Testosterone Therapy Is Associated With Increased Odds of Quadriceps Tendon Injury

Abstract

Background

Anabolic steroid use at supraphysiologic doses has been associated with an increased risk of tendon injury. However, the musculoskeletal effects of testosterone therapy in the clinical setting are not well understood.

Questions/purposes

(1) Is prescription testosterone associated with a higher odds of subsequent quadriceps muscle or tendon injury? (2) Is prescription testosterone associated with a higher odds of surgical repair of the quadriceps tendon?

Methods

The PearlDiver Database, which contains data on Medicaid, Medicare, and commercially insured patients, allows for a large representative sample of the US population including both publicly and privately insured patients. The database was queried for all patients between 2011 and 2018 who filled a testosterone prescription. Additionally, all quadriceps injuries using ICD-9 and ICD-10 codes between 2011 and 2018 were queried. Propensity score matching based on age, sex, Charlson comorbidity index, and specific comorbidities allowed us to create matched control groups. We used the t-test and chi-square analysis to compare the unmatched and matched cohorts. A total of 151,797 patients (123,627 male patients and 28,170 female patients) with a history of filled testosterone prescriptions were included in the study after matching with the control group, which was of equal size and representation of age, male-female proportions, and comorbidities. Chi-square and logistic regression analyses were performed to compare odds of quadriceps injury and quadriceps tendon repair among the testosterone groups to that of their respective control groups by age and sex.

Results

Within 1 year of filling prescriptions for testosterone, 0.06% (97 of 151,797) of patients experienced a quadriceps injury compared with less than 0.01% (18 of 151,797) of patients in the control group (OR 5.4 [95% CI 3.4 to 9.2]; p < 0.001). Within the sex-specific matched groups, filling a testosterone prescription was associated with an increase in the odds of quadriceps injury in male patients within 1 year of the prescription (OR 5.8 [95% CI 3.5 to 10.3]; p < 0.001). Additionally, patients who filled a testosterone prescription were at increased risk of having quadriceps tendon repair within a year of the injury than were patients in the matched control group (OR 4.7 [95% CI 2.0 to 13.8]; p = 0.001).

Conclusion

Considering these findings, it is important for physicians to counsel patients receiving testosterone replacement therapy of the substantially increased odds of quadriceps tendon injury. Future investigations into the mechanisms of influence of exogenous anabolic steroids on tendon injury remains of interest.

Level of Evidence

Level III, therapeutic study.

Introduction

Injuries to the quadriceps can be painful and debilitating. These injuries occur on a spectrum, ranging from mild strains or contusions of the muscle belly to more severe quadriceps tendon ruptures that can lead to surgery. Although strains of the musculature are more frequently seen in younger patients and athletes—typically high school and college age groups—tendon tears are more common in the older population as a result of trauma and overuse [4]. Clayton and Court-Brown [4] reported an incidence rate for quadriceps tendon ruptures of 1.37 per 100,000 person years in the United Kingdom with a mean age of 50.5 years and 51.7 years in male and female patients, respectively. Several factors are known to predispose patients to quadriceps tendon tears, including age, renal disease, diabetes, obesity, osteomalacia, systemic lupus erythematosus, and hyperparathyroidism [8]. Quadriceps injuries have also been associated with the illicit or nonmedical use of supraphysiologic doses of anabolic steroids. These compounds, which are synthetically derived forms of testosterone, are widely used to rapidly gain muscle mass in the setting of body building, weightlifting, and other sports where raw strength is desirable [17]. There have been several cases of anabolic steroid use as a predisposing factor for quadriceps tendon rupture in body builders [5,6,7, 14]. Although the steroid doses associated with tendon injuries in body builders are far higher than those prescribed for medical reasons, the possibility of injury, even at lower doses, is concerning.

