2.1 Study design and data source

We conducted a retrospective cohort study to evaluate the association between pre-operative hypothyroidism and postoperative outcomes following primary total knee arthroplasty (TKA) using de-identified electronic health records (EHR). Data were sourced from the TriNetX Research network (TriNetX LLC, Cambridge, MA), a US-based federated research platform comprising aggregated EHR data from 102 participating healthcare organizations (HCOs). This analysis included data available up to the query date of February 2, 2025. The database supports queries using standard coding systems including the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), Current Procedural Terminology (CPT), and Systematized Nomenclature of Medicine Clinical Terms (SNOMED-CT) codes. As this study relied solely on aggregated, de-identified patient records, it was deemed exempt from Institutional Review Board approval.

2.2 Participants

Adults (18–100 years) undergoing primary TKA were identified via CPT code 27447 and relevant ICD-10-PCS/SNOMED codes (e.g., 0SRC0JA, 0SRC069, 19063003). Patients required a documented visit code at least two years post-TKA. The Hypothyroidism Cohort included those with a diagnosis of hypothyroidism (ICD-10-CM: E03.9) within six months prior to surgery. Patients with any history of hyperthyroidism (ICD-10-CM: E05) were excluded. Controls had no history of thyroid dysfunction.

2.3 Propensity score matching

To reduce confounding, 1:1 PSM was performed using logistic regression and greedy nearest-neighbor matching. Prior to matching, the cohorts comprised 36,094 hypothyroid and 222,743 control patients. Covariates included demographics (age, sex, race, ethnicity) and comorbidities: hypertension, chronic kidney disease, tobacco use, obesity, diabetes, hip osteoarthritis, bone disorders, heart failure, and liver disease. Post-matching, each cohort included 35,318 patients. Balance was assessed using standardized mean differences (SMD), with SMD <0.1 indicating adequate matching.

2.4 Outcomes

Postoperative outcomes were evaluated within two predefined intervals following the index TKA: 1–90 days and 1–730 days (2 years). Outcome events were identified using relevant ICD-10-CM, CPT, and SNOMED-CT codes defined within the TriNetX platform. Short-term (90-day) and longer-term (2-year) medical complications included: infection following procedure (general), sepsis, pulmonary embolism (PE), deep vein thrombosis (DVT), acute myocardial infarction (MI), mortality, hospital readmission, acute renal failure, superficial surgical site infection (SSI), deep surgical site infection/periprosthetic joint infection (PJI), and emergency department (ED) visit. Mechanical and implant-related complications assessed included aseptic loosening, revision TKA, periprosthetic fracture (PPF), implant failure (general mechanical complication), and dislocation/instability. Patients were included in the risk analysis regardless of whether they had the outcome prior to the time window. A detailed list of codes defining each outcome is provided in Supplementary Table 1.

2.5 Statistical analysis

Analyses were conducted within TriNetX on the matched cohorts. Baseline characteristics were compared using t-tests and chi-squared tests, with SMDs to assess balance. Risk ratios (RRs), risk differences (RDs), and 95 % confidence intervals (CIs) were calculated for each outcome. Time-to-event analyses for mechanical complications used Kaplan-Meier survival curves and log-rank tests to derive hazard ratios (HRs). Counts between 1 and 9 were masked as 10* for privacy.

3.1 Patient cohort identification and matching

We initially identified 258,837 patients meeting eligibility criteria for primary TKA with at least two years of follow-up from 102 HCOs within the TriNetX Research network. Prior to propensity score matching, this included 36,094 patients in the hypothyroidism cohort and 222,743 patients in the control cohort. Before matching, patients comprising the hypothyroidism cohort were significantly older (mean 68.2 vs 65.1 years, p < 0.001, SMD = 0.305), more likely to be female (77.5 % vs 55.0 %, p < 0.001, SMD = 0.489), and exhibited a higher baseline prevalence of several comorbidities compared to controls, including hypertensive diseases (48.6 % vs 36.5 %, p < 0.001, SMD = 0.247), diabetes mellitus (17.7 % vs 12.5 %, p < 0.001, SMD = 0.147), overweight/obesity (20.1 % vs 13.8 %, p < 0.001, SMD = 0.169), and disorders of bone density and structure (12.9 % vs 6.6 %, p < 0.001, SMD = 0.215).

