Association Between Indices of Thyroid Hormone Resistance and Constipation Prevalence in the US Population

Abstract

Background:

Chronic constipation is a common and complex condition that significantly impairs the quality of life and health care costs. Identifying the underlying cause is crucial for effective management, and hypothyroidism is frequently implicated. Nevertheless, extensive studies are scarce regarding this correlation. The research aims to examine the association between indices of thyroid hormone resistance, particularly the Thyroid Feedback Quantile-based Index of Free Thyroxine (TFQIFT4), and the occurrence of constipation among the population of the United States.

Methods:

The analysis examined data collected from 6354 participants in the National Health and Nutrition Examination Survey (NHANES) conducted between 2007 and 2010. Chronic constipation was determined as the Bristol Stool Form Scale (BSFS) types of 1 to 2 or <3 bowel movements weekly. Thyroid hormone resistance was assessed using various indices. The statistical analysis comprised weighted logistic regression, restricted cubic splines (RCS), subgroup analysis, and sensitivity analysis.

Results:

Chronic constipation was diagnosed in 10% of participants, displaying distinct characteristics. A nonlinear association between TFQIFT4 and constipation was observed, with inflection points at −0.25 and 0.376. Above a TFQIFT4 value of −0.25, a significantly negative association with constipation was found, primarily in females (OR=0.21; 95% CI: 0.10-0.44). No such association was found in males.

Conclusions:

The study elucidates a complex correlation between thyroid hormone resistance, particularly TFQIFT4, and constipation. Gender-specific correlations were evident, with TFQIFT4 demonstrating a negative association with constipation, primarily in females. These findings underscore the need for further investigation into the underlying mechanisms involving central thyroid resistance and constipation across genders.

Chronic constipation, characterized by insufficient bowel movements, substantially impairs the quality of life, potentially triggering anxiety, depression, and cognitive decline.¹ It leads to nearly 1 million outpatient office visits annually.² Chronic constipation can be categorized as primary and secondary. The diagnosis of primary chronic constipation is established through the exclusion of secondary causes.³ Moreover, secondary constipation can arise from various factors, including organic disease (mechanical obstruction), dietary or medication influences (opioids or antidepressants), and metabolic disorders like hypothyroidism.³ The management of chronic constipation depends on the underlying cause and typically involves dietary adjustments, followed by pharmacological agents and biofeedback therapy.⁴ In severe cases, invasive surgical interventions may be considered.⁵ Constipation can only be effectively relieved once the underlying issue is resolved.⁶ While several reviews have mentioned that hypothyroidism is a common cause of constipation, they have provided limited in-depth analysis.^7,8

Although the association between hypothyroidism and constipation is widely reported,⁹ there has been a notable absence of comprehensive cross-sectional studies on a large scale. Limited evidence supports the recommendation of routinely checking thyroid function in patients primarily presenting with constipation.¹⁰ Furthermore, research has indicated that even after achieving euthyroidism through replacement therapy, many hypothyroidism patients continue to experience symptoms,¹¹ suggesting the necessity further to investigate the underlying connection between constipation and thyroid function.

Thyroid function evaluation commonly entails the assessment of thyroid-stimulating hormone (TSH), free triiodothyronine (FT3), and free thyroxine (FT4) concentrations.¹² Nevertheless, the hypothalamus-pituitary -thyroid (HPT) axis regulates the delicate balance of thyroid hormones, with FT3, FT4, and TSH mutually interacting in the body.¹³ For example, the FT3/FT4 ratio in peripheral tissues represents the conversion efficiency of FT4 to FT3 and indicates thyroid hormone availability.^12,14 The pituitary’s thyroid status, which reflects the central sensitivity to thyroid hormones, can be evaluated using the Thyroid-Stimulating Hormone Index (TSHI), the Thyrotropin Thyroxine Resistance Index, also known as the Total Thyroxine Resistance Index (TT4RI), and the Thyroid Feedback Quantile-based Index (TFQI).¹⁵ Impaired sensitivity to thyroid hormones, particularly central resistance, has been associated with metabolic syndrome, overweight, diabetes, mortality related to diabetes, and impaired kidney function.¹² However, the relationship between thyroid hormone resistance and constipation remains unclear.

Therefore, the study aims to investigate the connection between thyroid hormone resistance indices and the occurrence of constipation in the United States, using nationally representative data from the National Health and Nutrition Examination Survey (NHANES).

METHODS

Study Population

The NHANES (https://www.cdc.gov/nchs/nhanes.htm) is a study conducted by the National Center for Health Statistics (NCHS) to gather health and nutrition data on the American population. To ensure the representation of study participants, the organization employed a sampling design that involved stratification, multiple stages, and clustering based on probability. The study utilized demographic details, lab results, examination findings, and questionnaire responses from NHANES 2007–2010. A total of 20,686 individuals were initially included in the study. The exclusion criteria were: (1) individuals under the age of 20 due to the unavailability of the Bowel Health Questionnaire (BHQ) (n=8533), (2) participants with incomplete constipation profiles (n=1708), (3) subjects with missing thyroid function data (n=3916), (4) those lacking the first day dietary information (n=103), and (5) individuals with a history of gastrointestinal cancer diseases (n=56) or inflammatory bowel diseases (n=16). Consequently, a total of 6354 participants were ultimately incorporated into the examination (Fig. 1). The research adheres to the STROBE guidelines.¹⁶

FIGURE 1.

The flow chart for selecting participants.

