Levothyroxine Dosing in Older Adults: Recommendations Derived from the Baltimore Longitudinal Study of Aging

Abstract

Objective:

As thyroid hormone metabolism slows in advanced age, treatment dosing requirements change. Guidelines recommend titration from a low starting dose for older adults with hypothyroidism while providing weight-based estimates for younger populations. However, rapid replacement may be appropriate with acute onset of overt hypothyroidism. Therefore, a weight-based recommendation specific to older adults is needed.

Methods:

We determined mean levothyroxine dose by actual and ideal body weight ratios for the outcome of euthyroid on therapy relative to assay-specific and proposed age-specific ranges for independently living participants aged 65 and above in the Baltimore Longitudinal Study of Aging. We examined risk factors to identify those at highest risk for over-treatment using regression analyses adjusted for potential covariables and clustering to account for multiple visits per individual.

Results:

187 participants aged 65 and above were on levothyroxine at 645 eligible visits. At euthyroid visits, participants were on an average of 1.09 mcg/kg (1.35 mcg/kg IBW), with 84% of euthyroid individuals on a dose of less than 1.6 mcg/kg. Average euthyroid dose did not differ by sex using either actual or ideal body weight. For obese individuals, mean euthyroid dose was lower if calculated using actual body weight (0.9 mcg/kg v 1.14 mcg/kg, p<0.01), but similar using ideal body weight (1.42 v 1.32 mcg/kg-IBW, p=0.41) compared to those with BMI<30.

Conclusion:

Thyroid hormone dose per body weight estimates for replacement in older adults (1.1 mcg/kg) and accounting for obesity (0.9 mcg/kg) are one-third lower than current weight-based dose recommendations for younger populations.

Introduction:

Levothyroxine (LT4) is the primary therapy recommended for the treatment of hypothyroidism. The American Thyroid Association (ATA) states that the goals of treatment with LT4 are to resolve signs and symptoms, normalize thyroid stimulating hormone (TSH) level, and avoid over-treatment.¹ The importance of adequate treatment for hypothyroidism while avoiding over-treatment has been well-established.¹ Under-treatment has been associated with increased risk of hyperlipidemia² as well as atherosclerosis.^{3, 4} Over-treatment increases risk of atrial fibrillation⁵ and osteoporosis.⁶ All of these complications are of increased concern in older adults. Unfortunately, multiple studies have demonstrated that achieving a reference range TSH can be challenging, as up to 40% of adults on thyroid hormone replacement are not euthyroid, with 10-20% over-treated and a similar proportion under-treated.^7-11

The ATA currently recommends when appropriate using an initial full replacement dose estimate of 1.6 mcg/kg actual body weight (ABW) for adults with overt hypothyroidism, a widely accessible approach for clinicians in practice.¹ This dosing estimate is derived from studies after total thyroidectomy¹² as well as several with mixed populations that included participants with spontaneous primary hypothyroidism.^{13, 14} Some studies have shown that ideal body weight (IBW) may be a better estimate of volume of distribution for thyroid hormone,¹⁵ but this has not been widely adopted in clinical practice. In addition, prior studies have primarily enrolled younger adults and did not stratify by age. There are several age-related changes that are likely to contribute to lower dose requirements in older adults. For example, thyroid hormone clearance is reduced and thus half-life is extended, leading to higher steady state levels at a given dose.^{16, 17} In addition, the presence of other chronic medical conditions and associated medications can affect thyroid hormone absorption and further alter metabolism.^{18, 19}

The concern that cardiac disease may be acutely worsened by thyroid hormone replacement has contributed to an altered approach of initiating hypothyroidism treatment at a low dose followed by a slow titration in all older adults. However, full weight-based replacement dose initiation has been found to be safe in adults without coronary artery disease.²⁰ Furthermore, in those with an observed acute onset of hypothyroidism, the ATA currently supports rapid correction for all patients, although there remain caveats to continue taking into account individual patient comorbidities and clinical context.¹ Examples of situations where acute-onset hypothyroidism might be expected in older adults include thyroid surgery, RAI ablation, and drug-related thyroid dysfunction. The latter has become increasingly common over the past decade with the escalating use of tyrosine kinase and immune checkpoint inhibitors (ICI) for cancer treatment.²¹ Thyroiditis is the most common immune-related adverse effect with ICI therapy, with frank hypothyroidism rarely recovering to a euthyroid state.^{22, 23} As many oncology patients receiving therapy are older than 65 years, the growth of these therapies creates an impetus for a weight-based replacement dose recommendation for older adults in whom rapid replacement is appropriate.

