Risk Factors for Blood Transfusions in Primary Anatomic and Reverse Total Shoulder Arthroplasty for Osteoarthritis

Abstract

Background

The purpose of this study was to determine risk factors for blood transfusion in primary anatomic and reverse total shoulder arthroplasty (TSA) performed for osteoarthritis.

Methods

Patients who underwent anatomic or reverse TSA for a diagnosis of primary osteoarthritis were identified in a national surgical database from 2005 to 2018 by utilizing both CPT and ICD-9/ICD-10 codes. Univariate analysis was performed on the two transfused versus non-transfused cohorts to compare for differences in comorbidities and demographics. Independent risk factors for perioperative blood transfusions were identified via multivariate regression models.

Results

305 transfused and 18,124 nontransfused patients were identified. Female sex (p<0.001), age >85 years (p=0.001), insulin-dependent diabetes mellitus (p=0.001), dialysis dependence (p=0.001), acute renal failure (p=0.012), hematologic disorders (p=0.010), disseminated cancer (p<0.001), ASA ≥ 3 (p<0.001), and functional dependence (p=0.001) were shown to be independent risk factors for blood transfusions on multivariate logistic regression analysis.

Conclusion

Several independent risk factors for blood transfusion following anatomic/reverse TSA for osteoarthritis were identified. Awareness of these risk factors can help surgeons and perioperative care teams to both identify and optimize high-risk patients to decrease both transfusion requirements and its associated complications in this patient population.

Level of Evidence: III

Introduction

Total shoulder arthroplasty (TSA) is an effective treatment modality for various pathologies of the glenohumeral joint. Indications for anatomic TSA include avascular necrosis of the humeral head, inflammatory arthritis, proximal humerus fractures, and osteoarthritis (OA) whereas reverse TSA is typically indicated primarily for arthropathy secondary to irreparable/chronic rotator cuff tears.^1,2 With excellent clinical outcomes for both anatomic and reverse TSA in the treatment of glenohumeral OA, the popularity of TSA for the treatment of primary OA continues to increase.^3,4,5,6 Indeed, Trofa et al. report a 5.0-fold increase in TSA within a span of ten years for the treatment of glenohumeral joint OA.⁵

With both reverse and anatomic TSA’s increasing popularity, it is imperative to try and mitigate complications that may arise from this operative intervention. Minimizing blood transfusions following reverse/anatomic TSA may be one way to reduce complications to patients as blood transfusions have been associated with adverse events such as allergic reactions, delayed immune mediated reactions, iron overload, and cardiopulmonary adverse effects.⁷ Even with these potential complications, however, the rates of blood transfusion following TSA have been reported to range from 6.1% to as high as 43%.^8,9 As such, blood transfusions in TSA have been garnering increasing interest amongst surgeons.^8-16 However, to the author’s knowledge, most studies analyzing predictors of blood transfusion needs following TSA have been at single-institutions.^9-16 In addition, studies that utilize nationwide databases with larger patient samples examine risks for blood transfusion in reverse and anatomic TSA for various etiologies.^8,14 However, different etiologies of TSA requirements, such as trauma to the proximal humerus, are known to have increased rates of transfusion compared to others such as OA.^8,14 Anthony et al. had previously reported significant risk factors for morbidity and transfusions in a study analyzing TSA patients from 2005 to 2011; however, their study cohort included TSA patients with multiple etiologies, including both OA and upper extremity fractures.¹⁷ With OA being the primary leading diagnosis for TSA needs⁶, it is important to identify risk factors for blood transfusion utilizing a large sample size in the specific subset of patients who require TSA due to OA.

The current study sought to utilize the American College of Surgeons-National Surgical Quality Improvement Program (ACS-NSQIP) database to identify significant risk factors for blood transfusions following reverse/anatomic TSA for the treatment of OA from a larger cohort of patients undergoing surgery from 2005 to 2018. Although rates of blood transfusions have been reported to vary between reverse and anatomic TSA^9,11,15 we have analyzed risk factors for both reverse/anatomic TSA in aggregate similar to other studies.^10,13 In doing so, our study may shed light on how surgeons can better counsel high-risk patients, optimize management, and prevent complications secondary to blood transfusions.

