The Association Between Hypothyroidism and Cognitive Function Change in Women across the Menopause Transition: The Study of Women’s Health Across the Nation

Abstract

Background:

Patients treated for hypothyroidism with levothyroxine (LT4) monotherapy may present with persistent hypothyroidism symptoms, including cognitive symptoms, despite having a normal thyroid stimulating hormone (TSH) level. It remains unclear whether LT4 monotherapy is sufficient to normalize cognitive function outcomes over time.

Methods:

This is a multisite longitudinal study of a diverse group of women during midlife representing 5 ethnic/racial groups from 7 enrollment sites across the United States in the Study of Women’s Health Across the Nation. Women were screened for a history of thyroid disease and the use of LT4. The study consisted of two primary groups: women with LT4-treated hypothyroidism and control women without thyroid disease. Each participant completed up to 9 cognitive assessments over the study period testing processing speed, working memory, and episodic memory (immediate and delayed recall). Multivariable generalized linear mixed models of scores for each cognitive assessment were developed to determine the association between LT4-treated hypothyroidism and cognitive function trajectories. Covariates included sociodemographic, clinical characteristics, and menopausal status (pre/early peri, late peri, and surgical/post). Sensitivity analyses were conducted to assess the impact of abnormal TSH levels and practice effects (i.e., improvements in scoring after repeated testing).

Results:

Of the 2033 women who were included in the study, 227 (11.2%) met criteria for LT4-treated hypothyroidism. At baseline, both processing speed and working memory scores were higher in LT4-treated women (mean processing speed scores: 56.5 vs 54.4; p value = 0.006; mean working memory scores: 6.8 vs 6.4; p value = 0.018). However, when considering the effect of LT4-treated hypothyroidism over time, there were no significant differences in the rate of cognitive decline (in any measure) between the hypothyroidism and control groups with or without covariate adjustment. The results were similar when considering LT4-treated women with abnormal TSH levels or after minimizing practice effects.

Conclusions:

We observed no difference in cognitive decline between women with LT4-treated hypothyroidism and women without thyroid disease. For similar aged patients with cognitive complaints, if thyroid function testing is normal, clinicians should consider causes other than inadequate thyroid hormone treatment to explain these symptoms.

Introduction

Hypothyroidism is a metabolic disorder defined by low thyroid hormone levels that affects nearly 10% of the US population.^1,2 For many years, hypothyroidism was treated with replacement thyroid hormones—both thyroxine (T4) and triiodothyronine (T3)—mostly derived from animal thyroid glands.³ Treatment shifted when it was discovered that T4 is converted to T3 in the peripheral tissues via deiodination, thus providing a rationale for treatment with T4 alone.^4,5 As a result, levothyroxine (LT4) became the primary treatment for hypothyroidism and is now one of the most commonly prescribed medications in the United States.^6,7 Currently, conventional wisdom holds that patients with primary hypothyroidism taking LT4 daily with a normal thyroid stimulating hormone (TSH) level are adequately treated and should be symptom-free. However, the prevalence of patient dissatisfaction and persistent hypothyroid symptoms, including perceived cognitive deficits, with LT4 treatment^8,9 has led some experts to question the current paradigm.^10
–12

Clinicians are now faced with the difficult task of determining if these symptoms are related to inadequately treated hypothyroidism (despite a normal TSH level)—or something else. In particular, persistent symptoms related to cognitive function are commonly reported, with a decline in function that is not fully resolved with LT4 treatment.⁹ However, the effects of aging, comorbidities, and other clinical factors must be included when considering the change of cognitive function over time.¹³ A study evaluating cognitive function over time in LT4-treated individuals and their counterparts without thyroid disease can provide insight into the effectiveness of LT4 to treat the cognitive effects of hypothyroidism.

Several studies have sought to examine the differences in cognitive function in patients with treated hypothyroidism and healthy controls using cognitive and neuropsychiatric testing, with conflicting results.^{14

–25} Some studies have observed deficits in verbal memory and attention-heavy tasks,^16,17 but no clear symptom profile has emerged. Many of these studies have important caveats, including relatively young subjects, shorter periods of observation, or only measure cognitive function at a single time point. The Study of Women’s Health Across the Nation (SWAN) is a longitudinal, multidisciplinary study designed to examine key health changes in women from midlife into early old age. SWAN has identified several clinical factors associated with changes in cognitive function trajectories over time, including menopause symptoms,²⁶ sex hormone replacement,²⁷ metabolic syndrome,²⁸ and cardiovascular risk.^29,30 Using observations from SWAN, we have sought to compare the cognitive function trajectories in women with and without hypothyroidism in three key domains: (1) processing speed, (2) working memory, and (3) episodic memory; and we aim to compare scores of self-reported cognitive decline, assessed at the last visit of the study. Through this investigation, we aim to determine if LT4-treated hypothyroidism is associated with worse cognitive function decline in women at midlife and during menopause transition.

Materials and Methods

Study participants and design

This is a prospective cohort study comparing change in cognitive function over time between women with treated hypothyroidism and women without thyroid disease. We have analyzed data collected from SWAN, a longitudinal study of health outcomes in women throughout midlife. The initial SWAN cohort included 3302 women representing 5 ethnic/racial groups from 7 enrollment sites across the US: Black (Boston, MA; Chicago, IL; Detroit, MI area; Pittsburgh, PA), Chinese (Oakland, CA area), Hispanic (Newark, NJ), Japanese (Los Angeles, CA), and non-Hispanic White (all sites). Details on the overall study design, data collection, sampling, and recruitment have been published previously.³¹ Eligibility criteria at the time of enrollment (1996–1997) included 42–52 years old; having an intact uterus and at least 1 ovary; not pregnant, breastfeeding, or lactating; not using oral contraceptives or sex hormone replacement therapy; and reporting at least 1 menstrual cycle during the prior 3 months. This investigation includes data collected from the baseline assessment through the follow-up visits (1997–2017). Cognitive function was first tested at follow-up visit 4 (2000–2002), with 2429 women completing cognitive testing according to the study protocol. These women were eligible for study inclusion. Cognitive testing continued through follow-up visit 15.