Definitions of Constipation

The BHQ is a detailed interview tool in NHANES that assesses bowel habits and defecation function in adults, including fecal incontinence and stool frequency.¹⁷ The questionnaire also integrates the Bristol Stool Form Scale (BSFS) to evaluate stool consistency, with scores correlating to colonic transit time and incontinence severity.¹⁸ The BHQ was used to define chronic constipation based on self-reported stool type and frequency during the 2007–2008 and 2009–2010 survey cycles. Participants were presented with a visual card displaying 7 variants of the BSFS and requested to select the number corresponding to their typical or most common stool type in the preceding 12 months.¹⁹ Individuals who reported BSFS type 1 (separate hard lumps, akin to nuts) or type 2 (sausage-like, with lumps) as their regular stool consistency were classified as having persistent constipation. The frequency of stools was assessed by asking, “How many times per week do you typically have a bowel movement?” Individuals who responded <3 times per week were categorized as experiencing constipation.²⁰ Conversely, stool types falling within the range of 3 to 7 and a frequency of 3 or more bowel movements per week were classified as not being constipated.

Thyroid Hormones and Antibodies

Serum thyroid hormones were measured at the University of Washington (Seattle, WA) using a series of immunoenzymatic assays. The assays included a third-generation 2-site “sandwich” assay for human thyroid-stimulating hormone (hTSH) using the Access HYPERsensitive hTSH assay, known for its high sensitivity, specificity, and minimal cross-reactivity with other peptide hormones.²¹ For total and free T3 and T4, we employed competitive binding immunoenzymatic assays where T3 and T4 from the samples competed with biotinylated analogs for binding to specific antibodies, with the resulting complexes captured on streptavidin-coated solid phases.²² The assays for free T3 and T4 also involved alkaline phosphatase–conjugated antibodies and T3-alkaline phosphatase conjugates, respectively. The thyroglobulin (Tg) assay was conducted using a 1-step “sandwich” assay with a biotinylated mixture of monoclonal anti-Tg antibodies and streptavidin-coated paramagnetic particles. The thyroglobulin antibody (TgAb) and thyroid peroxidase antibody (TPOAb) assays utilized sequential 2-step “sandwich” assays with paramagnetic particles coated with the respective antigens and alkaline phosphatase conjugates for detection. All assays relied on the chemiluminescent substrate Lumi-Phos 530, and the light production measured by a luminometer was either directly or inversely proportional to the concentration of the analyte in the sample. The exact amount of each analyte was determined from a stored, multipoint calibration curve, allowing for precise measurement and comparison of thyroid hormone levels and antibody titers within the study population. Details were recorded on a website (https://wwwn.cdc.gov/Nchs/Nhanes/2007-2008/THYROD_E.htm). The reference ranges for normal thyroid function of FT4 and TSH were 7.74 to 20.64 pmol/L and 0.34 to 5.60 mIU/L, respectively.¹⁵

Indices of Thyroid Hormone Resistance

Increased peripheral thyroid hormone activity can be indicated by a higher FT3/FT4 ratio, which was determined as FT3 (pg/mL) divided by FT4 (pmol/L).¹² The TFQI systematically evaluates the HPT axis’s reaction to the fluctuation of FT4 levels in the blood, measuring discrepancies in pituitary reaction to thyroid hormone. And the value of TFQIFT4 can be computed in R using the following code: TFQIFT4=pnorm(FT4_cell, mean=10.075, sd=2.155, lower.tail=TRUE) + pnorm(log(TSH_cell), mean=0.4654, sd=0.7744, lower.tail=TRUE) − 1. The TFQI scale spans from −1 to 1. A value of 0 on the TFQI scale indicates average sensitivity, while negative and positive values represent sensitivity and resistance to FT4, respectively. The TT4RI is obtained by multiplying FT4 (pmol/L) with TSH (mIU/L).²³ The calculation of TSHI involves the code in R: TSHI=log TSH (mIU/L) + 0.1345 * FT4 (pmol/L).²⁴ The study calculated TFQIFT3 and thyrotropin triiodothyronine resistance Index (TT3RI)¹² based on the abovementioned formulas based on FT3. TFQI, TT4RI, TT3RI, and TSHI are parameters exhibiting a positive correlation with the resistance of the thyroid center.

Covariates

Demographic Attributes

Demographic attributes were obtained via interviews and comprised gender (categorized as male or female), age (grouped into ≥ 60 y, 40 to 59 y, and 20 to 39 y), race (divided into non-Hispanic white, non-Hispanic black, Other Hispanic, Mexican American, Other Race-Including Multi-Racial), educational levels (above high school, high school, and below high school), and income poverty ratio (<2 and ≥ 2). Body mass index (BMI) was calculated as body weight (kg) divided by height (m) squared, with measurements taken at the mobile examination center (MEC). Participants were categorized into 3 BMI (kg/m²) groups: under/normal weight (<25), overweight (25 to 29.9), and obese (≥ 30).

Lifestyle Characteristics

Lifestyle characteristics, ascertained through interviews, encompassed drinking, smoking, and physical activity levels. Drinking status was divided into non-drinking, moderate, and excessive drinking, with excessive drinking defined as a daily consumption exceeding 20 g for men and 10 g for women.²⁵ Smoking status was categorized as current (>100 cigarettes smoked in a lifetime and currently smoking), former (>100 cigarettes smoked in a lifetime but not currently smoking), and never (<100 cigarettes smoked in a lifetime). Physical activity levels were divided into active (≥500 MET-min/wk), less active (<500 MET-min/wk), and inactive (no MET record) based on the weekly metabolic equivalent of task (MET) minutes.²⁰

Medical Conditions

Diabetes was diagnosed if any of the conditions were met: (1) confirmation of diabetes mellitus by a doctor; (2) glycohemoglobin HbA1 ≥6.5%; and (3) participants were receiving intramuscular insulin or oral medication for diabetes treatment. Hypertension was defined as having an SBP ≥140 or/and DBP ≥90 mm Hg or a history of hypertension.²⁶ Individuals who scored 10 or higher on the Patient Health Questionnaire (PHQ-9) were identified as having depression.²⁷ Cardiovascular disease (CVD) was identified when participants reported stroke, myocardial infarction, coronary artery disease, angina, or congestive heart failure. Data on gastrointestinal cancers (colon, liver, pancreatic, and stomach cancers) were obtained from medical condition questionnaires. Information on inflammatory bowel disease (IBD), including Crohn’s disease and ulcerative colitis, was available solely for the 2009–2010 survey and used to exclude patients with IBD only for data collected during that period.