Anabolic steroid use in the form of testosterone replacement therapy has been on the rise in recent years, with a greater than threefold increase between 2001 and 2011 in the United States [1]. This therapy is seeing wider use worldwide as a way to treat the symptoms of hypogonadism, a clinical condition resulting from low sex hormone levels [16]. Hypogonadism is typically associated with aging in male patients and presents with decreases in bone mass, muscle mass, libido, and concentration, as well as increases in central body fat and osteoporosis [2]. Although testosterone replacement therapy has proven beneficial for these patients, it also increases the risk of prostate cancer, liver toxicity, congestive heart failure, ischemic stroke, myocardial infarction, worsening sleep apnea, skin disease, and infertility [2, 19]. Furthermore, exogenous testosterone use has been connected specifically to musculoskeletal injuries as well as decreased tensile strength of tendons [9, 13]. Although the link between anabolic androgenic steroid use and tendon injury has been widely documented, there is a paucity of information regarding the musculoskeletal injury risks from testosterone replacement therapy. In contrast to anabolic steroid users who reach supraphysiologic levels of testosterone, the patients evaluated in the present study received testosterone replacement therapy prescriptions of much lower doses.

We therefore asked: (1) Is prescription testosterone associated with a higher odds of subsequent quadriceps muscle or tendon injury? (2) Is prescription testosterone associated with a higher odds of surgical repair of the quadriceps tendon?

Patients and Methods

Study Design and Setting

This was a retrospective, comparative study drawn from a large, longitudinally maintained database. We queried PearlDiver (PearlDiver Technologies) deidentified data in the Mariner database for patients treated between 2011 and 2018. This database contains healthcare data for more than 150 million patients gathered from all Humana Inc insurance claims between 2010 and 2020. This allows for a large representative sample of the US population including both publicly and privately insured patients. This dataset provides the means to longitudinally track patients and their procedures, medical diagnoses, and medications. These data were used in this study to compare proportions of quadriceps muscle, fascia, and tendon injuries as well as subsequent surgical repair of the quadriceps tendon among patients with prior exogenous testosterone prescriptions with the same proportions in a control population.

Patients

Among the full M151Ortho dataset included in PearlDiver, 1,959,238 patients filled at least one prescription for testosterone between 2011 and 2020. We included patients between the ages of 35 and 75 years who filled a prescription for testosterone for a minimum of 3 consecutive months over an 8-year period between 2011 and 2018. This allowed for a minimum of 2 years of follow-up time for each patient. To ensure full medical records, we included only patients who were active (with existing insurance claims or no change in insurance provider) in the database for 1 year before and 2 years after the filled prescription. These inclusion criteria were met by 40% (776,974) of those patients. Patients with a diagnosis of connective tissue disease, mitochondrial disease, or cancer were excluded from the study (Supplemental Table 1; http://links.lww.com/CORR/B160). After exclusions, 27% (520,915) of the patients remained, of which we took a random sample of 10% (200,000) of the patients for the matching process. This was done to facilitate data handling of such large samples. The control group was created by generating a random sample of 6 million patients of all ages from 2010 to 2020. This group was then filtered to only capture patients between 35 and 75 years of age who did not meet any of the exclusion criteria and who had never filled a prescription for exogenous testosterone. After the inclusion and exclusion criteria were applied to the control sample of 6 million patients, 9% (544,702) were included for the matching process.

Matching the Testosterone and Control Groups

We used the R Statistical Package embedded in PearlDiver to perform propensity score matching of the overall, male, and female testosterone groups with their respective controls based on age, sex (in the overall cohorts), Charlson comorbidity index score [3], depression, diabetes, osteoarthritis, tobacco use, and osteoporosis. After matching of the testosterone group with the randomly generated control group, 151,797 patients (123,627 male patients and 28,170 female patients) were included in the analysis. The average age of patients in the testosterone cohort was 54 ± 11 years. We subsequently subdivided both the testosterone and control groups into sex- and age-specific subgroups.

Descriptive Data

Before matching, the two groups differed across all patient demographics including age, sex, Charlson comorbidity index score, tobacco use, and prevalence of comorbidities. After matching, the groups showed no appreciable clinical or demographic differences (Table 1).

Table 1.

