Effect of Comorbidities on Short-Term Postoperative Complications Following total Shoulder Arthroplasty: A TriNetX Study

Abstract

Purpose:

Although nonmodifiable medical comorbidities are known to increase postoperative risk following total shoulder arthroplasty (TSA), the impact of potentially modifiable behavioral risk factors is less established. This study evaluated the effects of ischemic heart disease (IHD), type II diabetes mellitus (T2DM), cannabis use disorder (CUD), and nicotine dependence (ND), on short-term postoperative complications following TSA.

Methods:

This retrospective cohort study used the TriNetX US Collaborative Network to identify patients undergoing primary TSA from 2004 to 2024 using CPT codes. Patients were stratified into cohorts according to ICD-10-CM diagnoses of IHD, T2DM, CUD, and ND. Propensity score matching (1:1) was done to balance baseline demographics and BMI. The primary outcome was 90-day complications.

Results:

Among 59,473 patients included, each comorbidity cohort showed notable increases in complications. Patients with IHD had greater odds of postoperative cardiac arrest (OR 6.825, CI, 3.513 to 13.26), acute kidney failure (OR 3.642, CI, 3.108 to 4.268), and pneumonia (OR 2.892, CI, 2.411 to 3.467). Patients with CUD had the highest odds of acute kidney failure (OR 2.539, CI, 1.556 to 4.143). T2DM was associated with increased skin infections (OR 2.151, CI, 1.806 to 2.562) and cardiac arrest (OR 3.159, CI, 1.692 to 5.898), whereas ND was associated with higher rates of pneumonia (OR 2.081, CI, 1.651 to 2.624) and acute kidney failure (OR 1.893, CI, 1.535 to 2.334).

Conclusion:

Both nonmodifiable comorbidities and modifiable risk factors markedly increase the risk of specific 90-day postoperative complications following TSA. These findings support targeted preoperative risk stratification and suggest optimizing CUD and ND may improve perioperative outcomes, although prospective studies are needed.

Total shoulder arthroplasty (TSA) is an increasingly used procedure, with the number of shoulder arthroplasties increasing more than 15-fold from 2000 to 2019.¹ There are now more than 100,000 shoulder arthroplasties performed yearly in the United States, with projections indicating continued substantial growth by 2040.^2,3 The increased incidence of TSA is multifactorial, with further adoption and advances in surgical technology and techniques, introduction of reverse TSA, expanded indications, and an aging population creating a greater need for the procedure. As the number of TSA procedures increases, preoperative optimization is paramount for mitigating postoperative complications. This is especially true when considering that the patient population tends to be older individuals who are more likely to have multiple comorbidities.^1,4

The aging population has led to an increasing proportion of multimorbid patients undergoing TSA. The prevalence of multimorbidity, defined as having two or more chronic medical conditions, is highest in patients older than 65 years, exceeding 73% of the population.⁵ With the average patient undergoing TSA being approximately 68 years, these patients are at an increased risk of postoperative complications compared with patients without any comorbidities.⁶ Ling et al reported markedly lower ASES and SAS scores in patients with multiple comorbidities, with similar findings reported by Esteras-Serrano et al and Bindi et al^7-9 In addition, patients with multiple comorbidities have higher rates of complications, increased episode of care cost, longer length of stays, and higher readmission rates following TSA, specifically those with diabetes, obesity, and cardiac disease.^10-14 Although the aforementioned comorbidities are in large part nonmodifiable risk factors, tobacco use and cannabis use are two modifiable patient behaviors that may also markedly impact their postoperative outcomes. Although the impact of risk factors such as diabetes and cardiac disease on TSA outcomes has been established, elucidating the impact of substance use disorders, particularly cannabis and tobacco, warrants further investigation.