Methods

This study was conducted and published in accordance with the STROBE Statement guidelines for case-control studies. As this study utilized a publicly available multiinstitutional surgical registry, no Institutional Review Board approval was required. No patient consent was required for the present study. The ACS-NSQIP database was queried to identify all patients who had undergone total shoulder arthroplasty (TSA) from 2005 to 2018 by Current Procedure Terminology (CPT) codes. Patients with CPT code 23472 were included in this study, which includes both reverse TSA and anatomic TSA. Patients who underwent shoulder hemiarthroplasties (CPT code 23470) and revision shoulder arthroplasties (CPT codes 23473 and 23474) were excluded from this study. Patients were further filtered based on ICD-9 and ICD-10 codes for diagnoses related to primary osteoarthritis. Patients were excluded if they had diagnoses related to rheumatoid arthritis, fractures/trauma, malunions, nonunions, dislocations, infections, bursae/tendon-related pathologies, ruptured rotator cuffs, traumatic arthropathies, and post-traumatic osteoarthritis. Patients with incomplete demographic and preoperative comorbidity data were also excluded to reduce the confounding effects of missing data. After all inclusion and exclusion criteria were applied, 18429 TSA patients with osteoarthritis were included in our study. These patients were then stratified into two separate cohorts: controls who had not received transfusions (n=18124) and those who received transfusions within the 72 hour period of procedure start (n=305). Transfusions were defined by the ACS-NSQIP as, “At least 1 unit of packed or whole red blood cells given from the surgical start time up to and including 72 hours postoperatively. If the patient receives shed blood, autologous blood, cell saver blood or pleurovac postoperatively, count this blood in terms of equivalent units. For a cell saver, every 500 ml’s of fluid will equal 1 unit of packed cells. If there are less than 250 ml of cell saver, round down and report as 0 units. If there are 250 cc, or more of cell saver, round up to 1 unit. The blood may be given for any reason”.^18,19

Patient demographics and preoperative comorbidities were analyzed to better characterize the TSA cohort with primary osteoarthritis. Demographic factors include age, sex, race, and body mass index (BMI). Preoperative comorbidities include diabetes mellitus, smoking history, chronic obstructive pulmonary disease (COPD), dyspnea, congestive heart failure, hypertension requiring medication management, acute renal failure, dialysis-dependence, disseminated cancer, open wound/wound infections, chronic steroid usage, hematologic disorders (i.e. Vitamin K deficiency, thrombocytopenia, hemophilias, etc.), systemic sepsis, and functional dependence (ie. requiring an orthotic device to walk, wheelchair bound, etc.). Perioperative variables such as type of anesthesia administered and American Society of Anesthesiologists (ASA) classification were also included. Complications were categorized as minor and major complications, following previously published guidelines in the orthopaedic literature.^20-23 Minor complications included superficial surgical site infections (SSI), wound dehiscence, pneumonia, urinary tract infections (UTI), and progressive renal insufficiency. Major complications included deep SSI, organ/space SSI, unplanned intubation, ventilator dependence, acute renal failure, cardiac arrest requiring CPR, myocardial infarctions, pulmonary embolisms, deep venous thromboembolisms, systemic sepsis, septic shock, unplanned readmissions, unplanned reoperations, and mortality occurring in the 30-day postoperative period.

To determine differences in patient demographic factors, preoperative comorbidities, and perioperative/postoperative outcomes, univariate analyses were first utilized in establishing differences in these variables between the transfused and non-transfused cohorts. Pearson’s chi-squared tests were implemented in analyzing categorical variables, while one-way analysis of variance (ANOVA) and independent T-tests were used to assess for differences in mean values of continuous variables such as age and time. Continuous variables are expressed as mean values with their respective standard deviations, while categorical variables are reported as a proportion representing incidence rates within the respective cohorts.