Comparison of patient demographics of unmatched and matched testosterone and control cohorts

Characteristic	Unmatched			Matched
	Testosterone (n = 200,000)	Control (n = 544,702)	p value	Testosterone (n = 151,797)	Control (n = 151,797)	p value
All patients
Sex, female	15 (29,499)	56 (304,327)	< 0.001	19 (28,170)	19 (28,170)	> 0.99
Age in years	55 ± 11	54 ± 10	< 0.001	54 ± 11	54 ± 11	> 0.99
CCI	0.9 ± 1.6	0.1 ± 0.5	< 0.001	0.3 ± 0.7	0.3 ± 0.7	> 0.99
Tobacco use	32 (63,109)	26 (140,315)	< 0.001	28 (42,101)	28 (42,101)	> 0.99
Diabetes	40 (80,438)	28 (153,304)	< 0.001	32 (48,830)	32 (48,830)	> 0.99
Male	n = 170,501	n = 240,316		n = 123,627	n = 123,627
Age in years	55 ± 11	54 ± 11	< 0.001	54 ± 11	54 ± 11	> 0.99
CCI	1.0 ± 1.6	0.2 ± 0.6	< 0.001	0.3 ± 0.7	0.3 ± 0.7	> 0.99
Tobacco use	33 (55,675)	31 (74,668)	< 0.001	29 (35,269)	29 (35,269)	> 0.99
Diabetes	43 (73,032)	30 (73,099)	< 0.001	34 (41,980)	34 (41,980)	> 0.99
Female	n = 29,499	n = 304,324		n = 28,170	n = 28,170
Age in years	56 ± 9	54 ± 12	< 0.001	56 ± 9	56 ± 9	> 0.99
CCI	0.4 ± 1.0	0.1 ± 0.5	< 0.001	0.3 ± 0.7	0.3 ± 0.7	> 0.99
Tobacco use	25 (7434)	22 (65,641)	< 0.001	24 (6832)	24 (6832)	> 0.99
Diabetes	25 (7406)	26 (80,193)	< 0.001	24 (6850)	24 (6850)	> 0.99

Data are presented as % (n) or mean ± SD; CCI = Charlson comorbidity index.

Identifying All Quadriceps Injuries and Subsequent Surgical Reattachment

We queried the M151Ortho dataset in PearlDiver for all quadriceps injuries using ICD-9 and ICD-10 codes between 2011 and 2018. The number of patients in each respective cohort who experienced a quadriceps injury within 1 year of and any time after their 3 consecutive months of filled testosterone prescriptions was recorded. Using the patients who experienced a quadriceps injury any time after the first timepoint of the initial analysis, we used Current Procedural Terminology codes to analyze the odds of those injuries that were indicated for surgical repair.

Ethical Approval

Ethical approval was not sought for the present study because the database accessed contained de-identified data.

Statistical Analysis

We used the t-test and chi-square analysis to compare the unmatched and matched cohorts. Multivariable logistic regression was used to calculate and compare the odds of quadriceps injury and subsequent surgical reattachment among the testosterone groups with those of their respective control groups while controlling for age, sex, Charlson comorbidity index score, depression, diabetes, osteoarthritis, tobacco use, and osteoporosis. We report ORs and 95% CIs for each comparison. A p value less than 0.05 was the cutoff for statistical significance. We used the R Statistical Package (R Foundation) embedded in PearlDiver for all statistical analyses.

Results

Association of Testosterone With Increased Odds of Quadriceps Injury

Receipt of a prescription for testosterone was associated with an increased odds of a patient being diagnosed with a quadriceps injury within 1 year of that prescription (OR 5.4 [95% CI 3.4 to 9.2]; p < 0.001). In the testosterone group, 0.06% (97 of 151,797) of patients experienced a quadriceps injury compared with 0.01% (18 of 151,797) in the control group. In the male patient-specific matched cohorts, filling a testosterone prescription was associated with an increase in the odds of quadriceps injury within 1 year of the prescription (OR 5.8 [95% CI 3.5 to 10.3]; p < 0.001) (Table 2). Female patients were not found to have an increased likelihood of quadriceps injury during the first year of filling testosterone prescriptions compared with their matched control cohort. Among male patients, each specific age cohort had an increased likelihood of developing a quadriceps injury; however, this was not seen in any of the age cohorts among female patients (Table 2). The same analysis was performed to assess the odds of quadriceps injury among the testosterone group at any time after the filled prescriptions for testosterone. Receipt of a prescription for testosterone was still associated with an increased odds of a patient being diagnosed with a quadriceps injury any time after that prescription compared with the control cohort (OR 1.9 [95% CI 1.6 to 2.2]; p < 0.001). Among male patients who filled prescriptions for testosterone, both the overall cohort and each age subdivision experienced increased odds of injury any time after the prescriptions (Table 3). Among female patients, the overall testosterone cohort experienced increased injury odds, although this association was not consistent among specific age subdivisions.