Determination of hypothyroidism treatment status and exclusion criteria

All participants were divided into two study groups: those with LT4-treated hypothyroidism and those with no history of thyroid disease. At baseline, study participants were asked to report if they had ever been diagnosed with thyroid disease or had thyroid surgery. At each follow-up visit, it was documented if they had been diagnosed since the last follow-up visit. Each participant also brought in their prescription medication bottles at each visit, which were reviewed by study personnel. Women who reported a history of thyroid disease or thyroid surgery prior to the first cognitive assessment while taking LT4 were eligible to be included in the LT4-treated hypothyroidism group. Women who never reported thyroid disease, thyroid surgery, or LT4 use during the study period were eligible for the control group.

Several exclusion criteria were implemented to ensure that women in the hypothyroidism group were treated with LT4 throughout the study period and that no women in the control group developed thyroid disease during the study period (no group cross-over). Any woman who reported a history of thyroid disease but reported taking LT4 for one or zero follow-up visits was deemed unlikely to have hypothyroidism and excluded. Women who did not report a history of thyroid disease but had an abnormal TSH level (<0.4 or >4.5 mIU/L) during the study period were excluded due to possibility of having untreated thyroid disease. Those with a history of thyroid cancer were excluded due to the possibility of intentionally suppressed TSH levels. Those taking other forms of thyroid hormone (e.g., liothyronine, desiccation thyroid extract) were also excluded. Because cognitive testing did not begin until follow-up visit 4, these criteria ensure that all women included in the study with treated hypothyroidism were taking LT4 for at least 3 years prior to study start. Finally, those who reported having a stroke prior to the first cognitive assessment were excluded, and those who reported having a stroke during the study period were right-censored at that time.

Evaluation of cognitive outcomes

Cognitive function was assessed under a standard protocol in three domains: processing speed, working memory, and episodic memory. Processing speed was assessed with the symbol digit modalities test (SDMT), which involved using a model to match as many unfamiliar symbols with digits as possible in 90 seconds.³² Working memory was assessed with digit span backward (DSB), which involved listening to and repeating back strings of 2–7 digits in reversed order.³³ Scoring was based on the number of correct strings. Episodic memory was assessed with the East Boston memory test (EBMT), which involved two variations (immediate and delayed recall).^34,35 Participants were asked to recall up to 12 ideas contained within the East Boston 36-word paragraph (10 minutes later for delayed recall). At follow-up visit 15, participants completed the ECog-12, a 12-item Likert-scale questionnaire designed to detect cognitive and functional decline over the last 10 years.³⁶ Translated tests were available if the participant indicated a preference for a language other than English (Cantonese, Japanese, or Spanish).

Sociodemographic and clinical covariates

To compare and account for anticipated differences between women with and without hypothyroidism, sociodemographic and clinical characteristics were evaluated. Sociodemographic covariates included age, race/ethnicity (linked to recruitment site as outlined above), education level, and financial strain (“difficulty paying for basics”). Clinical covariates, which were treated as time-varying in the analysis, included smoking status, body mass index (BMI), sex hormone use (estrogen and/or progesterone), number of comorbidities (myocardial infarction or angina, diabetes mellitus, osteoarthritis, hypertension, hyperlipidemia, osteoporosis), and menopausal status (peri/early perimenopause, late perimenopause, postmenopause/surgical menopause). TSH levels were also collected twice during the study period (follow-up visits 4 and 10) but were not included as a covariate in model development because TSH level was part of the defining criteria for hypothyroidism treatment status. In addition, the LT4 treatment group was subdivided into a controlled treatment group (normal TSH levels) and uncontrolled treatment group ( $\geq$ 1 abnormal TSH level as defined by TSH >4.5 mIU/L or <0.4 mIU/L) for the sensitivity analysis.

Statistical analysis

Descriptive statistics were run to examine all variables and describe the study population. Mean (SD) or median (IQR) was used for continuous variables and N (%) for binary/categorical variables. All variables were checked for normality or possible outliers. There was no need to transform any of the outcomes. Processing speed (SDMT) and working memory (DSB) raw scores were standardized to z-scores (by subtracting the baseline mean and dividing by the baseline standard deviation of the respective variable). Those scores were then analyzed longitudinally using mixed effects linear regression models with a random intercept and random slope to allow each woman to have her own starting point and slope over time. Episodic memory scores (EBMT) were reversed (so that the score was the number of mistakes made) and were analyzed longitudinally using Poisson regression, due to the nature of the distribution. Generalized linear mixed models using the maximum likelihood method were used for these outcomes. An advantage of this approach is that model construction is robust to missing data (which was rare). No imputation of missing data was required for model convergence. Self-reported cognitive decline (ECog-12) scores were analyzed cross-sectionally using linear regression. Time-dependent thyroid disease (i.e., when the subject reported a history of thyroid disease + LT4 use) was the primary predictor in all longitudinal models, and geographical site was included in the random effects with each participant nested within geographical site. The base model (model 0) included time-dependent thyroid disease (as defined above), time (since 1^st cognitive assessment), time (since 1^st cognitive assessment)*thyroid disease, and age at analytic baseline. Model 1 was an extension of model 0 + baseline sociodemographic variables (listed above) + time-dependent menopausal status. Model 2 was an extension of model 1 + physical and behavioral risk factors.

In addition, we performed two sensitivity analyses to evaluate the impact of having an abnormal TSH with LT4 treatment and to minimize practice effects related to the cognitive tests. We hypothesized that it would be possible that LT4-treated women with documented TSH levels outside the normal range during follow-up visits 4 and/or 10 may experience worse cognitive outcomes due to periods of overtreatment or undertreatment. Thus, a sensitivity analysis was completed that included only women with at least 1 abnormal TSH (<0.5 or >4.5 mIU/L) in the LT4-treated hypothyroidism group. In addition, to account for improvements in cognitive tasks with practice that have been observed in the SWAN cohort,³⁷ a sensitivity analysis was completed in which the first two cognitive assessments were excluded from the analysis. All analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC). P-values <0.05 were considered statistically significant.