Multivariate logistic regression models were utilized in identifying significant risk factors for blood transfusions. Significantly different demographic factors and preoperative comorbidities were entered into the regression model as covariates. The predicted probabilities given by the logistic regression model were used to generate a receiver operating characteristic (ROC) curve to assess the discriminatory ability of the regression model in assigning patients into the transfused or non-transfused cohort based on the controlled variables. Post-regression diagnostics were assessed with C-statistic and the Hosmer-Lemeshow Test. All statistical findings with p-values less than or equal to 0.05 were considered significant in this analysis. All statistical analyses were performed using the IBM® SPSS® Statistics Version 25 software (IBM Corporation, Armonk, NY).

Results

18,124 patients who had undergone TSA for primary osteoarthritis were included in this study, of which 305 (1.66%) had received blood transfusions within 72 hours of the start of their procedures. The transfused cohort was significantly older (x̄=72.53, SD 9.937; p<0.001) than the control cohort (x̄=68.84, SD 9.415) and was comprised of a larger proportion of female patients (73.11% vs. 53.11%; p<0.001) than the control cohort. The transfused cohort consisted of a smaller proportion of obese patients (41.31% vs. 51.79%; p<0.001) than the control cohort. No significant differences were observed in the distribution of self-reported race/ethnicity. (Table 1)

Table 1.

DEMOGRAPHICS and COMORBIDITIES in Transfused Versus Non-Transfused Cohorts

	Control (n=18124)		Transfusion (n=305)		P-Value
	18124		305
DEMOGRAPHICS
Age (Mean ± SD)a	68.84 ± 9.415		72.53 ± 9.937		<0.001
Age (years)					<0.001
x ≤ 55	1513	8.35%	15	4.92%
55 < x ≤ 65	4606	25.41%	49	16.07%
65 < x ≤ 75	7401	40.84%	108	35.41%
75 < x ≤ 85	4218	23.27%	111	36.39%
85 < x	386	2.13%	22	7.21%
Sex					<0.001
Female	9626	53.11%	223	73.11%
Male	8498	46.89%	82	26.89%
Race/Ethnicity					0.838
American Indian Alaska Native	or 70	0.39%	0	0.00%
Asian or Pacific Islander	115	0.63%	3	0.98%
Black or African American	842	4.65%	16	5.25%
Hispanic	4	0.02%	0	0.00%
White or Caucasian	15431	85.14%	259	84.92%
Other	1662	9.17%	27	8.85%
Body Mass Index (kg/m2)					<0.001
Normal	2856	15.76%	75	24.59%
Overweight	5881	32.45%	104	34.10%
Class I Obese	4848	26.75%	70	22.95%
Class II Obese	2622	14.47%	29	9.51%
Class III Obese	1917	10.58%	27	8.85%
PRE-OPERATIVE COMORBIDITIES
Diabetes Mellitus					<0.001
No Diabetes Mellitus	15157	83.63%	232	76.07%
Non-Insulin Dependent	2168	11.96%	46	15.08%
Insulin Dependent	799	4.41%	27	8.85%
Smoking History	1803	9.95%	30	9.84%	0.948
Dyspnea					0.015
No Dyspnea	16924	93.38%	274	89.84%
Moderate Exertion	1143	6.31%	28	9.18%
At Rest	57	0.31%	3	0.98%
Ventilator Dependence	0	0.00%	0	0.00%	-
COPD	1103	6.09%	34	11.15%	<0.001
Ascites	0	0.00%	1	0.33%	0.017
Congestive Heart Failure	74	0.41%	4	1.31%	0.040
Hypertension	12095	66.73%	225	73.77%	0.010
Acute Renal Failure	7	0.04%	2	0.66%	0.009
Dialysis	50	0.28%	6	1.97%	<0.001
Disseminated Cancer	26	0.14%	4	1.31%	0.001
Open Wounds/Wound Infections	58	0.32%	3	0.98%	0.080
Chronic Steroid Use	779	4.30%	16	5.25%	0.419
Weight Loss	25	0.14%	1	0.33%	0.352
Bleeding Disorders	432	2.38%	20	6.56%	<0.001
Transfusions (within 72 hours preop)	14	0.08%	3	0.98%	0.003
Systemic Sepsis	45	0.25%	2	0.66%	0.222
Functional Status					<0.001
Independent	17678	97.54%	283	92.79%
Partially Dependent	314	1.73%	18	5.90%
Totally Dependent	15	0.08%	2	0.66%