Cannabis and tobacco are commonly used throughout the world.¹⁵ In the United States, although cigarette smoking rates have decreased substantially, overall use has remained stable due to increased utilization of alternative nicotine products like e-cigarettes.¹⁵ Following legalization in many states, the prevalence of cannabis use is increasing among adults.^16,17 Likely due to high baseline prevalence of each substance and similar routes of administration, tobacco and cannabis couse is frequently observed.¹⁸ This couse pattern is clinically relevant, as the relationship is bidirectional. A 2024 review by Yimer et al found that baseline nicotine use was associated with more than 2.5-fold increased likelihood of cannabis initiation, and baseline cannabis use is similarly associated with a 3-fold increased initiation of nicotine. Furthermore, this couse pattern has been shown to interfere with cessation efforts of either cannabis or tobacco.¹⁵ Beyond frequent co-occurrence and difficulties with cessation, both cannabis and tobacco use disorders are independently associated with increased surgical complications in orthopaedic procedures, including prolonged wound healing and increased infection and readmission rates. However, unlike nonmodifiable comorbidities such as ischemic heart disease (IHD), both substance use disorders represent potentially modifiable risk factors that can be targeted through preoperative optimization programs.

Cannabis use disorder (CUD), involving the loss of control over cannabis use despite adverse consequences, affects approximately 36% of daily or near-daily users and is further associated with multiple medical and psychiatric comorbidities. Legalization of cannabis for medical and recreational use has also brought about increased use of high potency cannabis and higher rates of CUD.¹⁷ Despite growing recognition of cannabis-tobacco couse as a public health concern, the specific impact of these substance use disorders on postoperative complications following TSA remains understudied.

Given the aging surgical population, the high prevalence of multimorbidity patients, and increasing rates of both cannabis legalization and use, understanding the effect of such comorbidities on postoperative outcomes is essential for risk stratification and optimization before TSA. Therefore, the purpose of this study is to compare the impact of nonmodifiable comorbidities, including IHD and type II diabetes mellitus (T2DM), with modifiable risk factors, including CUD and nicotine dependence (ND), on short-term postoperative complications following TSA using the TriNetX database.

Methods

TriNetX is a large deidentified patient database with data from over 61 United States Healthcare Organizations (HCOs).^19,20 It contains deidentified aggregate patient information covering procedures, diagnoses, medications, vitals, genomics, and demographics (TriNetX citations). Because the TriNetX database does not involve patient identifiable information, it is exempt from institutional review board review and approval. HCOs involved in the TriNetX network contribute healthcare data in deidentified, pseudoanonymized, or limited data set formats, following local privacy regulations. These HCOs authorize the usage of these data for research purposes on the TriNetX platform. In return for providing data, HCOs incur no financial expenses and gain access to data query tools, analytics, visualization capabilities, and the necessary hardware for software execution. The deidentification process conforms to Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule standards, as verified by a qualified expert, meeting the requirements of Section §164.514(b)(1), ensuring HIPAA compliance.

The TriNetX US Collaborative Network was retrospectively queried on April 16, 2024. Patient cohorts were created using the Current Procedural Terminology (CPT) and International Classification for Disease, 10th Edition (ICD-10) codes (Table 1). IHD was identified using ICD-10 codes I20-I25, and T2DM with E11. ND and CUD were identified using ICD-10 codes F17 and F12, respectively. These diagnostic codes are based on clinical criteria reflecting a pattern of problematic use resulting in signs of impairment or distress (eg, tolerance, withdrawal, loss of control). Information on the frequency, quantity, or duration of substance use was not provided in the TriNetX database; thus, dose-dependent effects of either substance on postoperative outcomes was not assessed in this study. The cohort of TSA patients were stratified by comorbidity (IHD, CUD, T2DM, and ND) to create eight total cohorts (Figure 1).

Table 1.

CPT and ICD-10 Codes of TriNetX Inputs

Item	Codes
Procedure
Total shoulder arthroplasty	23472
Comorbidity
Ischemic heart disease	I20-I25
Diabetes mellitus type 2	E11
Cannabis use disorder	F12
Nicotine dependence	F17
Complications
Transfusion	36430
Diseases of veins, lymphatics, lymph nodes	I80-I89
Venous embolism and thrombosis	I82
Pulmonary embolism	I26
Postprocedural hematoma	M96.84
Infections of skin and subcutaneous tissue	L00-L08
Cardiac arrest	I46
Pneumonia	J18
Acute kidney failure	N17
Aplastic and other anemias and other bone marrow failure syndromes	D60-D64
Nerve injury	S44
Periprosthetic joint infection	T84.59XA
Subsequent hospital inpatient or observation care	1013668

Figure 1.

