Testosterone Deficiency, Weakness, and Multimorbidity in Men

The purposes of this study were to evaluate the association between total testosterone (TT) deficiency and weakness on multimorbidity in men. Analyses were performed to examine the prevalence of multimobidity among young, middle-aged, and older men, with and without testosterone deficiency. Multivariate logistic models were also used to determine the association between age-specific TT tertiles and multimorbidity, adjusting for key sociodemographic variables, as well as a secondary analysis adjusted for grip strength. Multimorbidity was more prevalent among men with testosterone deficiency, compared to normal TT in the entire group (36.6% vs 55.2%; p < 0.001); however, differences were only seen within young (testosterone deficiency: 36.4%; normal TT: 13.5%; p < 0.001) and older men (testosterone deficiency: 75.0%; normal TT: 61.5%; p < 0.001). Robust associations were found between the age-specific low-TT (OR: 2.87; 95%CI: 2.14–3.83) and moderate-TT (OR: 1.67; 95%CI: 1.27–2.20) tertiles (reference high-TT) and multimorbidity. Secondary analysis demonstrated that both low TT (OR: 1.82; 95%CI: 1.29–2.55) and moderate-TT (OR: 1.31; 95%CI: 1.01–1.69) were associated with multimorbidity, even after adjusting for obesity (OR: 1.75; 95%CI: 1.07–2.87) and NGS (OR: 1.21 per 0.05 unit lower NGS). Low TT and weakness in men were independently associated with multimorbidity at all ages; however, multimorbidity was more prevalent among young and older men with testosterone deficiency.

Weight was measured using a digital Toledo scale (Mettler-Toledo International, Inc., Columbus, OH), and participants only wore an underwear gown and foam slippers. Height was measured using a fixed stadiometer. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m 2 ). Standard categories were applied to determine if each participant was obese (≥30 kg/m 2 ). Individuals with BMI <18.5 kg/m 2 were excluded, due to the known association between underweight status and chronic disease risk 16 . Waist circumference was measured to the nearest 0.1 cm at the level of the iliac crest. Standard cut points for abdominal obesity in men (>102 cm) were used.

Self-Reported Health Conditions.
The NHANES collects self-reported primary health conditions by asking whether the participant has "ever been told by a doctor or health professional" that he or she has a health condition. For this study, the prevalence of 7 self-reported chronic conditions was evaluated, including diagnosis of: type 2 diabetes, arthritis, cardiovascular disease (congestive heart failure, coronary heart disease, and/or angina), stroke, pulmonary disease (emphysema), hypertension, and clinical depression. Clinical depression was dichotomized as a score of ≥10 on the 9-item Patient Health Questionnaire. Scores of 10 or higher have high sensitivity and specificity for identifying major depression in a primary care setting 17 . Examination and Laboratory Cardiometabolic Abnormalities. In addition to the aforementioned self-reported health conditions, participants were tested on routine cardiometabolic parameters with physical examination and laboratory assessments. Detailed descriptions of the laboratory protocols are provided in the NHANES Laboratory Procedures Manual (https://wwwn.cdc.gov/nchs/data/nhanes/2011-2012/manuals/2011-12_Laboratory_Procedures_Manual.pdf). Resting systolic and diastolic blood pressures were measured three to four times with a mercury sphygmomanometer by trained staff. Fasting and non-fasting measures of HDL-cholesterol, triglycerides, and glucose were measured. Non-fasting serum measures of glycated hemoglobin (HbA1c) were included as a diagnostic test for untreated diabetes, which reflects average plasma glucose for the previous ~three-months.
The diagnostic criterion for diabetes was defined as self-reported diabetes, elevated fasting glucose (≥126 mg/ dL), or HbA1c values ≥6.5% (≥48 mmol/mol) 18 . Participants with diabetes that were being treated with only insulin alone were excluded, as they were considered likely to have type 1 diabetes. Hypertension was defined as self-reported history of hypertension (i.e., physician diagnosis on 2 or more different visits), a systolic blood pressure ≥140 mmHg, or a diastolic blood pressure ≥90 mmHg. Hypertriglyceridemia was determined as ≥150 mg/dL, and low HDL-cholesterol was determined as <40 mg/dL and <50 mg/dL for men and women, respectively.

Multimorbidity.
Multimorbidity was defined as the presence of at least two chronic conditions among a list of the nine aforementioned self-reported chronic diseases or examination/laboratory cardiometabolic abnormalities (i.e., diabetes, hypertension, arthritis, cardiovascular disease, stroke, emphysema, hypertriglyceridemia, low-HDL cholesterol, and clinical depression). These conditions were selected in accordance with guidance from the literature pertaining to older adults and adults with disabilities [19][20][21] , and availability and reliability of the diagnosis within our clinical data.