Table 2.

Number of quadriceps tendon injuries within 1 year of filling prescriptions for exogenous testosterone

Population	Age group in years	Prior testosterone use		Control		OR (95% CI)	p value
		Total number of patients	Percentage of quad tendon injuries	Total number of patients	Percentage of quad tendon injuries
All patients		151,797	0.06 (97)	151,979	< 0.01 (18)	5.4 (3.4 to 9.2)	< 0.001
Male		123,627	0.08 (93)	123,627	0.01 (16)	5.8 (3.5 to 10.3)	< 0.001
	35 to 45	31,328	0.05 (17)	31,328	N/A (< 11)	5.7 (1.9 to 24.3)	0.006
	46 to 55	37,883	0.09 (33)	37,883	N/A (< 11)	6.6 (2.8 to 19.3)	< 0.001
	56 to 65	31,696	0.08 (24)	31,696	N/A (< 11)	4.8 (2.0 to 14.3)	0.001
	66 to 75	22,720	0.08 (19)	22,720	N/A (< 11)	6.3 (2.2 to 27.0)	0.003
Female		28,170	N/A (< 11)	28,170	N/A (< 11)	1.7 (0.4 to 8.1)	0.48
	35 to 45	3719	N/A (< 11)	3719	N/A (< 11)	N/A	> 0.99
	46 to 55	10,228	N/A (< 11)	10,228	N/A (< 11)	1.5 (0.3 to 11.4)	0.66
	56 to 65	10,052	N/A (< 11)	10,052	N/A (< 11)	N/A	> 0.99
	66 to 75	4171	N/A (< 11)	4171	N/A (< 11)	N/A	> 0.99

A comparison with a matched cohort of patients followed for 1 year; data presented as % (n). The number of patients in a cohort size less than 11 is not reportable per the Health Insurance Portability and Accountability Act.

Table 3.

Number of quadriceps tendon injuries any time after filling prescriptions for exogenous testosterone

Population	Age group in years	Prior testosterone use		Control		OR (95% CI)	p value
		Total number of patients	Percentage (n) of quad tendon injuries	Total number of patients	Percentage (n) of quad tendon injuries
All patients		151,797	0.34 (520)	151,979	0.18 (274)	1.9 (1.6 to 2.2)	< 0.001
Male		123,627	0.37 (455)	123,627	0.19 (239)	1.9 (1.6 to 2.2)	< 0.001
	35 to 45	31,328	0.32 (99)	31,328	0.19 (59)	1.7 (1.2 to 2.3)	0.002
	46 to 55	37,883	0.42 (160)	37,883	0.26 (100)	1.6 (1.3 to 2.1)	< 0.001
	56 to 65	31,696	0.39 (124)	31,696	0.18 (58)	2.1 (1.6 to 3.0)	< 0.001
	66 to 75	22,720	0.32 (72)	22,720	0.01 (22)	3.3 (2.1 to 5.4)	< 0.001
Female		28,170	0.17 (49)	28,170	0.10 (28)	1.8 (1.1 to 2.8)	0.02
	35 to 45	3719	N/A (< 11)	3719	N/A (< 11)	1.0 (0.3 to 3.2)	> 0.99
	46 to 55	10,228	0.17 (18)	10,228	0.11 (11)	1.6 (0.8 to 3.6)	0.20
	56 to 65	10,052	0.16 (16)	10,052	N/A (< 11)	1.8 (0.8 to 4.2)	0.17
	66 to 75	4171	N/A (< 11)	4171	N/A (< 11)	4.5 (1.2 to 29.6)	0.05

A comparison with a matched cohort of patients; data presented as % (n). The number of patients in a cohort size < 11 is not reportable per the Health Insurance Portability and Accountability Act.

Association of Testosterone With Increased Odds of Quadriceps Tendon Repair

Receipt of a prescription for testosterone was associated with increased odds of a patient undergoing quadriceps tendon repair within 1 year of injury compared with the matched control group (OR 4.7 [95% CI 2.0 to 13.8]; p = 0.001) (Table 4). In sex-specific subdivisions, this association held true for male patients; however, there was no difference in the odds of quadriceps injuries treated surgically among female patients.