Flowchart showing breakdown of patient selection process from TriNetX database with the numbers of patients in each cohort before and after matching.

Statistical analyses were done through the TriNetX platform. Propensity score matching was done using 1:1 nearest neighbor matching with a caliper of 0.1 standard deviations to balance baseline characteristics. For each comparison, cohorts were matched on age, sex, ethnicity, race, body mass index (BMI), and the three comorbidities not being studied. For example, in the IHD comparison, cohorts were matched on age, sex, ethnicity, race, BMI, T2DM, CUD, and ND. Measures of association analyses were done within the TriNetX platform to compare the risk of experiencing postoperative complications between cohorts with and without each morbidity. These complications were limited to within 90 days after the day of surgery. The defined complications were transfusion, diseases of veins, lymphatics, lymph nodes, venous embolism and thrombosis, pulmonary embolism, postprocedural hematoma, infections of skin and subcutaneous tissue, cardiac arrest, pneumonia, acute kidney failure (AKF), aplastic and other anemias and other bone marrow failure syndromes, nerve injury, periprosthetic joint infection, and subsequent hospital inpatient or observation care. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated by TriNetX to assess the odds of each outcome between the cohorts. A P value less than the alpha of 0.05 was considered significant.

Results

The total population of 59,473 TSA patients had an average age of 73 years, with 29,624 women (52.3%) and 24,474 men (42.2%). Most patients identified as White (81.7%), with the next most common identified race being Black or African American (6.1%) (Table 2). Of the total population of patients who underwent TSA, 30% had IHD, 2.4% had CUD, 27.8% had T2DM, and 16% were diagnosed with ND (Figure 1).

Table 2.

Patient Demographics for TSA in the US Collaborative Network in TriNetX

Demographic Characteristic	Number of Patients	Percent of Total Population
Total patients	59,473	—
Mean age (SD)	73 (10)	—
Female	29,624	52.29
Male	24,474	42.20
Not Hispanic or Latino	47,667	80.15
Hispanic or Latino	2099	3.53
White	48,607	81.73
Black	3604	6.06
Asian	493	0.83

Ischemic Heart Disease

Most patients with IHD identified as White (83.4%) and male (48.5%), compared with the non-IHD group identifying at White (81.5%) and female (55.2%). The average ages were 71.8 ± 8.66 years in the IHD group and 67.5 ± 9.81 years in the non-IHD group.

Aplastic and other anemias and other bone marrow failure syndromes (13.7%) as well as subsequent hospital inpatient or observation care (28.3%) were the most frequent complications in the IHD cohort. Postoperative cardiac arrest and AKF were markedly higher in patients with IHD (cardiac arrest OR 6.825, CI, 3.513 to 13.26; AKF OR 3.642, CI, 3.108 to 4.268, P < 0.0001). Notably, IHD patients were at a decreased risk of developing nerve injury when compared with the non-IHD cohort (OR 0.454, CI, 0.215 to 0.959, P = 0.0386). Furthermore, IHD patients experienced a higher risk of pulmonary embolism and pneumonia following surgery (pulmonary embolism OR 2.696, CI, 2.169 to 3.35; pneumonia OR 2.892, CI, 2.411 to 3.467, P < 0.0001). No notable differences observed for postprocedural hematoma with or without preoperative IHD (P = 0.0717; Table 3; Figure 2).

Table 3.