Covariate. Muscle strength was assessed using a hydraulic handgrip dynamometer (Takei Digital Grip
Strength Dynamometer, Model T.K.K.5401). Detailed descriptions of the protocol are provided in the NHANES Muscle Strength/Grip Test Procedure Manual 23 . A trained examiner explained and demonstrated the protocol to each participant, then adjusted the grip size of the dynamometer to the participant's hand size, and asked the participant to squeeze the dynamometer for a practice trial. Participants were randomly assigned to start the test with the dominant or non-dominant hand, and asked to squeeze the dynamometer with maximal effort, exhaling while squeezing. Each hand was tested three times. Grip strength was normalized (NGS) as strength per body mass.
Statistical analysis. All statistical analyses were performed using SAS 9.3 (SAS Institute, Cary, NC).
NHANES employs a multistage sampling design. Sample weights were used to adjust for oversampling, survey nonresponse, and post-stratification. Further, we took into account subsample weights since we conducted analyses on persons with non-fasting glucose measure. These weights were used to produce unbiased estimates. To obtain correct variance estimation, information on strata and primary sampling unit (PSU) were also utilized. Differences in these characteristics across age categories and for testosterone deficiency were tested using linear regression (proc surveyreg) and logistic regression (proc surveylogistic) for continuous and categorical variables respectively, after creating appropriate categories and dummy coding for each. Partial correlation statistics were conducted to examine the association between TT and NGS, while adjusting for age, race/ethnicity, education and household income.
To assess the odds of multimorbidity in the entire sample, we utilized the univariate and multivariate logistic regression modeling approaches. For model 1, only age-category-specific TT tertiles were entered as the primary exposure variable (reference: High TT tertile). For model 2, further adjustments for known demographic covariates, including age categories, race, education, annual income, and marital status, were included in each model. Lastly, in order to examine whether the association between TT and multimorbidity was explained in whole or part by obesity or NGS capacity, a third model was performed including (1) age-category-specific TT tertiles; (2) all sociodemographic variables; (3) obesity status; and (4) NGS (with units set to 0.05).

Results
The descriptive data of all 2,399 men are presented across age categories in Table 1. Obesity, waist circumference, and most cardiometabolic health parameters were significantly lower among young men (p < 0.001); however, in many cases, middle-aged men had worse cardiometabolic profiles than young and older men. Absolute grip strength and NGS were both significantly lower across higher age categories (p < 0.001). Moreover, TT was highest among young men (p < 0.001); however, there were no differences in TT between middle-aged and older men.
Both TT and NGS were robustly associated with multimorbidity (p < 0.001). Secondary analysis demonstrated that TT was significantly correlated to NGS (r = 0.35; p < 0.001) (Fig. 2), even after adjusting for age, race/ ethnicity, income and education. Unadjusted and adjusted logistic models revealed a robust association between the age-specific low-TT and moderate-TT tertiles (reference high TT tertiles) and multimorbidity (Table 3). In the final multiple logistic model, both low TT and moderate-TT tertiles were still significantly associated with multimorbidity even after adjusting for NGS (