Dietary Information

Information regarding dietary intake and dietary supplementation, such as total energy, total fiber, total fat, total carbohydrate, total protein, moisture (daily aggregates of water intake), caffeine, and alcohol, was gathered through two 24-hour dietary recalls. Values were summarized for each category, and in cases of missing data in the second recall, the average dietary intake from both recalls or solely data from the first 24-hour interview was utilized.

Measurement for Electrolyte and Complete Blood Count

Serum samples were collected and processed in accordance with NHANES protocols, then stored at −30°C and shipped to Collaborative Laboratory Services for analysis. Electrolyte concentrations, including phosphorus, iron, sodium, potassium, and chloride, were measured using Beckman Synchron LX20 and DxC800 systems with ion-selective electrode (ISE) methodology. Calcium was assessed by ISE, with ion activity referenced to a sodium electrode, and concentrations calculated using the Nernst equation. Iron levels were determined using the FerroZine method, monitoring absorbance at 560 nm, while phosphorus was quantified by forming a phosphomolybdate complex and measuring absorbance at 365 nm. Osmolality was inferred from the collective electrolyte concentrations. Complete blood count (CBC) parameters, including counts of red blood cell (RBC), white blood cell (WBC), and platelet (PLT), were derived using Beckman Coulter methodology, with the DxH 800 instrument in the NHANES MEC providing a CBC on blood specimens and a distribution of blood cells for all participants.

Statistical Analysis

The study utilized the dietary day one sample weight (WTDRD1) for weighted analysis according to NHANES guidance. Furthermore, the weight of the sample used in the ultimate analysis was equal to half the amount of “WTDRD1” due to the merging of 2 NHANES survey cycles. Statistical procedures entailed the representation of continuous variables using median and interquartile range (IQR), while categorical variables were expressed as survey-weighted percentages (95% CI). Survey-weighted χ² tests were employed for categorical variables to assess P values, while the Wilcoxon rank-sum test was utilized in cases of non-normal distributions for continuous variables. The association between thyroid resistance indices and chronic constipation was examined through univariable and multivariable logistic regression methods, generating odds ratios (ORs) and corresponding 95% CIs. Covariates found statistically significant in the univariable regression were subsequently integrated into the multivariable analysis. Three distinct models, labeled Model 1, Model 2, and Model 3, were constructed to account for the relationship between thyroid resistance indices and the occurrence of constipation. The crucial model was represented by Model 1, whereas Model 2 accounted for extra variables like age group, gender, race, education, and PIR (income poverty ratio). Model 3 went a step further by including additional factors such as BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, PLT, and RBC) in addition to those in Model 2. The nonlinear relationship between exposure X (thyroid resistance indices) and the prevalence of chronic constipation was captured using restricted cubic splines (RCS) with 4 specified nodes, while adjustments were made for covariates.²⁸ Subsequently, sensitivity analyses were conducted, excluding cases with abnormal thyroid function determined by FT4 and TSH. STATA (Stata v17.0, StataCorp LLC) and R (v4.4.2) were utilized for all statistical analyses. Regression models were performed utilizing the survey package²⁹ in R to accommodate the stratified and weighted design of NHANES. Visualizations were generated using the R package ggplot2 (RRID: SCR_014601). Statistical significance was determined at a threshold of P<0.05.

RESULTS

Prevalence of Constipation

Out of the 6354 participants, as outlined in Table 1, a diagnosis of constipation was observed in 641 individuals (10%). The constipation group exhibited several notable distinctions in comparison to the non-constipation group. These distinctions included younger age, a higher proportion of females, having less education, having a lower poverty ratio, having a lower BMI, being less physically active, having fewer alcohol consumers, and having a higher occurrence of depression. The constipation group exhibited lower intake levels of energy, carbohydrates, fiber, fat, and moisture in their daily diet. Concerning thyroid function–related measures, the constipation group exhibited decreased levels of TSH, TFQIFT4, TFQIFT3, TSHI, TT3RI, and TT4RI, with no significant impact observed on FT3 and FT4, as well as TT3 and TT4. Several laboratory examinations of serum also showed differences with constipation groups having lower iron, potassium, osmolality, and RBC count.

TABLE 1.

Characteristics and Thyroid Parameters of the US Population Represented in the NHANES Sample

