The study data, including demographic information, laboratory examination data, questionnaires, and disease history data, were derived from NHANES, a continuous survey focused on the U.S. civilian population that uses a complex, multistage, and probabilistic sampling design to assess the health and nutritional status of adults and children.

All NHANES data are publicly available at https://www.cdc.gov/nchs/nhanes/. Our study was based on two NHANES survey cycles from 2007 to 2010. The data for the 2007-2008 cycle comes from the following website: https://wwwn.cdc.gov/nchs/nhanes/continuousnhanes/default.aspx?BeginYear=2007; The data for the 2009-2010 cycle comes from the following website: https://wwwn.cdc.gov/nchs/nhanes/continuousnhanes/default.aspx?BeginYear=2009.

First, in this study, we did not include people under 20 years of age (n=8533) or those who were pregnant (n=125). Participants who lacked thyroid profile data (n=4927) or bowel health questionnaire data (n=600) were also excluded from the study. A further 668 participants with inflammatory bowel disease (n=19) or thyroid disease (n=649) were excluded. Moreover, those with FT4 or TSH levels outside the normal limits (n=312) and those with cancer (n=522) were also disqualified. Eventually, 4999 subjects were enrolled in this research ( Figure 1 ). All the subjects provided written informed consent. The procedures of the NHANES program were approved by the National Center for Health Statistics (NCHS) Ethics Review Board.

The participants were classified as having chronic diarrhea or chronic constipation according to their answers to the bowel health questionnaire. Participants were shown a card featuring colored illustrations and descriptions of the seven Bristol Stool Form Scale (BSFS) types and asked to tell interviewers “the number that corresponds with your usual or most common stool type.” Participants with a typical BSFS type 1 (separate hard lumps, like nuts) or type 2 (sausage-like, but lumpy) stool type were categorized as having chronic constipation, whereas those with a typical BSFS type 6 (fluffy pieces with ragged edges, a mushy stool) or type 7 (watery, no solid pieces) stool type were considered to have chronic diarrhea (14, 15). The remaining individuals are thought to have normal bowel habits. The question, “How often do you have bowel movements?” was also posed to the participants. The responses that were recorded as times per day were converted to times per week so that all the responses were presented as times per week.

The chemiluminescence technique was used to determine the levels of thyroid hormones. Total triiodothyronine (TT3), free triiodothyronine (FT3), and total thyroxine (TT4) concentrations were measured via a competitive binding immunoenzymatic assay. Free thyroxine (FT4) concentrations were measured via the Access FT4 assay, a two-step enzyme immunoassay. TSH concentrations were measured via a two-site immunoenzymatic assay (“sandwich”). According to previous research, the normal range of TSH is 0.34–5.60 mIU/L, and that of FT4 is 0.6–1.6 ng/dL (16). Participants were classified as euthyroid if their serum TSH and FT4 levels were within the normal limits (16, 17).

Covariates in the study included age, gender, race (Mexican American, Non-Hispanic white, Non-Hispanic black, other Hispanic, and other races), educational level (less than high school, high school or equivalent and more than high school), marital status (married/cohabiting with partner and others), Patient Health Questionnaire (PHQ-9) score, energy intake, total calcium, smoking status (smoked ≥ 100 cigarettes in life, yes or no), frequency of alcohol consumption (<12 drinks per year or ≥12 drinks per year), hypertension, diabetes, body mass index (BMI) (18), and physical activity.

Continuous variables are presented as the means ± standard deviations (SDs). Categorical variables are expressed as proportions. Comparisons among groups divided by bowel habits were evaluated by a weighted one-way ANOVA (for continuous variables) or a weighted chi-square test (for categorical variables). The relationships between thyroid function and abnormal bowel habits were examined via multivariate logistic regression. The relationships between thyroid function and the number of bowel movements were examined via multivariate linear regression. For multivariate tests, the following three models were employed. Model 1 was not adjusted for any variables. Model 2 was adjusted for age, gender, and race. Model 3 was adjusted for age, gender, race, educational level, marital status, BMI, alcohol use, smoking status, total calcium, energy intake, PHQ-9 score, hypertension status, diabetes status, and physical activity. Stratified subgroups were analyzed for gender, age, and BMI (<25 kg/m², 25–30 kg/m², and ≥30 kg/m²) via a multivariate regression model. Moreover, the nonlinear relationships of thyroid hormone (FT3, FT4) levels with the risk of developing chronic diarrhea and the number of bowel movements were investigated through the application of smoothed curve fitting. Threshold effect analysis was used to investigate the inflection points of thyroid hormone (FT3, FT4) levels with the risk of developing chronic diarrhea and the number of bowel movements. EmpowerStats (version 2.0) and R (version 3.4.3) software were used to analyze the statistics. P values < 0.05 were considered to indicate statistical significance.