Discussion
The principal findings of this study were that TT deficiency was robustly and independently associated with multimorbidity in a large population-representative sample of U.S. men. Moreover, there was strong evidence supporting a dose-response trend for TT and multimorbidity risk. Specifically, the lowest age-category-specific TT tertile and the middle age-category-specific TT tertile were associated a >3-fold and nearly 75% higher multimorbidity risk, when compared to the highest age-category-specific tertiles of TT. Our results support the findings of several other large population-based prospective studies that indicate increased all-cause mortality and cardiovascular risks in both young and older men within the lowest ranges of TT, in comparison to men in the highest [5][6][7]24 . The dose-response trend we observed remained significant even after adjusting for grip strength, a known, robust predictor of chronic disease and early mortality in men 25 reduced all-cause and cardiovascular mortality accompany each 1 standard deviation, or 1 nmol/L, incremental increase in TT, even for men in the eugonadal range 11,24,27 . Further, a large meta-analysis of observational studies (n = 16,184 men) reported a 55% higher all-cause mortality (RR: 1.55; 95% CI: 1.28 to 1.88) for studies with baseline TT ≤ 487 nd/dl 7 , suggesting a threshold for increased mortality that is very near our TT cutoffs for the highest age-category-specific tertiles. We have also uncovered interesting age disparities in testosterone deficiency and multimorbidity prevalence in our population-representative sample of U.S. men. Specifically, there was a much lower median TT in the youngest age category of men from our study (419 ng/dL), as compared to previous clinical cohort studies which harmonized TT reference ranges from healthy, non-obese men of the same age (533 and 529 ng/dL) 15 . This is very likely due to (1) the previous cohort populations, which were mainly men identified as white; (2) exclusion of ~5% of men who exhibited pituitary, testicular or adrenal disease, or who used medications that affect sex-steroid production from the European Male Aging Study, and those with BMI >30 kg/m 2 from the Belgium Siblings Study of Osteoporosis 15 ; and (3) the lack of adjustment for important sociodemographic factors such as race/ethnicity, income, education, and marital status. Moreover, the recently published harmonized 2.5 th , 5 th , 50 th , 95 th , and 97.5 th percentiles were 264, 303, 531, 852, and 916 ng/dL, respectively, for healthy, non-obese men aged 19-39 years. Currently used cutoffs to diagnose TT deficiency (<300 ng/dL [10.4 nmol/L]) would therefore correspond with the 5 th percentile of the reference range; however, in our study the 5 th percentile among young men aged 20-39 corresponded with a TT of 182 ng/dL. Moreover, multimorbidity was significantly more prevalent among men with testosterone deficiency, compared to normal TT in the entire group (36.6% vs 55.2%); and yet, these differences seem to have been largely driven by differences in multimorbidity for young and older men. Despite strong evidence of worse chronic disease risk profiles among middle aged men with testosterone deficiency, as compared to middle-aged men with normal TT, the differences in multimorbidity were not statistically different. Collectively, this demonstrates the importance of screening TT among young men, particularly those with existing obesity, diabetes, cardiovascular    In patients with multimorbidity, early detection can guide treatment as well as slow progression, or potentially halt disease processes entirely. In this study, among young men there was a strong association between low testosterone and multimorbity. It is unlikely that end-stage disease was responsible for this finding, and thus it can be hypothesized that low testosterone may play an early, causal role in chronic disease processes. This warrants further research as it could change screening guidelines for testosterone deficiency, as well as have a significant impact on the rates of disease processes such as hypertension and diabetes in men.
Not surprisingly, nearly all research related to the influence of testosterone deficiency to potentiate risk for secondary muscle and metabolic dysfunction in men has been conducted within an aging-related context. However, the underlying changes in hormonal and metabolic dysregulation leading to multimorbidity should be regarded as a gradual continuous process throughout the lifespan, rather than a discrete outcome or event. Thus, further evaluation of the temporal sequence of these consequences is of particular importance not only for screening efforts to reduce chronic disease, but also for informing early, targeted interventions to treat declining TT or testosterone deficiency in men. In this regard, debate surrounding the cardiovascular safety of testosterone replacement therapy (TRT) persists because the Testosterone in Older Men with Mobility Limitations (TOM) trial was discontinued due to a higher prevalence of cardiovascular-related events in the TRT vs placebo groups 29 (even though the study has received criticism primarily due to relatively poor classification of cardiovascular events 30 ), and the findings of two large retrospective studies indicated increased cardiovascular disease risk in men receiving TRT 31,32 . While these findings raised initial concerns about cardiovascular safety of TRT, the aforementioned retrospective studies have received extensive scientific criticism 33 and calls for retraction from numerous medical societies 34 . Moreover, they remain at odds with data from several recent meta-analyses that indicate TRT does not increase cardiovascular events in hypogonadal men [35][36][37] , and with findings of a very large retrospective study (n = 83,010 men with documented low T) that indicated TRT was associated with a 56% lower propensity matched all-cause mortality (Hazard Ratio [HR]: 0.44; 95%CI: 0.42-0.46) and 24-36% lower stroke risk (HR:0.64; 95% CI: 0.43 to 0.96) and MI risk (HR:0.76; 95% CI: 0.63 to 0.93), in comparison to untreated hypogonadal men 38 . It is also important to note that the recently published NIH-funded T-trials found no differences in cardiovascular-related or other adverse events among men receiving TRT vs placebo 39 , and there is also recent evidence that TRT is associated with lower mortality rates among diabetic men 40 .
Interestingly, despite that TT and muscle strength are known to be highly associated in men 41 , and that weakness is a known predictor of chronic disease and early mortality, our findings reveal an independent effect of muscle weakness and lower TT on risk for multimorbidity. Future research is needed to better understand the independent and joint effects of low TT and muscular weakness in young adulthood as a risk exposure for multimorbidity tracking into/throughout middle-age and older adulthood.
Limitations. The design of this study was limited by the cross-sectional design of NHANES, which posed challenges in causal inference, especially with respect to reverse causation. Thus, we are unable to deduce whether low TT leads to higher odds of multimorbidity, or conversely, whether poor chronic health profiles lead to declines in TT levels. Moreover, we were unable to determine if other competing risks or unmeasured confounding (e.g., dietary habits, medication use) may have influenced the observed estimates. Future longitudinal studies are needed to better understand how declines in TT contribute to unhealthy aging, as well as the extent to which other chronic conditions (e.g., obesity, diabetes) may mediate the association between low TT, and multimorbidity and mortality. Third, we did not have access to sex hormone-binding globulin or free testosterone for these men, and thus we could not determine the differential contribution of these compared to TT for prediction of multimorbidity across the age spectrum. Lastly, we found no statistical differences in TT and testosterone deficiency between middle aged and older men, a finding that might be due to an underrepresentation of institutionalized older men.

Conclusions
Our results suggest a much higher prevalence of testosterone deficiency occurs in men across the adult age-span than what has been previously reported, and that young and elderly men with testosterone deficiency exhibit a significantly higher multimorbidity risk than their eugonadal counterparts.