Characteristic	No constipation (90%) (unweight n=5713; weight N=112,746,807)			Constipation (10%) (unweight n=641; weight N=12,578,605)			P
	n	Proportion% (95% CI)	SE%	n	Proportion% (95% CI)	SE%
Gender							<0.001
Male	2960	49.8 (48.2–51.5)	0.8	189	26.3 (22.2–30.7)	2.1
Female	2753	50.2 (48.5–51.8)	0.8	452	73.7 (69.3–77.8)	2.1
Age group (y)							0.008
≥20, <39	1816	36.4 (33.8–39.1)	1.3	251	43.8 (39.7–47.9)	2.0
≥40, <59	1854	39.2 (37.1–41.3)	1.0	202	35.3 (30.9–39.9)	2.2
≥60	2043	24.4 (22.5–26.4)	0.9	188	20.9 (17.5–24.8)	1.8
Ethnicity							0.091
Mexican American	1046	8.7 (6.5–11.5)	1.2	95	7.6 (5.2–11)	1.4
Other Hispanic	617	4.8 (3.3–7.1)	0.9	86	6.3 (4.1–9.4)	1.3
Non-Hispanic white	2809	71.6 (66.3–76.4)	2.5	288	66.6 (56.6–75.3)	4.6
Non-Hispanic black	1017	9.7 (7.4–12.5)	1.2	147	13.7 (8.6–21.2)	3.0
Other race	224	5.2 (4.2–6.4)	0.5	25	5.9 (2.9–11.5)	2.0
Education							0.013
<High school	1653	18.6 (16–21.4)	1.3	214	23.4 (18.6–29)	2.6
High school	1363	24.3 (22–26.8)	1.2	182	29.3 (25.1–34)	2.2
>High school	2692	57 (52.8–61.2)	2.1	245	47.3 (40.5–54.1)	3.4
Poverty ratio (%)							<0.001
<2	2445	31.3 (28–34.9)	1.7	327	41.2 (35.2–47.6)	3.0
2	2799	62.3 (58.8–65.7)	1.7	262	51.4 (46.5–56.3)	2.4
BMI (kg/m2)							0.023
<25	1553	29.9 (28.1–31.8)	0.9	231	38.6 (32.8–44.6)	2.9
≥25, <29.9	1936	33.3 (31.5–35.2)	0.9	207	32.9 (28.7–37.4)	2.1
≥30	2139	35.4 (33.7–37.2)	0.9	192	26.9 (21.6–33)	2.8
Physical activity							0.009
Inactive	1987	26.6 (24.2–29.2)	1.2	257	35 (31.9–38.1)	1.5
Less active	603	11.1 (9.9–12.5)	0.6	74	13.5 (9.3–19.1)	2.4
Active	3113	62.0 (59.2–64.8)	1.4	308	51.4 (45.4–57.3)	2.9
Smoking status							0.200
Never	2940	51.9 (49.1–54.7)	1.4	371	55.6 (48.7–62.3)	3.4
Former	1516	25.3 (23.5–27.1)	0.9	125	20.4 (16.4–25)	2.1
Current	1255	22.8 (20.7–25.1)	1.1	145	24 (18.3–30.8)	3.1
Drinking status							0.001
None	3904	64.6 (61.2–67.9)	1.6	502	78.5 (72.8–83.2)	2.5
Moderate	553	10.0 (8.7–11.3)	0.6	56	8.1 (5.7–11.3)	1.3
Excessive	1256	25.4 (22.4–28.6)	1.5	83	13.5 (9.1–19.4)	2.5
Depression							<0.001
No	5172	92.1 (90.5–93.4)	0.7	513	80.1 (75.4–84.1)	2.1
Yes	541	7.9 (6.6–9.5)	0.7	127	19.8 (15.8–24.5)	2.1
Hypertension							0.700
No	3248	63.2 (61.3–65.1)	0.9	393	64.7 (60.6–68.7)	2.0
Yes	2464	36.8 (34.9–38.7)	0.9	248	35.3 (31.3–39.4)	2.0
Diabetes							0.140
No	4714	87.9 (86.6–89.1)	0.6	538	89.6 (86.5–92)	1.3
Yes	991	12.0 (10.8–13.3)	0.6	98	10.1 (7.7–13.1)	1.3
CVDs							0.400
0	5290	94.3 (93.2–95.2)	0.5	599	95.2 (91.6–97.4)	1.4
1	276	3.9 (3.2–4.7)	0.4	27	2.9 (1.8–4.6)	0.7
2	94	1.2 (0.9–1.6)	0.2	10	1.7 (0.5–5.5)	1.0
3	28	0.3 (0.2–0.5)	0.1	2	0.1 (0–0.6)	0.1

Characteristics	Median (25% quartile, 75% quartile)			Median (25% quartile, 75% quartile)			P
Energy (kcal)	1992 (1519, 2563)			1737 (1418, 2278)			<0.001
Protein (g)	77 (58, 101)			66 (51, 86)			<0.001
Carbohydrate (g)	237 (180, 308)			231 (179, 293)			0.140
Fiber (g)	15 (11, 21)			13 (9, 18)			<0.001
Fat (g)	73 (53, 102)			62 (47, 89)			<0.001
Caffeine (mg)	134 (53, 249)			113 (44, 251)			0.200
Moisture (g)	2719 (2049, 3594)			2194 (1694, 3104)			<0.001
Phosphorus (mmol/L)	1.23 (1.10, 1.36)			1.23 (1.13, 1.36)			0.200
Iron (μmol/L)	14.90 (11.30, 19.30)			13.40 (10.40, 18.80)			0.010
Sodium (mmol/L)	139.00 (138.00, 141.00)			139.00 (138.00, 141.00)			0.150
Potassium (mmol/L)	4.00 (3.80, 4.20)			3.90 (3.70, 4.10)			<0.001
Calcium (mmol/L)	2.35 (2.30, 2.43)			2.35 (2.30, 2.40)			>0.9
Chloride (mmol/L)	104.00 (102.00, 106.00)			104.00 (102.00, 106.00)			0.400
Osmolality (mmol/kg)	278.00 (275.00, 281.00)			277.00 (274.00, 281.00)			0.008
Platelet count (1000 cells/μL)	253.00 (215.00, 296.00)			262.00 (222.00, 311.00)			0.057
Red blood cell (million cells/μL)	4.68 (4.36, 5.03)			4.56 (4.25, 4.87)			<0.001
White blood cell (1000 cells/μL)	7.10 (5.80, 8.40)			7.00 (5.90, 8.40)			0.800
TgAb (IU/mL)	0.60 (0.60, 0.60)			0.60 (0.60, 0.60)			0.700
TG (ng/mL)	10 (6, 17)			10 (6, 18)			0.800
TPOAb (IU/mL)	1 (0, 2)			1 (0, 2)			>0.9
FT3 (pg/mL)	3.18 (2.90, 3.40)			3.10 (2.90, 3.40)			0.400
FT4 (pmol/mL)	10.30 (9.00, 11.60)			10.30 (9.00, 10.30)			0.069
TT3 (ng/dL)	112 (98, 126)			112 (98, 128)			0.300
TT4 (μg/dL)	7.60 (6.70, 8.70)			7.60 (6.90, 8.80)			0.200
TSH (μIU/mL)	1.67 (1.09, 2.50)			1.49 (1.01, 2.24)			0.013
TFQIFT4	0.00 (−0.22, 0.23)			−0.08 (−0.25, 0.14)			0.001
TFQIFT3	−0.48 (−0.69, −0.28)			−0.53 (−0.72, −0.33)			0.013
TSHI	1.87 (1.44, 2.29)			1.74 (1.34, 2.13)			0.002
FT3/FT4	0.32 (0.28, 0.36)			0.32 (0.29, 0.37)			0.400
TT3RI	5.3 (3.5, 7.9)			4.8 (3.1, 7.2)			0.008
TT4RI	17 (11, 25)			15 (10, 22)			0.004