Results

Of the total number of SWAN participants who completed baseline cognitive testing (n = 2429), a total of 2033 women comprised the two primary study groups (79.5%) (Figure 1). There were several differences between the included (n = 2033) and excluded (n = 1269) populations, including a higher proportion of Asian women (20% vs 9.9%), higher education levels (44.9% vs 37.8% with a college degree), and more financial security (37.0% vs 44.2% with difficulty paying for basics) in the included group. Of those included women, 227 (11.2%) met criteria for LT4-treated hypothyroidism, while the remainder (n = 1806; 88.8%) formed the control group. Two women had undergone thyroidectomy at baseline. In comparison, the treated hypothyroidism group had a greater proportion of non-Hispanic White women compared with the control population (59.9% vs 45.1%; p value <0.001). However, age (49.8 vs 50.0 years), education (50.7 vs 45.4% with college degree), and financial strain (37.2% vs 37.0% with difficulty paying for basics) were not significantly different between the two groups. The mean BMI (30.1 vs 28.5 kg/m²; p = 0.006) was higher in the treated hypothyroidism group, as was the proportion who had pre or early perimenopausal status at analytic baseline (54.2% vs 49.8%; p = 0.010). There was no difference between smoking status, number of comorbidities, and estrogen/progesterone use between the groups. The mean duration of follow-up for all participants was 13.0 years. The mean follow-up was 13.1 years for the hypothyroidism group and 12.9 years for the control group. A total of 1394 women (69% of the total study cohort) remained in the study through follow-up visit 15. Notably, of the treated hypothyroidism group, 105 (46.2%) were classified as controlled treatment (no abnormal TSH levels) and 122 (53.8%) were classified as uncontrolled treatment ( $\geq$ 1 abnormal TSH level).

FIG. 1.

Study flowchart and exclusion criteria. The dotted lines represent the subdivision of the LT4-treated group for the sensitivity analysis of women with uncontrolled treatment. Women with at least 1 abnormal TSH (<0.5 or >4.5 mIU/L) were considered uncontrolled. Of note, only 3 women excluded for untreated thyroid disease had two TSH levels >7.5 mIU/L. SWAN, Study of Women’s Health Across the Nation; LT4, levothyroxine.

Baseline cognitive function testing

At cognitive baseline, women with LT4-treated hypothyroidism exhibited higher processing speed scores, as measured by SDMT, with to control women (mean SDMT score: 56.5 vs 54.4; p value = 0.006) (Table 1). Similarly, working memory, as measured by DSB, was higher in those with LT4-treated hypothyroidism versus those without (mean DSB score: 6.8 vs 6.4; p value = 0.018). Episodic memory scores (both immediate and delayed recall as measured by EBMT) were not significantly different.

Table 1.

Baseline Characteristics and Cognitive Scores of those with LT4-Treated Hypothyroidism and the Control Population (without Thyroid Disease)

Participant characteristics	All participants(n = 2033)	Treated hypothyroidism(n = 227)	Control population(n = 1806)	p value
Age, years, mean (SD)	50.0 (2.7)	49.8 (2.6)	50.0 (2.7)	0.247
Race/Ethnicity, n (%)				0.0001
Non-Hispanic White	950 (46.7)	136 (59.9)	814 (45.1)
Black	579 (28.5)	49 (21.6)	530 (29.4)
Hispanic	98 (4.8)	7 (3.1)	91 (5.0)
Chinese	184 (9.1)	23 (10.1)	161 (8.9)
Japanese	222 (10.9)	12 (5.3)	210 (11.6)
Education level, n (%)				0.505
High school or less	440 (21.8)	46 (20.4)	394 (22.0)
Some College	650 (32.2)	65 (28.9)	585 (32.7)
College degree	440 (21.8)	54 (24.0)	386 (21.6)
Post College	486 (24.1)	60 (26.7)	426 (23.8)
Difficulty paying for basics, n (%)				0.940
Not at all	1254 (63.0)	135 (62.8)	1119 (63.0)
Very/Somewhat	736 (37.0)	80 (37.2)	656 (37.0)
Current smoker, n (%)	267 (13.5)	28 (12.5)	239 (13.6)	0.679
BMI, kg/m², mean (SD)	28.7 (7.3)	30.1 (8.2)	28.5 (7.2)	0.006
Number of comorbidities, mean (SD)	1.1 (1.1)	1.3 (1.2)	1.1 (1.1)	0.247
Hormone use, n (%)	408 (20.1)	44 (19.4)	364 (20.2)	0.861
Menopausal status, n (%)				0.010
Pre/Early peri	1022 (50.3)	123 (54.2)	899 (49.8)
Late peri	199 (9.8)	31 (13.7)	168 (9.3)
Post/Surgical	535 (26.3)	41 (18.1)	494 (27.4)
Undetermined	276 (13.6)	32 (14.1)	244 (13.5)
Thyroid function tests, median (IQR)
TSH (mIU/L), follow-up visit 4	1.9 (1.4, 2.7)	2.9 (1.2, 6.0)	1.9 (1.4, 2.6)	<0.001
TSH (mIU/L), follow-up visit 10	1.8 (1.2, 2.5)	2.2 (0.9, 4.1)	1.8 (1.3, 2.4)	0.009
Cognitive assessment scores
SDMT, mean (SD)	54.6 (11.7)	56.5 (10.6)	54.4 (11.8)	0.006
DSB, mean (SD)	6.5 (2.3)	6.8 (2.5)	6.4 (2.3)	0.018
EBMT immediate recall: total ideas, median (IQR)	10.0 (9.0,12.0)	10.0 (10.0,12.0)	10.0 (9.0,12.0)	0.071
EBMT delayed recall: total ideas, median (IQR)	10.0 (9.0,11.0)	10.0 (9.0,12.0)	10.0 (9.0,11.0)	0.100

In this analysis, the time-sensitive characteristics above (e.g., age, menopausal status, etc.) were determined at follow-up 4, which was the first follow-up during which the cognitive testing was performed.