Table 4.

Odds of surgical reattachment within 1 year of experiencing a quadriceps tendon injury

Population	Prior testosterone use		Control		OR (95% CI)	p value
	Total number of quad tendon injuries	Percentage of surgical reattachments	Total number of quad tendon injuries	Percentage of surgical reattachments
All patients	520	8 (44)	274	N/A (< 11)	4.7 (2.0 to 13.8)	0.001
Male	455	9 (40)	239	N/A (< 11)	7.4 (2.3 to 24.4)	< 0.001
Female	49	N/A (< 11)	28	N/A (< 11)	N/A	> 0.99

Data presented as % (n). The number of patients in a cohort size < 11 is not reportable per the Health Insurance Portability and Accountability Act.

Discussion

Injuries to the quadriceps muscle and tendon can be debilitating and affect people of all ages, from teens to the elderly. Various factors are known to predispose patients to quadriceps tendon injury such as diabetes, obesity, and osteomalacia [8]. Furthermore, the evidence points to the use of exogenous anabolic steroids as a predisposing factor [9, 10]. With the use of testosterone replacement therapy on the rise because of its benefits in treating the symptoms of hypogonadism, it is important to further explore its relationship with musculotendinous injury [1]. Our study demonstrates a strong association between prescribed testosterone and the likelihood of quadriceps injuries in male and female patients. These findings are consistent within a 1-year follow-up period after the initial 3 months of filled testosterone prescriptions, as well as any time after the filled prescriptions during the 10-year study period. This study determined that the risk of operative quadriceps tendon injuries was much higher in male and female patients receiving exogenous testosterone prescriptions, as well as in men alone than in their matched counterparts without a history of filling testosterone prescriptions. This information is useful for providers to consider when determining a patient’s candidacy for testosterone replacement therapy and for proper counseling of the associated risks.

Limitations

Inherent to any retrospective study of an administrative claims database, there are several limitations. First, the data depend on comprehensive and accurate coding and are susceptible to miscoding and noncoding errors; however, because of the large sample size, the effect of this was minimized. Second, quadriceps injuries are uncommon; although the risk of injury with testosterone replacement therapy is higher than without, this is a rare issue. The low prevalence of such an injury, even with a large sample size, may have affected our results through sparse data bias indicated by wide CIs surrounding some of the ORs. This can also be seen in the female-specific analysis (Table 3) where an association existed in the model that included all female patients but was lost in the specific age subgroups. With such a low prevalence of injury in these age subsets, statistical power was reduced, leading to loss of significance; however, we believe the association still holds true for the female cohort as a whole. Third, the definition of quadriceps injury in our study is based on ICD codes and can range from a muscle strain to a complete tendon rupture. Although this prevents us from differentiating diagnoses by type or severity of injury, our use of the procedural codes typically assigned to a quadriceps tendon repair or reconstruction allowed for a subanalysis of quadriceps tendon tears that were severe enough to warrant surgical intervention.

Fourth, our results may have been affected by unmeasured confounders that could influence experiencing quadriceps injury and treating it. We were unable to elucidate the indication for the testosterone prescription, the exact testosterone regimen the patients were prescribed, and the patients’ serum testosterone levels at the time of the quadriceps injury. We were also unable to account for the activity level of each patient. Active patients are likely at an increased risk of experiencing a quadriceps injury compared with those who live a sedentary lifestyle. This limitation, however, may offset the fact that patients who live a more sedentary lifestyle have decreased muscle mass, placing them at greater risk of a quadriceps tendon injury. Additionally, the decision to repair a tendon injury is influenced by patient preference, surgeon skill or preference, chronicity versus acuity, and function—information we could not obtain from an administrative claims database. Nonetheless, using propensity score matching and multivariable logistic regression, we could effectively balance other measured confounders including patient demographics and comorbidities between the two groups. Although there are several unmeasured variables that we were unable to include in our analysis, we did not consider that these confounders eliminated the association between exogenous testosterone use and quadriceps injury risk.