Rates of Complications Following TSA

Complication	Ischemic Heart Disease			Diabetes Mellitus Type 2			Cannabis Use Disorder			Nicotine Dependence
	OR	95% CI	P	OR	95% CI	P	OR	95% CI	P	OR	95% CI	P
Transfusion	2.453	(1.918-3.139)	<0.0001	1.296	(1.023-1.641)	0.0317	1.362	(0.68-2.727)	0.3832	1.182	(0.863-1.62)	0.2980
Diseases of veins, lymphatics, lymph nodes	1.755	(1.563-1.97)	<0.0001	1.508	(1.344-1.692)	<0.0001	1.239	(0.84-1.826)	0.2793	1.11	(0.952-1.295)	0.1824
Venous embolus and thrombosis	1.653	(1.408-1.941)	<0.0001	1.3	(1.106-1.528)	0.0015	1.118	(0.655-1.907)	0.6831	1.201	(0.963-1.498)	0.1040
Pulmonary embolism	2.696	(2.169-3.35)	<0.0001	1.388	(1.116-1.725)	0.0032	0.84	(0.43-1.641)	0.6104	1.67	(1.243-2.245)	0.0007
Postprocedural hematoma	1.518	(0.964-2.39)	0.0717	1.334	(0.866-2.057)	0.1914	1	(0.415-2.41)	1.0000	0.875	(0.487-1.572)	0.6546
Infections of skin and subcutaneous tissue	1.915	(1.609-2.279)	<0.0001	2.151	(1.806-2.562)	<0.0001	1.064	(0.653-1.734)	0.8034	1.646	(1.315-2.06)	<0.0001
Cardiac arrest	6.825	(3.513-13.26)	<0.0001	3.159	(1.692-5.898)	0.0003	1	(0.415-2.41)	1.0000	2.002	(0.937-4.28)	0.0733
Pneumonia	2.892	(2.411-3.467)	<0.0001	1.508	(1.263-1.8)	<0.0001	1.332	(0.831-2.136)	0.2342	2.081	(1.651-2.624)	<0.0001
Acute kidney failure	3.642	(3.108-4.268)	<0.0001	2.391	(2.05-2.788)	<0.0001	2.539	(1.556-4.143)	0.0002	1.893	(1.535-2.334)	<0.0001
Aplastic and other anemias and other bone marrow failure syndromes	2.003	(1.858-2.158)	<0.0001	1.512	(1.403-1.63)	<0.0001	1.303	(1.015-1.672)	0.0375	1.44	(1.3-1.596)	<0.0001
Nerve injury	0.454	(0.215-0.959)	0.0386	0.823	(0.406-1.671)	0.5904	1	(0.415-2.41)	1.0000	1	(0.416-2.404)	1.0000
Periprosthetic joint infection	1.41	(1.069-1.861)	0.0151	1.382	(1.06-1.802)	0.0167	1.127	(0.572-2.218)	0.7302	1.267	(0.921-1.744)	0.1462
Subsequent hospital inpatient or observation care	1.602	(1.521-1.688)	<0.0001	1.436	(1.363-1.514)	<0.0001	1.026	(0.868-1.213)	0.7650	1.259	(1.176-1.348)	<0.0001

Figure 2.

Forest plots of odds ratios and confidence intervals of matched cohorts to evaluate postoperative complications within 90 days of surgery.

Diabetes Mellitus

Before matching, the diabetes and the nondiabetes cohorts had similar age distributions with the average ages of each cohort being 69.4 ± 8.78 and 68.4 ± 9.91, respectively. Similarly, the sex distribution was comparable between the two cohorts with women making up 52.6% of the diabetes cohort and 52.7% of the nondiabetes cohort. Both cohorts were made up predominantly of people identifying as White (77.6% and 83.8%).

Diabetes has an increased odds ratio for infections of skin and subcutaneous tissue (OR 2.151, CI, 1.806 to 2.562), periprosthetic joint infection (OR 1.382, CI, 1.06 to 1.802), pneumonia (OR 1.508, CI, 1.263 to 1.8), deep vein thrombosis (OR 1.3, CI, 1.06 to 1.528), and pulmonary embolism (OR 1.39, CI, 1.116 to 1.725). Diabetic patients had higher readmission rates, cardiac arrests, postoperative transfusions, and acute renal failure (Table 3; Figure 2).