LT4, levothyroxine; BMI, body mass index; TSH, thyroid stimulating hormone; IQR, interquartile range; SDMT, symbol digit modalities test; DSB, digit span backward; EBMT, East Boston Memory Test.

Treated hypothyroidism and change in cognitive function over time

Scores of each cognitive test were modeled to examine the effects of treated hypothyroidism, time (as measured by study years), and an interaction term representing the effect of treated hypothyroidism on the rate of decline in cognitive function). In the unadjusted longitudinal model of processing speed (as measured by SDMT), LT4-treated hypothyroidism had a positive effect estimate of 0.173 $\pm$ 0.066 (p value = 0.009) (Table 2). Over the study period, annual decline of processing speed was −0.013 $\pm$ 0.003 (p value <0.001). However, the effect of LT4-treated hypothyroidism on the decline in cognitive function was not significant (−0.003 $\pm$ 0.003; p value = 0.294). Similar results were seen in models adjusted for sociodemographic and clinical covariates and menopausal status.

Table 2.

Change in Processing Speed (SDMT) over Time in Women with LT4-Treated Hypothyroidism and the Control Population (without Thyroid Disease)

	Unadjusted model			^aAdjusted model 1			^bAdjusted model 2
Effects	Coef	SE	p value	Coef	SE	p value	Coef	SE	p value
Intercept	2.174	0.385	<0.001	2.922	0.325	<0.001	2.852	0.332	<0.001
HT	0.173	0.066	0.009	0.133	0.057	0.019	0.141	0.057	0.013
Time	−0.013	0.003	<0.001	−0.017	0.003	<0.001	−0.014	0.057	<0.001
HT*time	−0.003	0.003	0.294	−0.003	0.003	0.354	−0.003	0.003	0.342

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, and menopausal status.

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, menopausal status, BMI, hormone use, smoking status, and number of comorbidities.

HT*time represents the interaction term between HT and time.

SDMT, symbol digit modalities test; Coef, coefficient; SE, standard error; HT, hypothyroidism; BMI, body mass index.

Similar to the processing speed outcome, in the unadjusted model of working memory (DSB), LT4-treated hypothyroidism had a positive effect estimate (0.159 $\pm$ 0.065; p = 0.015) (Table 3). In contrast, working memory scores did not decline significantly with time, and there was no significant effect of LT4-treated hypothyroidism over time. Neither treated hypothyroidism, time, nor treated hypothyroidism over time were estimated to have significant effects on working memory in the adjusted models. Similarly, no significant estimated effects were observed in unadjusted and adjusted models of episodic memory (Table 4).

Table 3.

Change in Working Memory (DSB) over Time in Women with LT4-Treated Hypothyroidism and the Control Population (without Thyroid Disease)

	Unadjusted model			^aAdjusted model 1			^bAdjusted model 2
Effects	Coef	SE	p value	Coef	SE	p value	Coef	SE	p value
Intercept	0.819	0.360	0.023	1.380	0.330	<0.001	1.167	0.343	<0.001
HT	0.159	0.065	0.015	0.097	0.061	0.110	0.103	0.061	0.093
Time	−0.0003	0.004	0.946	−0.004	0.004	0.326	−0.002	0.004	0.612
HT*time	−0.0008	0.004	0.843	−0.002	0.004	0.614	−0.001	0.004	0.752

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, and menopausal status.

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, menopausal status, BMI, hormone use, smoking status, and number of comorbidities.

HT*time represents the interaction term between HT and time.

DSB, digit span backward; Coef, coefficient; SE, standard error; HT, hypothyroidism; BMI, body mass index.

Table 4.

Change in Episodic Memory (EBMT, Immediate and Delayed Recall) over Time in Women with LT4-Treated Hypothyroidism and the Control Population (without Thyroid Disease)

	Unadjusted model			^aAdjusted model 1			^bAdjusted model 2
Effects	Coef	SE	p value	Coef	SE	p value	Coef	SE	p value
Immediate recall
Intercept	0.259	0.304	0.396	0.060	0.278	0.830	0.203	0.290	0.485
HT	−0.060	0.060	0.315	0.011	0.057	0.846	0.012	0.057	0.829
Time	−0.012	0.005	0.020	−0.008	0.006	0.156	−0.010	0.006	0.070
HT*time	−0.006	0.005	0.282	−0.004	0.006	0.463	−0.004	0.006	0.499
Delayed recall
Intercept	0.148	0.286	0.605	0.004	0.263	0.988	0.223	0.275	0.417
HT	−0.085	0.056	0.129	−0.022	0.053	0.684	−0.021	0.053	0.701
Time	−0.009	0.005	0.058	−0.007	0.005	0.182	−0.009	0.005	0.084
HT*time	−0.005	0.005	0.358	−0.005	0.005	0.348	−0.004	0.005	0.421

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, and menopausal status.

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, menopausal status, BMI, hormone use, smoking status, and number of comorbidities.

HT*time represents the interaction term between HT and time.

EBMT, East Boston Memory Test; Coef, coefficient; SE, standard error; HT, hypothyroidism; BMI, body mass index.

Self-reported cognitive decline scores and secondary analysis

Although examined cross-sectionally, unadjusted and adjusted regression models were constructed to estimate the effect of LT4-treated hypothyroidism on ECog-12 scores from follow-up visit 15. There were no statistically significant associations between LT4-treated hypothyroidism and self-reported cognitive decline (Table 5).

Table 5.