Continuous variables were presented as the median and interquartile range (IQR), and categorical variables were expressed as n (%). For categorical variables, P values were analyzed by the χ² test. For continuous variables, the Wilcoxon rank-sum test was used in the non-normal model.

BMI indicates body mass index; CVD, cardio-cerebral vascular disease.

Characteristics Based on TFQIFT4 Quartiles

Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/JCG/B159 also presents the characteristics of the participants based on the TFQIFT4 quartiles. Individuals with higher TFQIFT4 levels demonstrated several notable characteristics, such as advanced age, likelihood of being non-Hispanic white, higher education, elevated BMI, abstention from smoking and alcohol consumption, and lack of physical activity. They also reported a higher prevalence of conditions such as depression, hypertension, diabetes, multiple comorbidities, and increased constipation. Furthermore, an association was found between elevated TFQIFT4 scores and reduced energy, fiber, and fat intake. However, there were no notable disparities observed concerning gender. The sensitivity analysis conducted on euthyroid individuals, aimed at assessing the impact of thyroid function on the characteristics of participants across TFQIFT4 quartiles, revealed no significant differences in the distribution of characteristics when compared with the full cohort. The detailed results of this analysis can be found in Supplementary Tables S2 and S3, Supplemental Digital Content 1, http://links.lww.com/JCG/B159, which display the means and standard deviations for key demographic and clinical variables, respectively.

Factors Associated With Constipation

Supplementary Table S4, Supplemental Digital Content 1, http://links.lww.com/JCG/B159, displays the findings from the weighted univariable logistic regression analyses. The analysis identified TFQIFT4 (OR=0.56, 95% CI: 0.39–0.81), TFQIFT3 (OR=0.57, 95% CI: 0.39–0.88), and TSHI (OR=0.81, 95% CI: 0.71-0.93) as parameters negatively correlated with constipation. Several factors were identified as being inversely associated with constipation, including advanced age, higher levels of education, elevated BMI, excessive alcohol consumption, increased physical activity, and a diet rich in energy, fiber, and moisture. In addition, higher serum levels of iron, sodium, potassium, and elevated osmolality, as well as blood tests with higher RBC, were also found to exhibit a negative correlation with constipation. On the contrary, being female, suffering from depression, and having a higher PLT count were significantly and positively correlated with constipation. The odds ratios for these associations were 2.79 (95% CI: 2.14–3.62) for female gender, 2.87 (95% CI: 2.29–3.60) for depression, and 1.0018 (95% CI: 1.0002-1.0033) for each unit increase in PLT count, respectively. These findings underscore the multifactorial nature of constipation and highlight the importance of considering a comprehensive range of demographic, lifestyle, and biochemical factors in its assessment and management. However, no significant differences were observed in actual thyroid function indices, including TgAb, TG, TPOAb, FT3, FT4, TT3, TT4, TT3RI, TT4RI, and FT3/FT4, between the 2 groups. All covariates with significant correlation were included in subsequent analyses.

Thyroid Resistance and Constipation

The overall association between thyroid resistance indices (TFQIFT4, TFQIFT3, and TSHI) and constipation was not statistically significant after fully adjusting for potential confounders (Table 2) in the multivariable regression analysis. In the euthyroid population (Supplementary Table S5, Supplemental Digital Content 1, http://links.lww.com/JCG/B159), the results were similar. Nevertheless, the examination of TFQIFT4 and constipation revealed a significant correlation for the nonlinear test (P=0.0141), indicating that their association is not linear. Consequently, additional investigation is required to assess their relationship. As the RCS curve shown in Fig. 2, the curve was partitioned into 3 segments by 2 inflection points at −0.25 and 0.376.

TABLE 2.

Association Between Thyroid Hormones Resistance Indices and Constipation (N=6354)

	Model 1	Model 2	Model 3
Indices	OR (95%CI)	OR (95%CI)	OR (95%CI)
TFQIFT4	0.56 (0.39–0.81)	0.66 (0.45–0.97)*	0.69 (0.49–1.01)
TFQIFT3	0.57 (0.37–0.88)*	0.70 (0.46–1.08)	0.81 (0.51–1.28)
TSHI	0.81 (0.71–0.93)**	0.88 (0.77–1.001)	0.91 (0.79–1.05)

Model 1, crucial model. Model 2, adjusted for age group, gender, race, education, and PIR (income poverty ratio). Model 3, based on Model 2, adjusted for BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, platelet count, and red blood cell count).