In this study, 4999 participants were categorized into groups on the basis of normal bowel habits, chronic constipation, or chronic diarrhea status. A total of 7.32% of the population exhibited chronic constipation, and 7.72% exhibited chronic diarrhea. The participants’ average age was 47.63 ± 17.07 years. Among these participants, 53.69% were male and 46.31% were female. Abnormal bowel habits were more prevalent in females than in males (P<0.001). Compared with participants with chronic constipation and those with chronic diarrhea, those with normal bowel habits comprised a greater proportion of the population with higher educational levels and a lower PHQ-9 score. Compared with those in the normal group, participants in the chronic constipation group tended to be younger and have a lower BMI (P<0.05). In the chronic constipation group, FT4 levels and weekly bowel movements were greatly decreased (all P<0.05). Among the three groups, participants with chronic diarrhea tended to have more frequent bowel movements. They also had lower rates of moderate recreational activities and higher rates of smoking, hypertension, and diabetes (all P<0.05). No discernible differences in terms of race; vigorous/moderate work activities; or hormone levels of FT3, TT3, TT4, and TSH were detected among the three groups ( Table 1 ).

The results of logistic regression analysis in Model 3 revealed negative but nonsignificant relationships between thyroid hormone (FT4 and TSH) concentrations and the risk of developing chronic constipation ( Table 2 ). However, in Model 2, FT3 concentrations were significantly positively related to chronic constipation status in the 20–40-year-old population [1.59 (1.01, 2.49), P=0.045]. After comparing gender differences in thyroid function in different age groups, we found that men had significantly higher FT3 concentrations than women (P<0.001) ( Supplementary Table 1 ). Further analyzing the relationships between thyroid function and chronic constipation in different age groups, we found a significant negative correlation between FT3 concentration and chronic constipation in the male 60–80-year-old population (P<0.05) ( Supplementary Table 2 ).

Moreover, a positive correlation between FT3 concentrations and the risk of developing chronic diarrhea was observed in Model 3 [1.37 (1.00, 1.88), P=0.049]. When subgroups were analyzed by gender, the results suggested that the positive associations of FT3 [1.77 (1.12, 2.80), P=0.015] and TT3 [1.01 (1.00, 1.02), P=0.003] concentrations with the risk of developing chronic diarrhea were significantly positive in male participants, whereas higher FT4 concentrations were markedly positively associated with the risk of developing chronic diarrhea in women [3.33 (1.00, 11.08), P=0.049]. When comparing the relationships between thyroid function and chronic diarrhea in different age groups, we found that FT3 concentrations were positively associated with chronic diarrhea in the male 20–40-year-old population(all P<0.05). For women, FT4 concentrations were significantly positively associated with chronic diarrhea in the 40–60-year-old population (all P<0.05) ( Supplementary Table 3 ). In the 60–80-year-old population, the relationships between the above hormones and diarrhea were not significant. According to the subgroup analyses stratified by BMI, FT3 concentrations were significantly positively correlated with the risk of developing chronic diarrhea in the 25-30 kg/m² group [1.72 (1.10, 2.70), P=0.019] in Model 2 ( Table 3 ).

According to linear regression analysis, significantly positive correlations were identified between FT3 concentrations and the number of bowel movements [0.84 (0.39, 1.28), P<0.001]. When a subgroup analysis was performed on the basis of gender, FT3 concentrations were substantially positively correlated with the number of bowel movements in both the male and female subgroups in the three models (all P<0.05). FT4 concentrations were significantly negatively correlated with the number of bowel movements in the female group in Model 2 (P<0.05). In subgroup analyses stratified by age, within the 40–60-year age group, associations between FT3 concentrations and the number of bowel movements were positive according to the three models (all P<0.05). Within the 60–80-year age group, FT3 concentrations were positively correlated with the number of bowel movements (all P<0.05). In subgroup analyses stratified by BMI, within the BMI<25 kg/m², 25–30 kg/m², and ≥30 kg/m² groups, FT3 concentrations were positively correlated with the number of bowel movements in Model 1 and Model 2 (all P<0.05). In the BMI<25 kg/m² group, FT4 concentrations were negatively correlated with the number of bowel movements in three models (all P<0.05) ( Table 4 ).

By analyzing the subclinical hypothyroid and subclinical hyperthyroid populations separately, no statistical differences were observed in bowel habits ( Supplementary Tables 4, 5 ). After comparing the gender differences in bowel habits between these two populations, we found that in participants with subclinical hyperthyroidism, the prevalence of chronic constipation was higher in women than in men (20.36% vs. 0.95%), while the prevalence of chronic diarrhea was lower than in men (2.86% vs. 10.87%). The differences were not statistically significant in patients with subclinical hypothyroidism ( Supplementary Tables 6, 7 ). By analyzing the patients with and without subclinical hypothyroid, no statistical differences were observed in bowel habits ( Supplementary Table 8 ).

The associations of the levels of thyroid hormones (FT3 and FT4) with the risk of developing chronic diarrhea and the number of bowel movements were then described via smoothed curve fitting ( Figure 2 ). Smooth curve fitting revealed nonlinear relationships between FT4 concentrations and the risk of developing chronic diarrhea ( Figure 2C ). In addition, a two-stage regression model was employed to estimate the threshold effect ( Tables 5 , 6 ). For the association between FT4 concentrations and chronic diarrhea, we found an inflection point of 0.79 ng/dl. Below the inflection point, FT4 concentrations were not correlated with the risk of developing chronic diarrhea (OR=0.22, 95% CI: 0.04-1.34), but above the inflection point, there was a positive correlation (OR=5.33, 95% CI: 1.40-20.40).