Self-Reported Cognitive Decline in Women with LT4-Treated Hypothyroidism and the Control Population (without Thyroid Disease)

	Unadjusted model			^aAdjusted model 1			^bAdjusted model 2
Effects	Coef	SE	p value	Coef	SE	p value	Coef	SE	p value
Intercept	1.198	0.261	0.004	1.238	0.261	0.003	1.300	0.275	0.003
HT	0.025	0.034	0.463	0.020	0.034	0.553	0.004	0.034	0.915

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, and menopausal status.

Model adjusted for age at baseline, race/ethnicity, education level, financial hardship, menopausal status, BMI, hormone use, smoking status, and number of comorbidities.

Self-reported cognitive function was only assessed at follow-up visit 15.

Coef, coefficient; SE, standard error; HT, hypothyroidism; BMI, body mass index.

Sensitivity analyses

For the first sensitivity analysis, new models were constructed for each cognitive measure with the uncontrolled hypothyroidism treatment group serving as the experimental group (the control group was unchanged). In this group, 90 women had at least one high (>4.5 mIU/L) TSH level (median 6.7 mIU/L), and 32 women had at least one low (<0.4 mIU/L) TSH level (median 0.1 mIU/L). No significant differences were observed in the change in cognitive function scores over time in any model (data not shown). Similarly, in the second sensitivity analysis in which the first two cognitive assessments were excluded to minimize practice effects, no significant effects of LT4-treated hypothyroidism on the change in cognitive function over time were observed in any model.

Discussion

In this longitudinal study of cognitive function in women across the menopausal transition, we did not identify an association between LT4-treated hypothyroidism and change in processing speed, working memory, and episodic memory over time. In the cross-sectional analysis, there was no difference in self-reported cognitive decline between those with treated hypothyroidism and the control group at the last study visit. The results were consistent before and after adjustment with clinical and sociodemographic covariates. The results were also consistent after sensitivity analyses to account for uncontrolled thyroid hormone levels and practice effects. Of note, LT4-treated hypothyroidism was associated with improved processing speed and working memory scores at cognitive baseline. The effect persisted in the adjusted longitudinal models for processing speed. However, at cognitive baseline, the score difference did not meet a reported benchmark for clinically significant cognitive decline ( $\geq$ 4 points).³⁸

This study is unique in that it examines cognitive function in a relatively large (n = 227) sample of LT4-treated women over a period of 15 years. Due to the timing of the first cognitive tests in the SWAN cohort, all women in the hypothyroidism group were treated with LT4 at least 3 years prior to when this study began, ensuring a prolonged lead-in treatment period. Another strength was the relative lack of multimorbidity in both study groups, which has been identified as a potential confounder in LT4-treated individuals with persistent symptoms.⁹ Our study complements studies examining cognitive function before treatment and in the immediate posttreatment period (<12 months).^15,18,23 Our results are in line with several smaller cross-sectional studies which found that cognitive function in LT4-treated hypothyroidism is not significantly impaired, supporting the hypothesis that LT4 treatment is not associated with residual cognitive deficits attributable to hypothyroidism.^15,22
–24

It is important to recognize that the cognitive assessments included in the SWAN study were chosen as efficient, repeatable measures and indicators of early cognitive aging and maintenance of independence.^39
–41 As a result, the specificity of these tests to evaluate cognitive deficits associated with hypothyroidism must be considered. Memory deficits (specifically retrieval and recall) are the most consistently observed cognitive symptoms in hypothyroidism.^{15,18,42
–44} Studies of functional MRI and PET imaging show decreased hippocampal volume and decreased blood flow in regions of the brain responsible for information processing and working memory.^45

–49 Therefore, it is reasonable to expect that the cognitive testing conducted during SWAN, which emphasizes working and episodic memory, would be sensitive to detect cognitive deficits related to hypothyroidism.

Our findings agree with the study of long-term treatment effects by Quinque et al,¹⁹ in which 18 patients with hypothyroidism with a mean LT4 treatment duration of 4 years were recruited for neurocognitive testing and functional MRI (fMRI) imaging. Outcomes in verbal memory, working memory, psychomotor speed, sustained attention, and fMRI findings were similar between those patients and healthy controls. It is notable that the treated patients were relatively young (mean age 32 years). An analysis of an older hypothyroid population (mean age 74 years) from the Rancho Bernardo Study who were treated with LT4 for an average of 20 years did not show any cognitive function deficits on verbal fluency or the mini-mental state examination.²⁴

Multiple studies, however, have detected residual cognitive deficits in LT4-treated patients in some (but not all) neurocognitive measurements.^{15

–18,21,25} Key differences in participant characteristics (such as age or presence of thyroid autoimmunity), study design, methodologies, and outcome measures make it difficult to reach a conclusion on the effectiveness of LT4 to fully restore cognitive function. Irrespective of the literature findings, many patients have sought alternative treatment approaches, including desiccated thyroid extract.⁵⁰ This has led the major thyroid societies to call for additional investigations of patient outcomes with LT4 monotherapy and combination therapy, specifically in patients dissatisfied with their current therapy.^10,51 Approaches to navigating the evaluation and treatment of hypothyroid patients who are experiencing persistent symptoms have been developed, which include both exploring alternative causes and therapies (e.g., combination therapy).^52,53 Ultimately, our findings suggest that cognitive function decline during midlife is no worse for a woman treated for hypothyroidism than for a woman without thyroid disease. Clinicians should resist anchoring on inadequate treatment of hypothyroidism as the cause of these symptoms and may investigate other disease processes (e.g., iron deficiency, B12 deficiency, sleep apnea, celiac disease). Other healthcare-centric causes of patient dissatisfaction (e.g., unrealistic treatment expectations, somatic symptom disorder, Type D personality) that are more commonly seen in patients with hypothyroidism may be considered, as well.^9,52,54