Fifth, there is also the chance that patients who previously used exogenous testosterone to reach supraphysiologic levels are now being prescribed testosterone to treat the resultant hypogonadism. However, because of the large sample size, it is unlikely that this had a significant impact on the presented results. Lastly, although the patients in this study represent a large patient cohort, the data were generated using a single insurance provider, and therefore, it may not be a representative sample of patients with other insurance providers such as Medicare or Medicaid.

Association of Testosterone With Increased Odds of Quadriceps Injury

Receiving a prescription for testosterone was associated with a substantially higher risk of subsequent quadriceps muscle and tendon injury. In particular, males filling testosterone prescriptions were found to have a more consistent association with quadriceps injury than seen in female patients prescribed testosterone. Female patients were not at an increased risk of injury during the 1-year follow-up period; however, when evaluated for injury risk any time after the initial 3 months of filling testosterone prescriptions, an association was observed. Furthermore, there was no difference in the odds of surgical repair of the quadriceps tendon between the females who took testosterone and the control cohort, indicating that filling testosterone prescriptions may not be a risk factor for complete tendon rupture in females. Clayton and Court-Brown [4] reported a male-female ratio of 4.2:1 for quadriceps tendon rupture incidence in the United Kingdom. This distribution is similar to the male-female ratio of quadriceps injury in our study, which ranged from 8 to 1 in the control cohort to 9 to 1 in the testosterone group. We suspect a lower incidence of quadriceps muscle and tendon injury in females overall is one of the potential factors contributing to the absence of this association.

Association of Testosterone With Increased Odds of Quadriceps Tendon Repair

Prescription testosterone also was associated with a much higher risk of surgical repair of the quadriceps tendon within 1 year of injury. In a systematic review of anabolic steroids and tendon injury, Jones et al. [10] discussed a few proposed mechanisms to explain the relationship between anabolic androgenic steroids and tendon rupture, which include the biomechanical, structural, and biological effects of testosterone. They described that the increased stiffness and decreased elasticity seen in tendons with short-term anabolic androgenic steroid exposure contribute to the elevated risk of injury with activity. Alternate theories, which may contribute in a synergistic fashion, state that steroids lead to rapid muscular hypertrophy without allowing for adequate tendon compensation or that anabolic androgenic steroids directly damage the collagen structure of tendons at a molecular level, both of which make tendons vulnerable to rupture [10, 11]. The most common indication for prescribing exogenous testosterone is hypogonadism, a condition that presents with an array of symptoms including decreased libido and energy, as well as decreased bone and muscle mass [18]. Therefore, patients in our study may have hypogonadism and thus a potential for decreased muscle strength before treatment compared with their matched counterparts, possibly contributing to the higher odds of quadriceps injuries observed. A review of risk factors for lower extremity injury by Murphy et al. [15] described multiple studies that have demonstrated an association between decreased muscle strength and knee or leg injury. In addition, elevating testosterone levels in androgenic-deficient patients plays a role in weight loss, both as a function of restored vigor and motivation to return to activity, as well as the molecular effects on adipose tissue [12]. Nevertheless, because of the limited research conducted on this topic, we are unable to determine whether the quadriceps injury risk in testosterone prescription recipients in the present study is attributable to the same mechanisms that result from supraphysiologic dosing of testosterone seen in anabolic androgenic steroid users. As the rates of testosterone replacement therapy use continue to increase, it remains worthwhile to further explore the musculoskeletal risks associated with prescription testosterone use.

Conclusion

In a large-sample, retrospective database analysis, the present study demonstrated that patients who filled a prescription for testosterone replacement therapy were much more likely to experience a quadriceps muscle or tendon injury within 1 year of filling their prescription. This population was also at increased risk of undergoing surgical repair of the quadriceps tendon. Providers should use these findings to aid in determining which patients are candidates for testosterone replacement therapy, considering injury history and comorbidities that may predispose patients to quadriceps injury. Additionally, physicians who prescribe testosterone should counsel patients on the substantially increased odds of associated quadriceps tendon injury and provide appropriate education and guidance on safe exercise practices and injury prevention techniques. Given the limitations of our investigation, future prospective studies measuring actual use and serum levels of testosterone as well as patient activity levels may provide greater insight into the clinical significance of this association between testosterone replacement therapy and musculoskeletal health.