Two previous studies have quantitatively evaluated dose requirements for older adults. Rosenbaum and Barzel (1982) observed that older adults required 25% less than younger adults, with a mean dose of 118 mcg daily among 23 euthyroid adults on LT4 (average age 76), compared to 158 mcg daily in 44 younger adults (average age 48).¹⁶ A second smaller study reported an average dose one third less among older adults compared to previously published literature not stratifying by age.^{16, 24} More recently, a study in ICI-induced thyroiditis (n=66) that included older adults (mean age 65 years) reported a mean euthyroid dose of 1.45 mcg/kg, which was similar to athyreotic controls (1.53 mcg/kg NS), while euthyroid controls with conventional hypothyroidism averaged 1.25 mcg/kg (p<0.01), consistent with both near-total destruction of the thyroid in ICI-thyroiditis and lower dose requirements in older cohorts compared to the 1.6 mcg/kg benchmark.²⁵ Thus, to meet the need for a weight-based dose estimate when rapid normalization of thyroid function is indicated in older adults, we examined the relationship between dose and TSH at visits made by healthy participants aged 65 and older in the Baltimore Longitudinal Study of Aging (BLSA).

Materials and Methods:

The BLSA is an ongoing prospective cohort study of adults living independently at the time of enrollment.²⁶ Participants are seen every 2 years between age 65-79, and then annually for a three-day stay at the NIA Clinical Research Unit at Harbor Hospital in Baltimore, MD and are rescheduled if acutely ill, so that non-thyroidal illness does not complicate TSH interpretation. Medical history and medication use are obtained by structured interviews conducted by study staff. Thyrotropin (TSH) has been routinely measured in the BLSA since 2003 on Day 2 in serum collected after an observed overnight fast and before morning medication administration to avoid diurnal variation. Thyroid treatment status categories were defined according to the assay-specific reference ranges (supplemental table 1): euthyroid: TSH within range; over-treated: TSH below range; under-treated: TSH above range. We also examined whether age-specific reference ranges would affect dosing recommendations using two published suggestions for the 97.5^th percentile in older adults of 5.6 mU/L²⁷ and 5.9 mU/L.²⁸ Body weight and height are measured at every visit. Ideal body weight (IBW) was calculated from height using the Devine formula,²⁹ which is widely used in the literature. BMI was calculated as kg/m². Obesity categories based on BMI applied standard definitions as follows: <18.5 underweight; 18.5-24.9 normal weight; 25-29.9 overweight; 30-34.9 class 1 obesity; 35-39.9 class 2 obesity; and ≥40 class 3 obesity. We also analyzed BMI as a binary variable at the threshold definition of obesity (<30 or ≥30).

We evaluated data from visits made by participants aged 65 years or older and on thyroid hormone replacement at which serum TSH was available (n=194, visits=680). We excluded 9 visits made by 3 participants with a history of thyroid cancer, since TSH goals may be different in this population. We also excluded 12 visits at which participants were on combination hormone replacement with T3 as well as LT4 or desiccated thyroid hormone preparations. Thyroid hormone dosing information was incomplete for an additional 14 visits (2.1%). This gave a final cohort of 187 participants and 645 visits, with a mean of 3.7 visits per participant. This study was performed under IRB approval from the Johns Hopkins University and the National Institutes of Health Intramural Research Program.

We analyzed the relationships between participant factors and euthyroid status versus over- or under-treatment. Statistical significance for between-group comparisons of visit characteristics (dose, TSH, sex, age, race, BMI) according to thyroid treatment status were carried out with linear regression for continuous variables, and logistic regression for categorical variables. For continuous predictors, the linearity of the outcome-predictor relationship was explored on the appropriate scale using lowess smoothing curves. Adjustments were made to the modeling of such predictors (BMI categorized by obesity, and age with a linear spline). Because the dataset includes multiple observations per subject, the standard errors for the regression parameters were estimated taking into account potential within-subject outcome correlations using a robust cluster variance estimator approach. Multi-regression models included variables that had univariable significance with an association of p<0.05 or deemed biologically necessary. A sensitivity analysis was performed excluding visits on potentially interfering meds (anti-thyroid drugs n=0 visits, steroids n=9 visits, estrogen n=16 visits, amiodarone n=6 visits, carbamazepine n=2 visits) and active smokers (n=7), which did not affect recommended doses, and so these visits were included in the reported analyses. All statistical analyses were performed using STATA^® (Statistical Data Analysis Package version 16.0 IC, College Station, TX).