This study is not without limitations. As mentioned above, the cognitive assessments included as outcomes in this study were not intended to be a comprehensive evaluation of all aspects of cognitive function. More intensive neuropsychiatric testing may have increased sensitivity for more subtle cognitive changes.^55,56 This may explain why no significant differences were observed over time between the study groups. Change over time in cognitive skills other than processing speed, working memory, and episodic memory may be different between LT4-treated women and controls. This offers a possible explanation for the unexpected finding that women with uncontrolled thyroid hormone levels faired no worse than controls. Second, if there is a true yet small effect of LT4-treated hypothyroidism on cognitive outcomes, the analysis may not have been adequately powered to detect it. However, this is a relatively large cohort compared to prior investigations, taking repeated outcome measures enhances statistical power,⁵⁷ and nearly all observed effect coefficients were close to zero. It is possible that the higher education levels achieved in the study population (vs excluded women) may have served as a protective effect against cognitive decline in both the hypothyroidism and control populations, biasing toward a null result. However, there is conflicting evidence on whether education level is associated with the rate of cognitive decline, which was assessed in this study.^58,59 Third, while it was known which women underwent thyroidectomy, tests of thyroid autoimmunity were not available. Thus, we cannot determine if the presence of thyroid antibodies (in either study group) impacted cognitive outcomes. In addition, TSH levels at the time of hypothyroidism diagnosis were not available; thus, overdiagnosis in some women is possible,² which could contribute to the null findings. However, those women with abnormal TSH levels during the study period are likely to have true hypothyroidism, and no differences were found between this group and controls, as well. Finally, given the study design, we cannot completely rule out the effect of confounding covariates (measured or unmeasured) on the study outcomes.

Conclusion

To summarize, we found no association between treated hypothyroidism and change in processing speed, working memory, or episodic memory across the menopause transition and into early old age. With repeated measures, the cognitive function of women with treated hypothyroidism declined at a similar rate to those without thyroid disease. In addition, at the final study visit, self-reported cognitive function scores were no different between groups. These findings provide evidence that LT4 monotherapy for hypothyroidism is adequate to maintain cognitive function over time on par with those without thyroid disease. In LT4-treated patients with persistent cognitive symptoms, clinicians should not anchor on inadequate treatment as the only explanation.

Footnotes

Authors’ Contributions

Conceptualization, methodology, analysis, and writing—original draft by M.D.E. and K.K. Conceptualization, methodology, review of analysis, review and editing of this article by I.J. Conceptualization and editing of this article by S.M.B. and L.H.P.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The Study of Women’s Health Across the Nation (SWAN) has grant support from the National Institutes of Health (NIH), DHHS, through the National Institute on Aging (NIA), the National Institute of Nursing Research (NINR) and the NIH Office of Research on Women’s Health (ORWH) (Grants U01NR004061; U01AG012505, U01AG012535, U01AG012531, U01AG012539, U01AG012546, U01AG012553, U01AG012554, U01AG012495, and U19AG063720). The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIA, NINR, ORWH or the NIH.

References

Aoki

, Belin

, Clickner

, et al. Serum TSH and total T4 in the United States population and their association with participant characteristics: National Health and Nutrition Examination Survey (NHANES 1999–2002). Thyroid, 2007; 17(12):1211–1223; doi: 10.1089/thy.2006.0235

Brito

, Ross

, El Kawkgi

, et al. Levothyroxine use in the United States, 2008–2018. JAMA Intern Med, 2021; 181(10):1402–1405; doi: 10.1001/jamainternmed.2021.2686

McAninch

, Bianco

. The history and future of treatment of hypothyroidism. Ann Intern Med, 2016; 164(1):50–56; doi: 10.7326/M15-1799

Hennessey

. The emergence of levothyroxine as a treatment for hypothyroidism. Endocrine, 2017; 55(1):6–18; doi: 10.1007/s12020-016-1199-8

Luongo

, Dentice

, Salvatore

. Deiodinases and their intricate role in thyroid hormone homeostasis. Nat Rev Endocrinol, 2019; 15(8):479–488; doi: 10.1038/s41574-019-0218-2

Rodriguez-Gutierrez

, Maraka

, Ospina

, et al. Levothyroxine overuse: Time for an about face? Lancet Diabetes Endocrinol, 2017; 5(4):246–248; doi: 10.1016/S2213-8587(16)30276-5

Johansen

, Marcinek

, Doo Young Yun

. Thyroid hormone use in the United States, 1997–2016. J Am Board Fam Med, 2020; 33(2):284–288; doi: 10.3122/jabfm.2020.02.190159

Mehuys

, Lapauw

, T’Sjoen

, et al. Investigating levothyroxine use and its association with thyroid health in patients with hypothyroidism: A community pharmacy study. Thyroid, 2023; 33(8):918–926; doi: 10.1089/thy.2023.0066

Perros

, Nagy

, Papini

, et al. Hypothyroidism and somatization: Results from E-Mode patient self-assessment of thyroid therapy, a cross-sectional, international online patient survey. Thyroid, 2023; 33(8):927–939; doi: 10.1089/thy.2022.0641

10.

Jonklaas

, Bianco

, Cappola

, et al. Evidence-based use of levothyroxine/liothyronine combinations in treating hypothyroidism: A consensus document. Thyroid, 2021; 31(2):156–182; doi: 10.1089/thy.2020.0720

11.

Peterson

, Cappola

, Castro

, et al. An online survey of hypothyroid patients demonstrates prominent dissatisfaction. Thyroid, 2018; 28(6):707–721; doi: 10.1089/thy.2017.0681

12.

Saravanan

, Chau

, Roberts

, et al. Psychological well-being in patients on ‘adequate’ doses of l-thyroxine: Results of a large, controlled community-based questionnaire study. Clin Endocrinol (Oxf), 2002; 57(5):577–585; doi: 10.1046/j.1365-2265.2002.01654.x

13.