^* P<0.05.

^** P<0.01.

^*** P<0.001.

OR indicates odds ratio.

FIGURE 2.

Nonlinear relationship evaluation using the restricted cubic splines (RCS) model on different measures (N=6354). The relationship between the prevalence of constipation in the whole population with TFQIFT4 (A), TFQIFT3 (B), and TSHI (C). Adjustments were made for age group, gender, race, education, PIR (income poverty ratio), BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, platelet count, and red blood cell count).

Compared with TFQIFT4=0, a pronounced increase in the OR was observed when TFQIFT4 was below −0.25, transitioning from a negative to a positive association until reaching its peak value. Conversely, when TFQIFT4 exceeded −0.25, the OR rapidly declined, gradually stabilizing as TFQIFT4 approached 0.376. The positive relationship diminished throughout the decline and gradually became a negative correlation.

Subgroup Analysis

Subgroup analysis (Table 3) delineated significant correlations between constipation and the TFQIFT4 among specific demographic groups, notably in females (OR=0.52; 95% CI: 0.35–0.77), individuals aged 40 to 59 years (OR=0.46; 95% CI: 0.27–0.80), and those with a BMI ≥30 kg/m² (OR=0.40; 95% CI: 0.22–0.75). The TSHI also demonstrated significant associations with constipation within the 40 to 59 years age bracket (OR=0.77; 95% CI: 0.59–0.99) and among individuals with a BMI ≥30 kg/m² (OR=0.63; 95% CI: 0.49–0.81). In addition, the TFQIFT3 was significantly associated with constipation in individuals with a BMI ≥30 kg/m² (OR=0.45; 95% CI: 0.21–0.96).

TABLE 3.

Association Between Thyroid Hormones Resistance Indices and Constipation in Individuals of Different Subgroups, Fully Adjusted Model (N=6354)

Subgroups	TFQIFT4	TFQIFT3	TSHI
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Gender
Male	1.29 (0.66–2.55)	1.20 (0.47–3.08)	1.11 (0.79–1.55)
Female	0.52 (0.35–0.77)**	0.70 (0.42–1.15)	0.87 (0.75–1.01)
P for interaction	0.005**	＜0.002**	0.500
Age group (y)
≥20, <39	0.96 (0.53–1.71)	0.96 (0.46–1.98)	1.04 (0.83–1.29)
≥40, <59	0.46 (0.27–0.80)**	0.61 (0.26–1.40)	0.77 (0.59–0.99)*
≥60	0.55 (0.28–1.09)	0.92 (0.45–1.89)	0.90 (0.73–1.12)
P for interaction	0.010*	>0.9	0.030*
BMI (kg/m2)
<25	1.17 (0.64–2.13)	1.30 (0.57–2.94)	1.12 (0.87–1.44)
≥25, <29.9	0.63 (0.36–1.12)	0.87 (0.42–1.81)	0.96 (0.78–1.19)
≥30	0.40 (0.22–0.75)**	0.45 (0.21–0.96)*	0.63 (0.49–0.81)***
P for interaction	0.002**	0.500	<0.001***

Fully adjusted model, adjusted for age group, gender, race, education, PIR (income poverty ratio), BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, platelet count, and red blood cell count).

^∗ P<0.05.

^∗∗ P<0.01.

^∗∗∗ P<0.001.

BMI indicates body mass index; OR, odds ratio.

The correlation between TFQIFT4 and constipation was significantly influenced by gender (P for interaction=0.005), age groups (P for interaction=0.010), and BMI (P for interaction=0.002). Sensitivity analysis within euthyroid populations (Supplementary Table S6, Supplemental Digital Content 1, http://links.lww.com/JCG/B159) confirmed these interactions, with significant findings across gender (P for interaction <0.001), age groups (P for interaction=0.003), and BMI (P for interaction<0.001).

Investigations into the potential associations between actual thyroid function indices (Supplementary Table S7, Supplemental Digital Content 1, http://links.lww.com/JCG/B159) and constipation across the entire cohort and the euthyroid population, with a focus on gender-specific associations, revealed no significant associations with constipation for any of the thyroid function indices among males in both cohorts. Conversely, females exhibited a significant inverse relationship with FT4 in the entire cohort, with an OR of 0.91 (95% CI: 0.86–0.97), suggesting a negative effect against constipation. This association was further substantiated by a significant interaction with gender (P for interaction=0.026), highlighting a differential effect of FT4 between males and females. However, within the euthyroid cohort, although females continued to show a significant inverse association with FT4 (OR=0.84, 95% CI: 0.76–0.94) and TSH (OR=0.83, 95% CI: 0.71–0.98), the interaction with gender was no longer significant (P for interaction=0.2). This suggests that the gender-specific effects of actual thyroid function indices observed in the entire cohort may be confounded by the inclusion of individuals with abnormal thyroid function.

The TFQIFT4, demonstrating gender-specific correlations in both the entire cohort and the euthyroid subgroup, and the absence of such associations with FT4 in the euthyroid subgroup, positions TFQIFT4 as a potentially valuable indicator for further investigation into the complex relationship between dietary factors, demographic characteristics, and constipation.