3.2 Baseline characteristics of matched cohorts

Following 1:1 propensity score matching based on the specified demographic and comorbidity variables, each cohort consisted of 35,318 patients. Post-matching, the demographic and clinical characteristics were well-balanced between the hypothyroidism and control cohorts (Table 1). The mean age was identical between groups (68.2 years, p = 0.761, SMD = 0.002). The proportion of females was exactly matched (77.5 % in both groups, p = 1.000, SMD<0.001). Race distribution showed no significant differences, with White patients representing 82.1 % vs 82.4 % (p = 0.245, SMD = 0.009). Similarly, the prevalence of matched comorbidities was well-balanced, including diabetes mellitus (17.7 % vs 17.4 %, p = 0.239, SMD = 0.009), hypertensive diseases (48.6 % vs 48.7 %, p = 0.690, SMD = 0.003), chronic kidney disease (7.2 % vs 6.5 %, p = 0.001, SMD = 0.025), and overweight/obesity (20.1 % vs 19.9 %, p = 0.529, SMD = 0.005). All matched variables met the criterion for negligible difference (SMD <0.1), with the largest residual SMD observed for heart failure (SMD = 0.031).

3.3 90-Day postoperative outcomes

Within the 90-day postoperative period, patients in the hypothyroidism cohort experienced a significantly higher risk of several adverse events compared to the matched control cohort (Table 2). Among medical complications, hypothyroidism was associated with increased risk of ED visit (8.2 % vs 7.3 %, RR: 1.115 [95 % CI: 1.059, 1.173]; p < 0.001), deep SSI/PJI (1.7 % vs 1.2 %, RR: 1.400 [95 % CI: 1.239, 1.583]; p < 0.001), DVT (2.2 % vs 1.8 %, RR: 1.204 [95 % CI: 1.085, 1.337]; p < 0.001), sepsis (0.6 % vs 0.4 %, RR: 1.422 [95 % CI: 1.152, 1.755]; p = 0.001), acute renal failure (2.2 % vs 1.9 %, RR: 1.161 [95 % CI: 1.047, 1.287]; p = 0.005), and general postoperative infection (1.1 % vs 0.9 %, RR: 1.183 [95 % CI: 1.021, 1.371]; p = 0.025). Regarding early implant-related events, significantly higher risks were observed for general implant failure (1.9 % vs 1.4 %, RR: 1.317 [95 % CI: 1.173, 1.478]; p < 0.001), periprosthetic fractures (0.3 % vs 0.2 %, RR: 1.532 [95 % CI: 1.150, 2.041]; p = 0.003), and aseptic loosening (0.4 % vs 0.3 %, RR: 1.330 [95 % CI: 1.031, 1.717]; p = 0.028).

No statistically significant difference was observed between the hypothyroidism and control groups within 90 days for pulmonary embolism (1.1 % vs 1.0 %, RR: 1.044 [95 % CI: 0.905, 1.205]; p = 0.554), hospital readmission (1.6 % vs 1.5 %, RR: 1.078 [95 % CI: 0.958, 1.214]; p = 0.210), superficial SSI (0.2 % vs 0.2 %, RR: 1.197 [95 % CI: 0.852, 1.681]; p = 0.299), myocardial infarction (0.5 % vs 0.5 %, RR: 1.013 [95 % CI: 0.814, 1.259]; p = 0.911), mortality (<0.1 % [*10] vs <0.1 % [*11], RR: 0.909 [95 % CI: 0.386, 2.140]; p = 0.827), dislocation/instability (0.3 % vs 0.2 %, RR: 1.310 [95 % CI: 0.992, 1.731]; p = 0.057), or revision TKA (0.5 % vs 0.4 %, RR: 1.219 [95 % CI: 0.987, 1.507]; p = 0.066).