Cullum

, Huppert

, McGee

, et al. Decline across different domains of cognitive function in normal ageing: Results of a longitudinal population-based study using CAMCOG. Int J Geriat Psychiatry, 2000; 15(9):853–862; doi: 10.1002/1099-1166(200009)15:9<853::aid-gps211>3.0.co;2-t

14.

Ritchie

, Yeap

. Thyroid hormone: Influences on mood and cognition in adults. Maturitas, 2015; 81(2):266–275; doi: 10.1016/j.maturitas.2015.03.016

15.

Osterweil

, Syndulko

, Cohen

, et al. Cognitive function in non-demented older adults with hypothyroidism. J Am Geriatr Soc, 1992; 40(4):325–335; doi: 10.1111/j.1532-5415.1992.tb02130.x

16.

Wekking

, Appelhof

, Fliers

, et al. Cognitive functioning and well-being in euthyroid patients on thyroxine replacement therapy for primary hypothyroidism. Eur J Endocrinol, 2005; 153(6):747–753; doi: 10.1530/eje.1.02025

17.

Samuels

, Schuff

, Carlson

, et al. Health status, psychological symptoms, mood, and cognition in L-thyroxine-treated hypothyroid subjects. Thyroid, 2007; 17(3):249–258; doi: 10.1089/thy.2006.0252

18.

Correia

, Mullally

, Cooke

, et al. Evidence for a specific defect in hippocampal memory in overt and subclinical hypothyroidism. J Clin Endocrinol Metab, 2009; 94(10):3789–3797; doi: 10.1210/jc.2008-2702

19.

Quinque

, Karger

, Arelin

, et al. Structural and functional MRI study of the brain, cognition and mood in long-term adequately treated Hashimoto’s thyroiditis. Psychoneuroendocrinology, 2014; 42:188–198; doi: 10.1016/j.psyneuen.2014.01.015

20.

Jorde

, Waterloo

, Storhaug

, et al. Neuropsychological function and symptoms in subjects with subclinical hypothyroidism and the effect of thyroxine treatment. J Clin Endocrinol Metab, 2006; 91(1):145–153; doi: 10.1210/jc.2005-1775

21.

Djurovic

, Pereira

, Smit

JWA

, et al. Cognitive functioning and quality of life in patients with Hashimoto thyroiditis on long-term levothyroxine replacement. Endocrine, 2018; 62(1):136–143; doi: 10.1007/s12020-018-1649-6

22.

Giannouli

, Toulis

, Syrmos

. Cognitive function in Hashimoto’s thyroiditis under levothyroxine treatment. Hormones (Athens), 2014; 13(3):430–433; doi: 10.14310/horm.2002.1507

23.

Miller

, Parsons

, Whybrow

, et al. Memory improvement with treatment of hypothyroidism. Int J Neurosci, 2006; 116(8):895–906; doi: 10.1080/00207450600550154

24.

Kramer

, von Muhlen

, Kritz-Silverstein

, et al. Treated hypothyroidism, cognitive function, and depressed mood in old age: The Rancho Bernardo Study. Eur J Endocrinol, 2009; 161(6):917–921; doi: 10.1530/EJE-09-0606

25.

Jensen

, Isaksen

, Ahlberg

, et al. Association of DIO2 and MCT10 polymorphisms with persistent symptoms in LT4-treated patients in the UK Biobank. J Clin Endocrinol Metab, 2024; 109(2):e613–e622; doi: 10.1210/clinem/dgad556

26.

Greendale

, Wight

, Huang

, et al. Menopause-associated symptoms and cognitive performance: Results from the study of women’s health across the nation. Am J Epidemiol, 2010; 171(11):1214–1224; doi: 10.1093/aje/kwq067

27.

Greendale

, Huang

, Wight

, et al. Effects of the menopause transition and hormone use on cognitive performance in midlife women. Neurology, 2009; 72(21):1850–1857; doi: 10.1212/WNL.0b013e3181a71193

28.

Kazlauskaite

, Janssen

, Wilson

, et al. Is midlife metabolic syndrome associated with cognitive function change? The study of women’s health across the nation. J Clin Endocrinol Metab, 2020; 105(4):e1093–e1105; doi: 10.1210/clinem/dgaa067

29.

Derby

, Hutchins

, Greendale

, et al. Cardiovascular risk and midlife cognitive decline in the study of women’s health across the nation. Alzheimers Dement, 2021; 17(8):1342–1352; doi: 10.1002/alz.12300

30.

Janssen

, Powell

, Dugan

, et al. Cardiovascular health, race, and decline in cognitive function in midlife women: The study of women’s health across the nation. J Am Heart Assoc, 2024; 13(9):e031619; doi: 10.1161/jaha.123.031619

31.

Maryfran

, Sybil

, Barbara

, et al. SWAN: A multicenter, multiethnic, community-based cohort study of women and the menopausal transition. Menopause, 2000; 175–188; doi: 10.1016/b978-012453790-3/50012-3

32.

Smith

. Symbol Digit Modalities Test (SDMT) Manual (Revised). Los Angeles; 1982. Available from: https://www.wpspublish.com/sdmt-symbol-digit-modalities-test [Last accessed: February, 14 ].

33.

Hilbert

, Nakagawa

, Puci

, et al. The digit span backwards task. European Journal of Psychological Assessment, 2014; 31(3):174–180; doi: 10.1027/1015-5759/a000223

34.

Albert

, Smith

, Scherr

, et al. Use of brief cognitive tests to identify individuals in the community with clinically diagnosed Alzheimer’s disease. Int J Neurosci, 1991; 57(3–4):167–178; doi: 10.3109/00207459109150691

35.

Wilson

, Beckett

, Barnes

, et al. Individual differences in rates of change in cognitive abilities of older persons. Psychol Aging, 2002; 17(2):179–193.

36.

Tomaszewski Farias

, Mungas

, Harvey

, et al. The measurement of everyday cognition: Development and validation of a short form of the everyday cognition scales. Alzheimers Dement, 2011; 7(6):593–601; doi: 10.1016/j.jalz.2011.02.007

37.