^aValues expressed as Mean ± Standard Deviation (SD) All other values expressed as (%) and N COPD: Chronic obstructive pulmonary disease

A larger proportion of the transfused cohort presented with diabetes mellitus (p<0.001), dyspnea (p=0.015), COPD (p<0.001), ascites (p=0.017), congestive heart failure (p=0.040), hypertension requiring medication management (p=0.010), acute renal failure (p=0.009), dialysis (p<0.001), disseminated cancer (p=0.001), hematologic disorders (p<0.001), preoperative blood transfusions (p=0.003), and functional dependence (p<0.001). (Table 1)

On multivariate logistic regression analyses, numerous risk factors for blood transfusions were identified. Age (> 85 years) (OR 3.192, 95% CI 1.604-6.350; p=0.001), female sex (OR 2.258, 95% CI 1.738-2.935; p<0.001), insulin-dependent diabetes mellitus (IDDM) (OR 2.045, 95% CI 1.331-3.141; p=0.001), acute renal failure (OR 9.178, 95% CI 1.630-51.684; p=0.012), dialysis dependence (OR 4.816, 95% CI 1.874-12.371; p=0.001), disseminated cancer (OR 7.915, 95% CI 2.514-24.919; p<0.001), hematologic disorders (OR 1.923, 95% CI 1.173-3.154; p=0.010), functional dependence (OR 2.244, 95% CI 1.377-3.656; p=0.001), and ASA ≥ 3 (OR 1.835, 95% CI 1.394-2.417; p<0.001) were shown to be significant independent risk factors for blood transfusions. (Table 2) The Hosmer-Lemeshow Test had a significance of 0.257, while the C-statistic or Area Under ROC (AUROC) was 0.728 (p<0.001), indicating relatively strong predictability and discriminatory ability of the logistic regression model in predicting blood transfusions based on the entered variables. (Figure 1)

Table 2.

SIGNIFICANT RISK FACTORS for Intra-op/Post-op Blood Transfusions

RISK FACTORS	OR	95% CI		P-value
Age
x ≤ 55	Reference	-	-	-
55 < x ≤ 65	0.899	0.499	1.619	0.723
65 < x ≤ 75	1.107	0.637	1.923	0.718
75 < x ≤ 85	1.674	0.958	2.927	0.071
85 < x	3.192	1.604	6.350	0.001
Gender
Male	Reference	-	-	-
Female	2.258	1.738	2.935	<0.001
Obesity (≥ 30 kg/m2)	0.585	0.457	0.750	<0.001
Diabetes Mellitus
No-DM	Reference	-	-	-
NIDDM	1.344	0.964	1.876	0.082
IDDM	2.045	1.331	3.141	0.001
Dyspnea
No Dyspnea	Reference	-	-	-
Moderate Exertion	0.971	0.637	1.48	0.891
At Rest	1.766	0.516	6.044	0.365
COPD	1.362	0.922	2.012	0.120
Ascites	>999.99	0.000	-	0.999
CHF	1.327	0.446	3.944	0.611
Hypertension	1.051	0.797	1.385	0.727
Acute Renal Failure	9.178	1.630	51.684	0.012
Dialysis	4.816	1.874	12.371	0.001
Disseminated Cancer	7.915	2.514	24.919	<0.001
Hematologic Disorders	1.923	1.173	3.154	0.010
Transfusion (within 72 hours of start of surgery)	3.538	0.856	14.626	0.081
Functional Status
Independent	Reference	-	-	-
Partially or Totally Dependent	2.244	1.377	3.656	0.001
ASA Classification
ASA 1, 2	Reference	-	-	-
ASA 3, 4, 5	1.835	1.394	2.417	<0.001

OR: Odds Ratio; CI: Confidence Interval; DM: Diabetes-mellitus; NIDDM: Non-insulin dependent diabetes mellitus; IDDM: Insulindependent diabetes mellitus; COPD: Chronic obstructive pulmonary disease; CHF: Congestive Heart Failure; ASA: American Society of Anesthesiologists

Figure 1.