Nonlinear Correlation in Females

Subsequent research conducted on female participants revealed a statistically significant nonlinear correlation (P=0.0016) between the level of TFQIFT4 and the occurrence of constipation, as illustrated in Figure 3. The observed trend of the RCS curve in females closely resembled that of the entire study population. Two distinct inflection points were identified at −0.25 and 0.376 using the RCS curve. After adjusting for all relevant covariates (Table 4), a significantly negative association (OR=0.10; 95% CI: 0.03-0.32) was observed between constipation and the middle segment of TFQIFT4 (−0.25 to 0.376). However, no significant association was found when TFQIFT4 values were below −0.25 or above 0.376. Similar findings were observed in the entire study population, with TFQIFT4 within the middle segment (−0.25 to 0.376) showing a significantly negative association with constipation (OR=0.24; 95% CI: 0.10-0.62). However, no statistically significant correlation was found among males. The population was further stratified based on TFQIFT4=−0.25 to facilitate practical clinical application. Multivariable regression analysis revealed that when TFQIFT4 exceeded −0.25, it served as a negatively associated factor for both the entire study population (OR=0.45; 95% CI: 0.24–0.87) and females (OR=0.20; 95% CI: 0.09–0.44). However, in males, no correlation was observed between segmented TFQIFT4 and the occurrence of constipation. Nevertheless, within the euthyroid group, the same subgroup analysis consistently showed the trend mentioned above only in the total population and females (Supplementary Table S8, Supplemental Digital Content 1, http://links.lww.com/JCG/B159).

FIGURE 3.

Nonlinear relationship evaluation using the restricted cubic splines (RCS) Model in different genders. The relationship between TFQIFT4 and the prevalence of constipation in subgroups of females (A, N_A=3205) and males (B, N_B=3149). Adjustments were made for age group, gender, race, education, PIR (income poverty ratio), BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, platelet count, and red blood cell count).

TABLE 4.

Association Between TFQIFT4 and Constipation Within Diverse Segments Divided by Inflecting Points Determined by Restricted Cubic Spines (RCS) (N=6354)

Segments of TFQIFT4	Total population	Males	Females
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Inflecting points
(−1.00, −0.25)	0.96 (0.25–3.65)	4.15 (0.36–47.7)	0.61 (0.12–3.04)
(−0.25,0.376)	0.24 (0.10–0.62)**	1.92 (0.47–7.76)	0.10 (0.03–0.32)**
(0.376,1.00)	0.66 (0.05–8.48)	2.47 (0.01–706)	0.63 (0.02–19.8)
Divided by −0.25
(−1.00, −0.25)	0.96 (0.25–3.65)	4.15 (0.36–47.7)	0.61 (0.12–3.04)
(−0.25,1.00)	0.45 (0.24–0.81)**	2.22 (0.83–5.92)	0.21 (0.10–0.44)***

Fully adjusted model, adjusted for age group, gender, race, education, PIR (income poverty ratio), BMI, activity level, drinking status, depression, dietary factors (energy, protein, fiber, fat, and moisture), and laboratory examinations (iron, sodium, potassium, osmolality, platelet count, and red blood cell count).

^* P<0.05.

^** P<0.01.

^*** P<0.001.

OR indicates odds ratio.

DISCUSSION

The current study observed a complicated association between the TFQIFT4 and constipation. Analysis of the RCS curve revealed a nonlinear relationship characterized by an inverted V-shaped pattern, with 2 key inflection points identified at −0.25 and 0.376. The finding suggested that TFQIFT4 was intricately associated with constipation. The results revealed a negative correlation between constipation and TFQIFT4 values of −0.25 to 0.376, particularly among females. After dividing the population based on TFQIFT4=−0.25, it was found that higher TFQIFT4 levels than −0.25 had a robustly negative relationship with constipation for the entire population and females, while males showed insignificant results. These gender-specific variances in the correlation between TFQIFT4 and constipation underscore the intricate interaction of this relationship, which may have important implications for clinical practice.

Constipation is associated with a detrimental effect on quality of life, affecting an estimated 10% of the global population.³⁰ It is particularly prevalent among the elderly, and insufficiencies in dietary fiber and fluid intake,³¹ along with reduced physical activity, are recognized as contributing factors to its elevated occurrence.³² Furthermore, constipation has been reported frequently associated with anxiety and depression.³³ Female gender has been recognized as an independent risk factor for constipation, with females being approximately twice as likely to experience constipation compared with males.^34,35 In addition, people with constipation tend to have lower education levels than those without it.³⁶ This study confirms previous research indicating that constipation is influenced by various unfavorable factors such as aging, depression, insufficient elementary dietary intake, as well as an unhealthy lifestyle characterized by a lack of physical activity. Interestingly, individuals in the constipation group were found to have lower proportions of alcohol consumption, suggesting that the stimulation of the gut caused by alcohol may contribute to defecation rather than inhibiting it. Moreover, contrary to the conventional understanding that hypothyroidism exacerbates constipation,³⁷ the findings reveal a contrasting association as evidenced by lower serum TSH levels in the constipation group.

Prior research’s primary focus has been exploring the relationship between hypothyroidism and constipation.³⁸ However, it was found that there was no elevated risk of hypothyroidism in children suffering from chronic constipation.³⁹ Research projects rarely involved euthyroid individuals. Although TSH measurement is frequently utilized as a convenient marker for thyroid activity, its reliability can be influenced by various factors.⁴⁰ In the study, FT4 was found to be negatively correlated with females specifically, not the general population, while TSH showed a negative correlation with females only among euthyroid individuals. This suggests that specific contexts may limit the application of actual thyroid hormone indices. Therefore, thyroid function should be evaluated using multiple tests, such as mass spectrometry, to ensure accuracy. In light of the inconsistent findings, it is essential to consider the balance of the HPT axis and the dynamic nature of thyroid hormones.