Results:

Among adults aged 65 or older and on thyroid hormone replacement with LT4, participants were euthyroid at 78.9% of visits (mean TSH 2.34), with a low TSH (over-treated) at 5.6% (mean TSH 0.22), and elevated TSH (under-treated) (mean TSH 6.16) at 15.5% of visits (Table 1). The cohort was 76.9% white, with a mean age of 80.5 years, mean creatinine of 1.03 mg/dl, mean weight of 72.8 kg, and mean BMI of 26.9 at included visits, none of which differed by thyroid treatment status (under-treated, over-treated, or euthyroid). Women made 62% of the visits but were over-represented among over-treated visits (91.7%) and under-represented among under-treated visits (53.5%, p=0.01).

Table 1:

Participant Characteristics at Included Study Visits

	Participants (n=187)	All visits (n=645)	Over-treated visits (n=36)	Euthyroid visits (n=509)	Under-treated visits (n=100)	p-value
Female (n, %)	116 (62%)	400 (62%)	33 (91.7%)	313 (61.5%)	54 (53.5%)	<0.01
White (n, %)	144 (77%)	496 (76.9%)	33 (91.7%)	392 (77.0%)	71 (71.0%)	0.12
Black (n, %)	28 (15%)	95 (14.7%)	3 (8.3%)	75 (14.7%)	17 (17%)
Othere	15 (8%)	54 (8.4%)	0 (0%)	42 (8.3%)	12 (12%)
Number of visits	3.9 (1-15)
TSH (mean, range)		2.81 (<0.01-24.9)	0.22 (<0.01-0.47)	2.34 (0.39-5.47)	6.16 (4.01-24.9)	<0.001
Age (mean, range)		80.5 (65-101)	81.9 (66-90)	80.3 (65-101)	80.9 (66-98)	0.42
Weight in kg (mean, SDa; range)		72.8 (15.5; 39.1-136.2)	74.3 (12.3)	72.6 (15.4)	73.0 (16.7)	0.79
IBWb in kg (mean, SD; range)		57.9 (9.8; 34.1-81.1)	54.8 (5.8)	57.9 (9.6)	58.7 (11.4)	0.06
BMI (mean, SD; range)		26.9 (4.7; 17.2-43.5)	28.3 (5.3)	26.8 (4.7)	26.7 (4.5)	0.36
LT4:ABWc (mean, SD)		1.1 (0.5)	1.6 (0.5)	1.1 (0.5)	1.0 (0.5)	<0.001
LT4:IBWd (mean, SD)		1.4 (0.6)	2.1 (0.6)	1.3 (0.6)	1.2 (0.7)	<0.001

^aSD: standard deviation

^bIBW: ideal body weight

^cLT4:ABW levothyroxine dose in mcg/kg of actual body weight

^dLT4:IBW: levothyroxine dose in mcg/kg of ideal body weight

^eAsian, Pacific Islander, Native American

The mean dose-to-body-weight ratio among euthyroid participants was 1.09 mcg/kg. The dose was significantly higher when participants were over-treated (1.58 mcg/kg) and significantly lower when under-treated (0.97 mcg/kg, p<0.001 Figure 1; Table 2, supplemental Table 2). The current weight-based dose recommendation of 1.6 mcg/kg actual body weight was at the 84^th percentile for the euthyroid population. The proportion of participants over-treated increased with increasing dose and the proportion under-treated decreased (Figure 2). Across tertiles (mean doses of 0.56, 1.06 and 1.66 mcg/kg), proportion of those under-treated fell from 20.9% to 12.0% while the proportion of those over-treated rose from 0.9% to 12.5% (p<0.001). Out of 36 visits at which TSH was low, only 4 occurred at doses lower than 1.1 mcg/kg, including one participant weighing 77 kg with a fully suppressed TSH on 25 mcg. Using proposed age-based reference ranges to define the bounds of euthyroidism at 5.6 or 5.9 mU/L rather than 4.5 mU/L (from the most common assay) did not change the mean dose among those classified as euthyroid.

Figure 1. Distribution of weight-based levothyroxine dose by thyroid treatment status.