ROC Curve Assessing Multivariate Logistic Regression Model for Risk Factors for Transfusions Following Total Shoulder Arthroplasty

Discussion

Blood transfusions have been associated with a multitude of different complications that can have adverse effects on patients in the perioperative period.⁷ As such, taking steps to mitigate blood transfusion needs remains an important way that orthopaedic surgeons can help to minimize avoidable risks. By identifying high risk patients, perioperative management can be optimized and better informed patient counseling can be done prior to surgery. The present study identified age > 85 years, female sex, IDDM, acute renal failure, dialysis dependence, disseminated cancer, hematologic disorders, functional dependence, and ASA > 3 as independent risk factors for blood transfusion requirements following TSA for OA.

Advanced age and female gender have previously been reported as risk factors for blood transfusion requirements in various orthopaedic procedures including total knee arthroplasty. (TKA), total hip arthroplasty (THA), and spinal fusion.^24-31 Age has been analyzed as a risk factor for blood transfusion following TSA. The majority of the literature consistently demonstrates age to increase the risk of blood transfusion needs in reverse/anatomic TSA.^8,9,14,15,16 Decreased hematopoiesis and decreased physiologic reserve as bone marrow function declines with aging is likely responsible for the increased rates of transfusion that are seen following TSA and other various orthopaedic procedures.^24,25,26,32 The present study’s results regarding increased risk of transfusion with female patients have been reported previously in the literature. Hemoglobin levels are generally lower at baseline in female patients as opposed to male patients due to multiple factors including different hormonal balances and higher rates of iron deficiency anemia from menstrual loss.³³ A decreased baseline hemoglobin in female patients may lead to increased blood transfusion levels in surgical procedures, such as reverse/anatomic TSA.

Unsurprisingly, the present study identified hematologic disorders as an independent risk factor for increased transfusion in reverse/anatomic TSA in line with the current literature. Kandil et al. reports a 3.5 times increased risk of blood transfusion in deficiency anemia patients following reverse/anatomic TSA.⁸ Makhni et al. also report lower preoperative Hgb levels as an independent risk factor for blood transfusion in reverse/anatomic TSA.¹¹ Hypervigilance for patients with underlying anemia or hematologic disorders should be exercised for all operations, including TSA.

To the author’s knowledge, the effects of acute renal failure and dialysis dependence as independent risk factors for transfusion in TSA have not been well studied. Although the risk of acute renal failure/acute kidney injury as a result of blood transfusions is a subject of current debate,³⁴ the reciprocal relationship of acute renal failure increasing blood transfusion needs has not been demonstrated. Regarding dialysis dependence, Cancienne et al. previously reported over a four-fold increased incidence of dialysis-dependent patients undergoing TSA from 2005 to 2014.³⁵ As such, it is increasingly relevant and essential to further elucidate how this preoperative comorbidity affects the reverse/anatomic TSA patient population. End stage renal disease has been previously demonstrated to increase the risk of blood transfusion in lower extremity procedures such as TKA.³⁶ As such, the results of the present study also support this relationship in reverse/anatomic TSA patients for OA management. As the kidneys are responsible for erythropoietin production and bone marrow stimulation, development of anemia in dialysis dependent patients is common.³⁷ With baseline anemia, it is plausible that these patients would be at higher risk for transfusion needs in the perioperative period.

Similarly, there is limited literature regarding disseminated cancer as a risk factor for transfusion in TSA. Newman et al. previously reported significantly higher rates of postoperative transfusion following THA in patients with various hematologic malignancies.³⁸ Menendez et al. also demonstrated that multiple myeloma was associated with an increased the risk of blood transfusions following total joint arthroplasty, possibly due to impaired bone marrow function.³⁹ However, to the author’s knowledge, no previous relationship between malignancy and increased blood transfusion needs in the anatomic/reverse TSA patient populations has been reported. Although the exact mechanism as to how malignancy increases blood transfusion needs following TSA is likely multifactorial, further work elucidating the relationship between transfusion needs and different types of cancers/stages of disease is required.