3.4 Two-year postoperative outcomes

Evaluation of outcomes extending up to two years post-TKA demonstrated persistent associations between pre-operative hypothyroidism and adverse events, particularly concerning implant survivorship and mechanical complications (Table 3). Compared to controls, the hypothyroidism cohort exhibited significantly elevated 2-year risks for aseptic loosening (1.2 % vs 0.8 %, RR: 1.526 [95 % CI: 1.310, 1.778]; p < 0.001), revision TKA (2.0 % vs 1.4 %, RR: 1.371 [95 % CI: 1.225, 1.535]; p < 0.001), periprosthetic fractures (0.7 % vs 0.5 %, RR: 1.506 [95 % CI: 1.237, 1.834]; p < 0.001), general implant failure (4.3 % vs 3.2 %, RR: 1.360 [95 % CI: 1.261, 1.467]; p < 0.001), and dislocation/instability (1.0 % vs 0.7 %, RR: 1.423 [95 % CI: 1.208, 1.677]; p < 0.001). The increased risk of deep SSI/PJI also persisted at 2 years (2.9 % vs 2.1 %, RR: 1.372 [95 % CI: 1.251, 1.505]; p < 0.001), while no significant difference was found for superficial SSI (0.4 % vs 0.3 %, RR: 1.220 [95 % CI: 0.948, 1.571]; p = 0.122).

Kaplan-Meier survival analysis corroborated these findings, showing significantly lower event-free survival rates in the hypothyroidism group for aseptic loosening (Log-Rank p < 0.001; HR: 1.528 [95 % CI: 1.311, 1.782]), revision TKA (Log-Rank p < 0.001; HR: 1.374 [95 % CI: 1.226, 1.540]), and general implant failure (Log-Rank p < 0.001; HR: 1.368 [95 % CI: 1.267, 1.478]) over the 2-year follow-up period. The proportional hazards assumption was met or borderline met for these analyses (p ≥ 0.062). Additionally, several medical complications demonstrated significantly higher risks at 2 years in the hypothyroidism group, including MI (RR: 1.122, p = 0.045), hospital readmission (RR: 1.118, p = 0.019), DVT (RR: 1.208, p < 0.001), renal failure (RR: 1.270, p < 0.001), sepsis (RR: 1.354, p < 0.001), general infection (RR: 1.261, p < 0.001) and ED visits (RR: 1.105, p < 0.001), while differences for PE (p = 0.062) and mortality (p = 0.349) remained non-significant.

4.1 Summary of findings

In this large, propensity-matched cohort study, we evaluated the impact of preexisting hypothyroidism on early (90-day) and late (2-year) outcomes following primary total knee arthroplasty (TKA). Hypothyroidism was significantly associated with a broad range of complications, including increased risks of sepsis, anemia, renal failure, and pneumonia within 90 days, and mechanical complications such as prosthetic joint infection (PJI), periprosthetic fracture, and implant failure by two years. Leveraging a large national dataset and robust matching, our study highlights hypothyroidism as a clinically meaningful risk factor influencing TKA outcomes well beyond the perioperative period.

4.2 Comparison with existing literature

Our findings regarding increased short-term risks align with previous research. Buller et al. (2018) demonstrated higher odds of 90-day complications and increased costs in hypothyroid TKA patients.¹⁰ Tan et al. (2016) identified hypothyroidism as an independent predictor of PJI in a mixed TJA population.¹⁴ Jing et al. (2021) also reported elevated 90-day complication rates in subclinical hypothyroid patients.¹² Although we could not differentiate subclinical from overt hypothyroidism, our data reinforce that disturbances in thyroid homeostasis warrant clinical attention before major orthopedic procedures.

Our study significantly expands the literature by demonstrating an association between hypothyroidism and adverse mechanical outcomes at two years. This contrasts with Tsotsolis et al. (2023), who found no significant differences in mechanical complication rates post-TJA.¹³ Differences in sample size, follow-up duration, and mechanical complication definitions may explain this discrepancy. Additionally, our findings align with recent evidence from total hip and shoulder arthroplasty cohorts,^15,16 suggesting systemic vulnerability in hypothyroid patients undergoing major joint replacement.

