9
Person-years	3,857,750	124,990	142,140		3,857,750	12,366	11,725
Age-adjusted	1 (ref)	1.33 (1.02–1.75)	1.09 (0.85–1.40)	0.39	1 (ref)	1.18 (0.52–2.64)	1.74 (0.89–3.38)	0.09
Multivariable-adjusted*	1 (ref)	1.06 (0.81–1.40)	0.81 (0.63–1.04)	0.11	1 (ref)	1.16 (0.51–2.61)	1.69 (0.86–3.30)	0.12
Multivariable-adjusted^†	1 (ref)	1.03 (0.78–1.35)	0.80 (0.62–1.03)	0.08	1 (ref)	1.14 (0.51–2.57)	1.68 (0.86–3.28)	0.12
Multivariable-adjusted^‡	1 (ref)	0.82 (0.54–1.27)	0.78 (0.59–1.03)	0.04	—	—	—	—
Multivariable-adjusted^∣∣, BMI < 25 kg/m^2	1 (ref)	1.24 (0.76–2.03)	0.96 (0.59–1.57)	0.95	—	—	—	—
Multivariable-adjusted^∣∣, BMI ≥ 25 kg/m^2	1 (ref)	1.03 (0.74–1.43)	0.77 (0.57–1.04)	0.09	—	—	—	—

∗Multivariable core model is stratified by age (in months), two-year cycle at risk, and cohort (NHS/NHSII) and is adjusted for body mass index (kg/m²; continuous linear variable and missing data indicator), cigarette smoking (continuous linear variable of pack-years of smoking and missing data indicator), family history of uterine cancer (yes/no), diabetes (yes/no), parity (continuous linear variable of number of children), oral contraceptive use history (never, <1, 1–4, 5–9, 10+ years of use, missing data indicator), age at last birth (30 or greater yes/no and missing data indicator), age at menarche (continuous linear variable and missing data indicator), age at menopause (continuous linear variable and missing data indicator), use of estrogen only postmenopausal hormone therapies (never user, past user with <5 years duration, past user with ≥5 years duration, current user with <5 years duration, current user with ≥5 years duration), use of estrogen plus progestin only postmenopausal hormone therapy (never user, past user with <5 years duration, past user with ≥ 5 years duration, current user with <5 years duration, current user with ≥5 years duration) and use of other postmenopausal hormone therapy (never user, past user with <5 years duration, past user with ≥5 years duration, current user with <5 years duration, current user with ≥5 years duration), and infertility (none, self is infertile, spouse is infertile, missing).
†To core model, additionally adjusted for greater health surveillance that may be associated with thyroid screening (report of physical exam and high cholesterol).
‡Hypothyroidism defined as being reported at least twice during followup; adjusted for the same variables as in core model. Not enough cases were available for analyses of the alternative definition for hyperthyroidism.
∣∣Stratified analyses by body mass index (BMI) categories defined by the median (25 kg/m²); BMI is a strong risk factor for both hypothyroidism and endometrial cancer. 𝑃 for interaction with BMI was 0.15 for hypothyroidism status and 0.23 for duration of having a history of hypothyroidism.
¶Duration of having a history of hypothyroidism (median = 8 years) or hyperthyroidism (median = 6 years).