Cannabis Use Disorder

The CUD cohort comprised 55.4% males and 38.5% females, with an average age of 61 ± 9.9 years. White (70.5%) and Black or African American (16.7%) were represented in the CUD cohort. The non-CUD group was more female (53.1%) than male (42.7%). The average age in the non-CUD group was 68.8 ± 9.51 years with the largest represented race being White (82.3%).

CUD was associated with higher rates of AKF (OR 2.539, CI, 1.556 to 4.143, P = 0.0002). Higher incidences of aplastic and other anemias and other bone marrow failure syndromes were observed following surgery among patients with CUD as compared with non-CUD patients, as indicated by the odds ratio of 1.303 and CI, 1.015 to 1.672 (P = 0.0375). No other notable differences were identified in complication rates with and without CUD, indicating similar risks across both groups for the remaining complications (P > 0.05; Table 3; Figure 2).

Nicotine Dependence

Within the ND cohort, the average age was 64.1 ± 9.72 years old with 49.7% being male and 45.4% being female. The largest race was White, making up 78.4% of the cohort. In the non-ND cohort, the average age was 69.5 ± 9.33 years with 41.7% being male and 54.1% being female. Again, the largest race represented was White, making up 82.7% of the cohort.

ND was associated with significant increases in pulmonary complications, including pneumonia (OR 2.081, CI, 1.651 to 2.624) and pulmonary embolism (OR 1.67, CI, 1.243 to 2.245), and superficial infections (OR 1.646, CI, 1.315 to 2.06). The risk of deep vein thrombosis was not notable (OR 1.201, CI, 0.963 to 1.498) nor the risk of periprosthetic joint infection (OR 1.267, CI, 0.921 to 1.744). Subsequent hospital care in the ND group had an odds ratio of 1.259 and CI (1.176 to 1.348), indicating greater odds of needing readmission following surgery (Table 3; Figure 2).

Discussion

As the population ages and indications for TSA increases, the presence of nonmodifiable medical comorbidities and modifiable risk factors are important considerations when counseling patients on arthroplasty procedures. The purpose of this study was to evaluate the relationship between medical comorbidities, in particular ND and CUD, on postoperative complications and readmissions following TSA. Using TriNetX, the results of this study support the hypothesis that among patients undergoing TSA, patients with medical comorbidities, including IHD, T2DM, CUD, and ND, experienced elevated rates of complications within 90 days of surgery.

The population growth during the second half of the 20th century in combination with medical advancements, public health initiatives, and technological revolution has resulted in a larger global population than ever before with the proportion of individuals older than 60 years to be at an all-time high. The proportion of individuals older than 60 years has nearly doubled over the past half century, with estimates that the proportion of the population older than 60 years will be 32% by 2050.²¹ As patients live longer, surgeons are faced with higher rates of multimorbid patients, those with two or more chronic conditions, with nearly two thirds of patients older than 65 years now multimorbid. Two of the most common comorbidities include IHD and T2DM, with a reported prevalence of 38.1% and 35.6%, respectively. The results of this study are in line with the current shoulder arthroplasty literature and found patients with IHD and T2DM to be more likely to suffer postoperative complications, including DVT/PE, superficial and periprosthetic infection, cardiac arrest, kidney failure, pneumonia, and rate of readmission within 90-days. Duey et al²² found similar results in a study using the National Readmission Database in patients undergoing TSA having higher rates of readmission, postoperative infection, and DVT. Oladeji et al²³ reported the most likely predictors of postoperative myocardial infarction following TSA to include heart disease and diabetes, in line with the results of this study. In addition, in the hip and knee surgery literature, IHD and T2DM have been linked to prolonged length of stay, acute kidney injury, need for transfusion, and thromboembolic events, and a nearly three times higher rate of postoperative myocardial infarction and 54% greater odds of mortality within 1 year.^24-27 The development of osteoarthritis, a common indication for shoulder arthroplasty, has also been associated with increased risk of development of IHD and is likely the result of decreased activity level and mobility due to pain.²⁸ This creates a dilemma for surgeons as patients with severe osteoarthritis may struggle to improve their cardiovascular health due to their limited mobility, creating a challenging scenario for preoperative conditioning and optimization.