Resistance to thyroid hormone (RTH) refers to a widespread lack of response to thyroid hormones in tissues, resulting from mutations in TRα or TRβ,^41,42 leading to symptoms associated with hypothyroidism such as fatigue, parched skin, mood swings, and constipation. The general population, particularly older individuals, may experience a slight acquired resistance. As organisms age, hormonal changes are seen as a way to adapt and conserve energy by shutting down nonessential processes for survival.⁴³ A dynamic balance is maintained by these hormones. Consequently, individual hormone levels like FT4 or TSH might not accurately indicate the impact on metabolism. Composite indices provide a systemic and interconnected perspective on regulating HPT axis feedback control.¹⁴

TFQIFT4, or the thyroid feedback quantile-based index of FT4, is derived from the empirical joint distribution of FT4 and TSH, designed to prevent extreme values in instances of thyroid gland dysfunction. It indicates the feedback capacity of the HPT axis to FT4. Concurrent elevated TSH and FT4 levels define an elevated TFQI value.⁴⁴ Research has indicated that higher TFQI levels are linked to raised levels of serum uric acid, especially in females.⁴⁵ The renal dysfunction had a stronger correlation with the composite TFQI index compared with TSH or FT4 individually.¹² In older euthyroid individuals, impaired responsiveness to thyroid hormones is associated with osteoporosis and fractures, regardless of other traditional risk factors.⁴⁶ Obesity, metabolic syndrome, diabetes, and diabetes-related mortality are correlated with increased RTH indices.¹⁵ The distribution of diverse characters across quartiles of TFQIFT4 in the study yielded consistent findings, indicating a significant correlation between TFQIFT4 and diminished quality of life. The study notably demonstrated that a TFQIFT4 value above −0.25 is negatively correlated with the prevalence of constipation. It is suggested that the level of TFQIFT4 might be a result of feedback mechanisms rather than an initial factor contributing to unfavorable conditions such as constipation. A higher TFQIFT4 indicates a stronger central resistance to thyroxine, thereby maintaining serum thyroxine homeostasis. In individuals with fluctuating TSH levels, TFQIFT4 acts as a surveillance mechanism and exerts protection against diseases.

The gender-dependent association between TFQIFT4 and constipation was found to be noticeable. The relationship was significant in females and consistent with that in the entire population. Currently, no established mechanism exists to account for gender-based differences in the prevalence of chronic constipation.⁴⁷ The cross-effect between the HPT axis and the hypothalamic-pituitary-gonadal (HPG) axis may be responsible for these discrepancies, as the connection between thyroid hormones and sex hormones is inseparable.⁴⁸ The differential effects of sex steroids, such as estrogens and androgens, on thyroid hormone metabolism and function are likely responsible for the observed gender-dependent associations. Estradiol (E2), a predominant female sex hormone, is known to modulate the activity of deiodinases, enzymes crucial for the peripheral conversion of thyroxine to triiodothyronine, and may also interact with estrogen receptors in the intestinal epithelium, potentially influencing gastrointestinal motility and secretion. Conversely, androgens, including testosterone, have been shown to decrease the serum concentration of thyroxine-binding globulin (TBG), leading to a reduction in total thyroxine levels while leaving FT4 levels unchanged. This androgen-induced TBG modulation may contribute to gender-specific differences in thyroid hormone metabolism, thereby influencing gastrointestinal function.⁴⁹ The interplay between thyroid hormones and sex hormones may modulate the expression and function of transporters and receptors in intestinal epithelial cells, such as estrogen receptor alpha and beta, thereby potentially influencing gastrointestinal motility and secretion, which could underlie the gender-specific associations with constipation.⁵⁰ Similar gender-specific impacts have been reported in conditions such as thyroid nodular goiter, Hashimoto’s disease, and papillary thyroid carcinoma.⁵¹

Sensitivity analyses were finally performed in euthyroid individuals to validate the findings, and the slight variances observed when compared with the thyroid dysfunction groups were deemed reasonable, attributing them to the exclusion of extreme outliers in thyroid function indices.

Implementing the understanding of central RTH on human health, the study explored the association between RTH and constipation for the first time. Furthermore, an adequately sized sample of participants was examined. To mitigate the impact of various metabolic factors related to gender, the participants were organized into subgroups in the study. An RCS analysis was performed to enhance the credibility of the results by examining potential nonlinear associations. Moreover, NHANES implemented rigorous quality control measures to guarantee the accuracy and legitimacy of the data.

The study using NHANES data sheds light on the relationship between constipation and thyroid hormone resistance but has limitations. The retrospective, observational nature of NHANES does not permit causality determination. Exclusion of incomplete data may cause selection bias, and missing clinical exams like abdominal and rectal evaluations limit our analysis. In addition, the lack of information on the use of medications and diagnosis of anxiety that could explain secondary constipation in our surveyed population restricts our ability to draw comprehensive conclusions about the causes of constipation in this study. The lack of digestive enzyme and microbiome data are significant oversights, as both are crucial for understanding constipation’s metabolic aspects. Recall bias is a potential limitation in our study due to reliance on self-reported medical conditions and symptoms. Our definition of constipation was based on self-reported data from the BHQ, which did not include questions about straining, the sensation of incomplete evacuation, the sensation of anorectal blockage, or the need for digital maneuvers. These omissions may have affected the accuracy of self-reported constipation status. Furthermore, our diagnosis did not incorporate advanced diagnostic techniques such as magnetic resonance defecography (MR defecography) or anorectal manometry, which could provide a more objective assessment of constipation and related disorders.⁵² These limitations suggest that our findings should be interpreted cautiously and suggest areas for future research with more detailed assessments and diagnostics.

CONCLUSIONS

To summarize, the research demonstrated central resistance to thyroid hormone that is strongly associated with the occurrence of constipation, regardless of other traditional risk factors. We proposed a negative correlation range of turning points above −0.25 for TFQIFT4, particularly in females, concerning the incidence of constipation. The findings still required validation through more rigorously controlled studies. Further investigation is required to better understand the mechanisms of central thyroid resistance on constipation in individuals of different genders, given the unique connections between TFQIFT4 and constipation in both males and females.