A-C. Distribution of participants by LT4 (levothyroxine) dose among those who were (A) under-treated (elevated TSH), (B) euthyroid (TSH within reference range), and (C) over-treated (low TSH) with mean dose represented by dashed vertical line. D. Distribution of LT4 dose by thyroid status, with light grey representing under-treated population, grey representing euthyroid population, and black representing over-treated population.

Table 2:

Mean (95% CI) Levothyroxine Dose (mcg/kg) by Treatment Status, Sex and BMI

	All	Women	Men	p- value	BMK<30	BMI≥30	p- value
Actual body weight-based dose
Over-treated	1.58 (1.38-1.78)	1.60 (1.39-1.80)	1.30 (0.48-2.12)	0.47	1.72 (1.44-1.99)	1.36 (1.10-1.62)	0.06
Euthyroid	1.09 (1.01-1.17)	1.11 (1.02-1.2)	1.06 (0.92-1.12)	0.49	1.15 (1.06-1.24)	0.90 (0.77-1.03)	0.002
Under-treated	0.95 (0.81-1.09)	1.01 (0.83-1.19)	0.88 (0.67-1.09)	0.37	0.99 (0.84-1.14)	0.77 (0.49-1.04)	0.15
Ideal body weight-based dose
Over-treated	2.10 (1.85-2.35)	2.15 (1.89-2.41)	1.56 (0.73-2.39	0.17	2.05 (1.74-2.36)	2.18 (1.79-2.57)	0.58
Euthyroid	1.35 (1.24-1.45)	1.41 (1.29-1.54)	1.25 (1.09-1.42)	0.13	1.33 (1.22-1.43)	1.43 (1.20-1.65)	0.44
Under-treated	1.17 (1.00-1.34)	1.27 (1.03-1.51)	1.05 (0.82-1.29)	0.20	1.17 (0.98-1.35)	1.21 (0.77-1.63)	0.86

Figure 2. Distribution of thyroid status by levothyroxine dose (mcg/kg).

X-axis: tertile (mean levothyroxine dose per tertile was 0.56, 1.06 and 1.66 mcg/kg ABW in 0, 1, and 2, respectively)

Y-axis: percentage of participants

Using ideal body weight (IBW), the mean dose at euthyroid visits was 1.35 mcg/kg-IBW, 2.10 mcg/kg-IBW at over-treated, and 1.20 mcg/kg-IBW at under-treated visits respectively (Table 2). Weight-based dosing among euthyroid participants did not differ significantly by sex for either actual or ideal body weight. Obese participants (BMI >=30) were euthyroid at a significantly lower dose-to-weight ratio (0.90 mcg/kg) compared to participants with BMI<30 (1.14 mcg/kg, p<0.01). However, the mean euthyroid dose estimate did not differ by BMI when calculated using IBW: 1.32 mcg/kg-IBW and 1.43 mcg/kg-IBW for non-obese and obese respectively (p=0.41).

Risk factors for over-treatment in univariable analyses included dose, sex, and race (Table 3). LT4 dose in mcg/kg (multivariable OR=8.70; 95% CI 2.14-35.22) and female sex (multivariable OR=8.20, 95% CI 2.42-27.85) were strongly associated with over-treatment in both univariable and fully adjusted models, with the association strengthened in the adjusted models. White race was associated with over-treatment (OR=3.47, 95% CI 1.03-11.6) and marginally significant in univariate analysis only. Obesity was associated with risk of over-treatment, which was significant only in adjusted models (multivariable OR=2.32, 95% CI 1.6-3.36). Increasing age was a modest risk factor for over-treatment up to age 90, after which it was highly protective of over-treatment.

Table 3:

Risk Factors for Over-Treatment with Levothyroxine

	Univariate odds ratio	95% CI	p-value	Multivariablea odds ratio	p-value	95% CI
Dose (mcg/kg)	6.24	2.56-15.2	<0.001	8.70	<0.01	2.14-35.22
White vs Non-Whiteb	3.47	1.03-11.6	0.04	2.48	0.149	0.72-8.48
Female vs Male	7.25	2.18-24.18	0.001	8.20	0.001	2.42-27.85
BMI >/=30 vs BMI <30	1.39	0.97-2.01	0.08	2.32	<0.001	1.6-3.36
Age 65-90 yearsc	1.06	1.01-1.14	0.03	1.09	0.02	1.01-1.18
Age >90 yearsc	0.10	0.02-0.55	0.009	0.009	0.002	0.001-0.17

^aModel includes all variables in the table: dose, race, sex, BMI, and age.

^bNon-White includes Black, Asian, Pacific Islander, Native American.