The selection of ND and CUD as focal modifiable risk behaviors in this study was driven by several factors. For one, unlike nonmodifiable comorbidities like IHD and T2DM, substance use disorders represent specific targets for behavioral intervention during preoperative optimization. In addition, these risk factors are increasingly relevant to TSA outcomes due to the common use of these substances across age groups and within arthroplasty patient populations.²⁹ Self-reported cannabis use in patients undergoing total joint arthroplasty increased from 1% to 11% following legalization, and documentation of substance use disorder is increasing in patients undergoing shoulder procedures.³⁰ In addition, the bidirectional relationship between cannabis and tobacco use suggests that these substances may have synergistic effects on surgical outcomes that warrant investigation.³¹

Although comorbidities like IHD and DM are nonmodifiable lifelong diseases, nicotine use and cannabis use are modifiable risk factors affecting the rate of postoperative outcomes. Smoking has long been known to be one of the most damaging behaviors an individual can partake, increasing the rate of periprosthetic joint infection, medical and wound complications, requiring higher rates of revisions, and readmissions postoperatively.^32-38 Fortunately, the rate of smoking in the United States is near an all-time low; however, the use of nicotine has markedly increased with the advent of new nicotine pouches and flavored electronic cigarettes. Patel et al³⁹ found a 3% increase in the use of oral nicotine pouches from September 2020 to May 2022, with a trend toward even further increased use. Even nontobacco nicotine use is associated with postoperative complications, with the lower extremity literature finding higher rates of wound complication, infection, deep vein thrombosis/pulmonary embolism, and periprosthetic joint infection (PJI) following total knee arthroplasty.⁴⁰ Similarly, higher rates of wound problems were found following Achilles tendon repair, and patients remain at elevated risk within 1 year of cessation of nicotine.^41,42 However, there is little literature examining the use of nontobacco nicotine use and impact on TSA. This study found patients with ND to have a markedly greater risk of PE, superficial infection, pneumonia, acute kidney failure, anemia, and readmission. These findings are in line with the current published literature finding nicotine use to be associated with increase postoperative narcotic use, early loosening and need for revision, and higher rates of PJI, wound, and medical complications.^{33,34,38,43,44}

Although viewed by many to be “safer” than smoking cigarettes, the literature presents mixed results regarding the impact of cannabis use in the perioperative window on postoperative complications. Remily et al found a higher risk of PJI in cannabis users following TSA, whereas Rahmon et al found a cannabis user to have a shorter length of stay and increased rate of discharge to home after arthroplasty procedures.^45,46 The results of our study found that patients with CUD were at a greater risk of acute renal failure and anemia postoperatively. The observed association between CUD and acute renal failure (OR 2.539) aligns with recent evidence showing cannabis users facing increased rates of AKI after major elective surgery.^47,48 Proposed mechanisms include cannabinoid hyperemesis syndrome leading to severe dehydration and prerenal injury, although the inability to distinguish synthetic from natural cannabinoids in administrative data limits mechanistic conclusions.^49,50 This finding warrants further investigation with prospective studies with detailed substance use characterization. The increased rates of postoperative anemia (OR 1.303) in the CUD cohort may be reflective of perioperative blood loss, nutritional deficiencies associated with substance use disorders, or unmeasured confounders rather than a direct hematologic effect of cannabis. Thus, due to multiple possible contributors, further investigation is required on this association.