Karlamangla

, Lachman

, Han

, et al. Evidence for cognitive aging in midlife women: Study of women’s health across the nation. PLoS One, 2017; 12(1):e0169008; doi: 10.1371/journal.pone.0169008

38.

Morrow

, Drake

, Zivadinov

, et al. Predicting loss of employment over three years in multiple sclerosis: Clinically meaningful cognitive decline. Clin Neuropsychol, 2010; 24(7):1131–1145; doi: 10.1080/13854046.2010.511272

39.

Salthouse

. Trajectories of normal cognitive aging. Psychol Aging, 2019; 34(1):17–24; doi: 10.1037/pag0000288

40.

Finkel

, Reynolds

, McArdle

, et al. Age changes in processing speed as a leading indicator of cognitive aging. Psychol Aging, 2007; 22(3):558–568; doi: 10.1037/0882-7974.22.3.558

41.

Wadley

, Bull

, Zhang

, et al. Cognitive processing speed is strongly related to driving skills, financial abilities, and other instrumental activities of daily living in persons with mild cognitive impairment and mild dementia. J Gerontol A Biol Sci Med Sci, 2021; 76(10):1829–1838; doi: 10.1093/gerona/glaa312

42.

Miller

, Parsons

, Whybrow

, et al. Verbal memory retrieval deficits associated with untreated hypothyroidism. J Neuropsychiatry Clin Neurosci, 2007; 19(2):132–136; doi: 10.1176/jnp.2007.19.2.132

43.

Burmeister

, Ganguli

, Dodge

, et al. Hypothyroidism and cognition: Preliminary evidence for a specific defect in memory. Thyroid, 2001; 11(12):1177–1185; doi: 10.1089/10507250152741037

44.

Mennemeier

, Garner

, Heilman

. Memory, mood and measurement in hypothyroidism. J Clin Exp Neuropsychol, 1993; 15(5):822–831; doi: 10.1080/01688639308402598

45.

Cooke

, Mullally

, Correia

, et al. Hippocampal volume is decreased in adults with hypothyroidism. Thyroid, 2014; 24(3):433–440; doi: 10.1089/thy.2013.0058

46.

Bauer

, Silverman

, Schlagenhauf

, et al. Brain glucose metabolism in hypothyroidism: A positron emission tomography study before and after thyroid hormone replacement therapy. J Clin Endocrinol Metab, 2009; 94(8):2922–2929; doi: 10.1210/jc.2008-2235

47.

Constant

, de Volder

, Ivanoiu

, et al. Cerebral blood flow and glucose metabolism in hypothyroidism: A positron emission tomography study. J Clin Endocrinol Metab, 2001; 86(8):3864–3870; doi: 10.1210/jcem.86.8.7749

48.

, Ma

, Pan

, et al. Functional magnetic resource imaging assessment of altered brain function in hypothyroidism during working memory processing. Eur J Endocrinol, 2011; 164(6):951–959; doi: 10.1530/eje-11-0046

49.

Zhang

, Zhao

, Chen

, et al. Structural and functional alterations of hippocampal subfields in patients with adult-onset primary hypothyroidism. J Clin Endocrinol Metab, 2024; 109(7):1707–1717; doi: 10.1210/clinem/dgae070

50.

Ettleson

, Ibarra

, Wan

, et al. Demographic, healthcare access, and dietary factors associated with thyroid hormone treatments for hypothyroidism. J Clin Endocrinol Metab, 2023; 108(12):e1614–e1623; doi: 10.1210/clinem/dgad331

51.

Ahluwalia

, Baldeweg

, Boelaert

, et al. Use of liothyronine (T3) in hypothyroidism: Joint British Thyroid Association/Society for endocrinology consensus statement. Clin Endocrinol (Oxf), 2023; 99(2):206–216; doi: 10.1111/cen.14935

52.

Casula

, Ettleson

, Bianco

. Are we restoring thyroid hormone signaling in levothyroxine-treated patients with residual symptoms of hypothyroidism? Endocr Pract, 2023; 29(7):581–588; doi: 10.1016/j.eprac.2023.04.003

53.

Biondi

, Celi

, McAninch

. Critical approach to hypothyroid patients with persistent symptoms. J Clin Endocrinol Metab, 2023; 108(10):2708–2716; doi: 10.1210/clinem/dgad224

54.

Perros

, Nagy

, Papini

, et al. Hypothyroidism and Type D personality: Results from E-MPATHY, a cross-sectional international online patient survey. J Clin Endocrinol Metab, 2024; doi: 10.1210/clinem/dgae140

55.

Elkana

, Eisikovits

, Oren

, et al. Sensitivity of neuropsychological tests to identify cognitive decline in highly educated elderly individuals: 12 months follow up. J Alzheimers Dis, 2016; 49(3):607–616; doi: 10.3233/jad-150562

56.

De Jager

, Hogervorst

, Combrinck

, et al. Sensitivity and specificity of neuropsychological tests for mild cognitive impairment, vascular cognitive impairment and Alzheimer’s disease. Psychol Med, 2003; 33(6):1039–1050; doi: 10.1017/s0033291703008031

57.

Vickers

. How many repeated measures in repeated measures designs? Statistical issues for comparative trials. BMC Med Res Methodol, 2003; 3:22; doi: 10.1186/1471-2288-3-22

58.

Seblova

, Berggren

, Lövdén

. Education and age-related decline in cognitive performance: Systematic review and meta-analysis of longitudinal cohort studies. Ageing Res Rev, 2020; 58:101005; doi: 10.1016/j.arr.2019.101005

59.

Clouston

SAP

, Smith

, Mukherjee

, et al. Education and cognitive decline: An integrative analysis of global longitudinal studies of cognitive aging. J Gerontol B Psychol Sci Soc Sci, 2020; 75(7):e151–e160; doi: 10.1093/geronb/gbz053