^cAge was analyzed with a linear spline.

Discussion:

Current weight-based LT4 dosing recommendations by major societies do not include specific recommendations for older adults in whom rapid replacement for acute hypothyroidism may be indicated. Here we used a heterogenous population of community-dwelling older adults to establish such a target. We found that a dose of 1.1 mcg/kg actual body weight (1.35 mcg/kg IBW) represented the mean LT4 dose in euthyroid older adults (average age 80.5 years). Participants were over-treated at a dose below that level at only 4 visits (0.6%). In contrast, we found that the current benchmark recommendation of 1.6 mcg/kg would exceed the dose needed to achieve euthyroid levels in 83% of euthyroid visits. Interestingly, our results are consistent with the older literature suggesting that older adults required 67-75% of the dose of younger adults.^{16, 24} Expanding the definition of euthyroid visits to those within proposed age-specific reference ranges for TSH did not change the mean euthyroid dose. This can be explained by the significant overlap between doses at under-treated and euthyroid visits (Figure 1D), so that the combination of these curves did not shift the mean. This overlap also means that our dose recommendation errs on the side of avoiding hyperthyroidism and its associated adverse effects^{5, 6} while accepting a higher rate of initial under-treatment.³⁰ Thus, our results are likely to provide a relatively conservative, but more rapid titration to euthyroid results compared to initiation at a set low dose of 25 mcg. This should be tested in future prospective studies.

One limitation of this study is that the indication for thyroid hormone use in our participants is not known, and so some participants may have had considerable thyroid reserve, for example if treatment was initiated for mild disease. This would bias our hormone replacement recommendation towards a lower dose than would be required in athyreotic persons. However, such a bias decreases the risk of over-treatment from using 1.1 mcg/kg as an initial estimate for a replacement dose in older adults, meaning this recommendation is a conservative estimate. Aging populations develop a wide range of functional capacities and disabilities which might also impact on the risk of over-treatment and the clinical applicability of this approach. Our population is ambulatory and relatively healthy, and thus represents those in whom a weight-based initial dose estimate is most appropriate to use, and therefore a reasonable population from which to derive this estimate.

Ideal body weight (IBW) may be a better estimate of the volume of distribution for thyroid hormone than ABW, and therefore dose requirement calculated with IBW has been proposed as more accurate in those with higher body water, including women^{15, 31, 32} and those who are obese.^{17, 33} While we found that the euthyroid dose was slightly lower for men than for women, the difference was not significant either by actual or ideal body weight. Excluding visits at which participants were taking estrogenic therapies did not impact these findings. Thus, it may be that changes in body composition associated with aging dominate over those associated with sex in this cohort.

In contrast, we did find a significant difference in dose related to obesity. Those at a BMI ≥30, with a mean euthyroid dose of 0.9 mcg/kg ABW, would be more likely to be over-treated if the dose estimate was done using the total population ABW-based dose of 1.1 mcg/kg. In contrast, the IBW dose of 1.35 mcg/kg-IBW was not statistically different across BMI categories. Ideally dose estimates should use an ideal body weight calculation. However, in the absence of ready clinical access to IBW metrics, the lower actual body weight dosing estimate should be applied to patients with obesity. Both strategies provide an adjustment for the altered volume of distribution that results from increased BMI.

The high rates of over-treatment we observed among obese individuals and women in particular suggest that factors other than TSH may be driving dosing decisions in these populations. Given that the risks associated with over-treatment exacerbate risks already present in these populations, our findings suggest a need for increased clinician awareness of the goals for thyroid hormone treatment in older adults.

Conclusion:

By providing an estimate for weight-based thyroid hormone replacement in a population of community-dwelling adults aged 65 and older, clinicians now have additional information with which to personalize the treatment decisions for their older adults, some of whom may benefit by more rapid thyroid hormone dose titration.

Highlights.

• Levothyroxine dose requirements are lower in older adults

• A dose of 1.1 mcg/kg can achieve rapid correction of hypothyroidism in older adults

• A dose of 1.35 mcg/kg ideal body weight is preferred in older adults at higher BMI

Clinical Relevance:

The ATA Guidelines recommend a weight-based dose of levothyroxine (1.6 mcg/kg) for treatment of hypothyroidism in younger populations. In this paper, we recommend a weight-based dose of 1.1 mcg/kg for older adults when rapid correction of hypothyroidism is indicated, rather than following the typical “start low and go slow” method.