In the hip and knee arthroplasty literature, cannabis use has been associated with higher odds of DVT, PE, cardiac complications, and revisions without a decrease in opioid use postoperatively.⁴⁷ Similar results were reported in the spine literature with cannabis use found to be an independent risk factor for hospital readmission within 90 days following posterior cervical fusion.⁵¹ Bhattacharjee et al⁵² reported increased risk of DVT, PE, and infection in patients using cannabis who underwent knee and shoulder arthroscopy. Cannabis use has been linked with an increased risk of aseptic revision and periprosthetic joint infection following TSA, but this was not demonstrated in the results of this study.^46,53 This may be in part to the scope of this study being limited to the first 90 days postoperatively. There are studies indicating potential benefits of cannabis use in the perioperative period; Moon et al⁵⁴ found a notable decrease in mortality and stroke in cannabis users undergoing total knee, hip, and shoulder arthroplasty. Although cannabis has some analgesic effect, in the perioperative period, cannabis users require greater amounts of anesthesia, higher levels of postoperative pain, and require greater amount of narcotics for pain control after surgery.^55,56

Based on emerging evidence demonstrating increased complications across multiple orthopaedic procedures,^44,57-59 cannabis use should be formally incorporated into preoperative risk assessment like tobacco use. The American Society of Regional Anesthesia and Pain Medicine recommends universal screening for cannabis use, including dose, frequency, route of administration, and time of last use during preoperative evaluation.⁶⁰ Current guidelines also recommend counseling patients on the impact of cannabis use on postoperative pain control in patients undergoing surgery.⁶¹ Former tobacco users have displayed comparable postoperative outcomes as nonsmokers after TSA in measures of pain scores, complication rates, and functional outcomes, indicating that preoperative tobacco cessation can improve outcomes.^37,62 However, specific evidence-based protocols for cannabis cessation timing before TSA remain undefined and require prospective investigations. ND and CUD are modifiable risk factors that, if addressed before surgery, may allow surgeons to optimize patients and limit preventable complications.⁴⁴

Using the TriNetX database in this study has several limitations. The first is a limited level of detail available according to ICD-10 diagnosis and retrospective data in TriNetX. For example, the extent of damage due to a “periprosthetic joint infection” is unknown. Those with a superficial skin infection and those with deeper infections will be placed in the same category. In addition, it was not always possible to further classify patients within each comorbidity by the extent of their condition. For instance, controlled and uncontrolled T2DM are placed in the same cohort by the program. The TriNetX database does not provide information on the frequency, quantity, or duration of substance use. Therefore, dose-dependent effects of nicotine or cannabis on postoperative outcomes could not be assessed in this study.

In addition, patients in this study likely have multiple risk factors present at one time, which could confound the ability to independently assess the impact of each comorbidity. Propensity score matching was done to balance cohorts on the three other comorbidities being studied such that, for example, in the ND comparison, cohorts were matched on baseline age, sex, ethnicity, race, BMI, IHD, T2DM, and CUD. Still, residual confounding from the co-occurrence of multiple risk factors cannot be entirely excluded. The bidirectional relationship between cannabis and tobacco use with frequent co-use patterns is particularly relevant and these substances may be synergistic with one another on postoperative outcomes that was not captured in this analysis.

The results of this study are also limited by the level of detail in a patient's chart and how a physician chooses to code-specific procedures and diagnoses. Finally, TriNetX is an active database and is constantly being updated, causing the sizes of each cohort to change over time. All the queries in this study were run on the same day, to minimize the impact of the changing database.

Even with its limitations, this study used a large, national cohort of TSA patients and details the impact of many major comorbidities. Many studies investigate a single comorbidity or a single complication, but the use of TriNetX allowed for the investigation into four comorbidities, specifically including ND and CUD, and 13 postoperative complications following TSA. Building on the findings of this study, future studies can explore the impact of multimorbidity on postoperative complication and recovery, given the climbing rates of comorbidity and multimorbidity in the United States. Research establishing evidence-based alterations to treatment protocol to improve patient recovery is needed to provide optimal patient care.

Conclusion

Both nonmodifiable comorbidities (IHD and T2DM) and modifiable risk factors (CUD and ND) markedly increase 90-day postoperative complications following TSA. ND was associated with increased pulmonary complications, superficial infections, and readmissions, whereas CUD was associated with acute kidney failure and anemia. These findings support universal preoperative screening for substance use and targeted efforts for optimization of modifiable risk factors. Surgeons should counsel patients on documented risks and provide cessation resources during preoperative evaluation for TSA. Shared decision making regarding modifiable risk factors can help optimize patients for surgery and improve perioperative outcomes.