4.3 Biological mechanisms and pathophysiology

Thyroid hormones modulate immune function, bone turnover, and tissue healing.^7–9 Hypothyroidism may impair host defenses, predisposing to infections like PJI, while altered bone remodeling could compromise implant stability, leading to loosening and fractures. The autoimmune milieu in conditions like Hashimoto's thyroiditis may further influence bone metabolism.¹⁷ Though our dataset lacked biological markers, the observed complication patterns suggest systemic effects of hypothyroidism that could impair TKA outcomes across multiple domains.

4.4 Clinical implications

The association between hypothyroidism and increased post-TKA complications carries significant clinical implications. Firstly, these findings argue for incorporating thyroid status assessment into routine preoperative risk stratification for TKA candidates, alongside standard evaluations of cardiovascular, pulmonary, and metabolic health. Identifying patients with known or suspected thyroid dysfunction may allow for targeted optimization efforts, although it must be stressed that prospective studies are required to confirm whether achieving a specific euthyroid state preoperatively mitigates the observed risks. Secondly, based on the heightened risks found in our study, particularly for PJI and mechanical failure, clinicians should engage in explicit preoperative counseling with hypothyroid patients regarding their potentially elevated risk profile. Tailored postoperative monitoring strategies, perhaps involving closer wound surveillance or extended vigilance for signs of infection or instability, might be warranted. Shared decision-making informed by these potential risks is essential for managing patient expectations and optimizing care pathways for this potentially vulnerable population undergoing TKA.

4.5 Strengths and limitations

This study possesses several strengths, including the use of a large, nationwide database (TriNetX) encompassing a diverse patient population, enhancing the generalizability of our findings. The propensity score matching methodology employed helped to mitigate bias by balancing numerous measured baseline comorbidities and demographic factors between the hypothyroid and euthyroid cohorts. Furthermore, the inclusion of a two-year follow-up period specifically allowed for the investigation of longer-term mechanical complications, addressing a critical knowledge gap identified in the existing literature.¹³ This represents one of the most comprehensive analyses of both early medical and later mechanical TKA outcomes associated with hypothyroidism to date.

However, this study also has important limitations inherent to its retrospective, observational design using administrative data that must temper interpretation. Firstly, while demonstrating association, the study cannot establish causality. Secondly, reliance on ICD codes for identifying hypothyroidism, while necessary for large database analysis, has known limitations. We could not ascertain the etiology, duration, or biochemical severity of hypothyroidism (lacking TSH, fT4/T3 levels), nor could we confirm patient adherence to thyroid hormone replacement therapy or their functional thyroid status (e.g., euthyroid vs. inadequately treated) at the time of surgery. The “hypothyroidism” cohort therefore likely represents a heterogeneous group in terms of underlying disease activity and control, potentially masking nuances or leading to an underestimation or overestimation of the true effect depending on the distribution of disease severity and treatment adequacy within the cohort. This lack of granularity prevents analysis of potential dose-response relationships. Thirdly, details regarding the specific reasons for revision TKA or the management of observed complications were unavailable. Fourthly, administrative databases typically lack patient-reported outcome measures (PROMs), data on potentially important lifestyle factors (e.g., smoking nuances, detailed activity levels), and information regarding surgeon or institutional experience and specific surgical techniques beyond CPT codes (e.g., alignment targets, cementing techniques, implant design specifics). Finally, despite rigorous propensity score matching on measured covariates, the potential for residual confounding by unmeasured factors (such as detailed preoperative functional scores like KOOS/WOMAC, precise measures of bone mineral density or quality, physical activity level, or subtle socioeconomic influences) remains. These limitations necessitate cautious interpretation of the findings and underscore the need for further validation. Prospective studies incorporating biochemical thyroid markers and treatment adequacy are needed to determine whether perioperative optimization can mitigate these risks and improve long-term implant survivorship.