Additional risk factors identified by the present study include IDDM, ASA ≥ 3, and functional dependence. Consistent with these findings, Fu et al. also report increased rates of transfusion in IDDM patients following reverse/anatomic TSA.⁴⁰ Anthony et al. and Dacombe et al. also report increased risk of blood transfusion in patients with a higher ASA grade.^17,41 To the authors’ knowledge, functional dependence and its role as a risk factor for blood transfusions has not been previously demonstrated in reverse/anatomic TSA. However, increased perioperative complications, including blood transfusions, have been reported in other orthopaedic procedures, such as total hip arthroplasty, in patients with increased functional dependence.⁴²

The present study highlights various risk factors, both modifiable and non-modifiable, for increased blood transfusion in reverse/anatomic TSA. By identifying non-modifiable risk factors, such as age and gender, orthopaedic surgeons can better counsel these high-risk patients on the risks of blood transfusion for reverse/anatomic TSA for OA treatment. A thorough discussion between provider and patient should strongly consider whether elective TSA is the best treatment modality in high-risk patients. Increased vigilance in the postoperative period for potential transfusion needs should also be exercised in patients with numerous non-modifiable risk factors. Surgeons should work with patients with modifiable risk factors together to best optimize the patient preoperatively in order to minimize the risk of transfusion. Similar to prior reports in the literature,^{14,27,43,44,45} the present study has also demonstrated increased complication rates, unplanned reoperation/readmission rates, and mortality following blood transfusions for reverse/anatomic TSA. By understanding the risk factors for blood transfusions in reverse/anatomic TSA, blood transfusions and their associated complications can be minimized.

Several limitations are present in this retrospective study. Patients with missing data in any of the variables analyzed were excluded to reduce the confounding effects of missing data; however, these patients still represent those with osteoarthritis who were indicated for TSA. Excluding these patients may potentially result in small variations in our results, though these effects seem minimal given the large cohort that was analyzed. Another limitation with the present study is the number of patients with missing pre-operative hemoglobin levels. Intuitively, lower pre-operative hemoglobin levels would logically be the greatest risk factor for increased transfusion risks. However, less than 50% of the patients who met inclusion criteria had pre-operative hemoglobin/hematocrit (Hgb/Hct) values recorded. With only a minority of patients with these pre-operative lab values, it is difficult to accurately ascertain at which specific pre-operative Hgb/Hct value greatly increases risk of transfusion. Future studies are warranted to better ascertain at what pre-operative Hgb/Hct levels greatly increases transfusion risk for TSA. In addition, another limitation specific to this large-scale database study is the lack of recorded transfusion thresholds. Lack of such data may introduce potential confounding variables that this study could not account for. The present study also utilized CPT codes that were not intended for research purposes. As a result, potential data skewing based on financial incentive cannot be controlled for.⁴⁶ Specific to this study, the ACS-NSQIP database cannot differentiate between reverse TSA and anatomic TSA as they are both identified by the same CPT codes. Although previous reports have demonstrated increased rates of transfusion in reverse TSA compared to anatomic TSA,^9,11,15 the present study has analyzed reverse and anatomic TSA as a whole similar to other reports in the literature.^10,13 Finally, another limitation inherent to all database studies is the inability to confirm the accuracy of the information inputted into the nationwide database.

As the incidence of TSA continues to increase, it is important to recognize and identify complications that can arise from this operative intervention. One way to curtail complications is to prevent precipitating events, such as blood transfusions. Previous studies have analyzed risk factors for blood transfusions in TSA for a variety of indications in aggregate. However, the present study analyzed risk factors for blood transfusion needs in reverse/anatomic TSA for the management of OA. As OA continues to be the leading indication for TSA, analyzing risk factors for transfusion in this specific patient population is warranted. The present study identified various risk factors for increased blood transfusion requirements following reverse/anatomic TSA for the management of osteoarthritis , including age > 85 years, female sex, IDDM, acute renal failure, dialysis dependence, disseminated cancer, hematologic disorders, increased functional dependence, and ASA Class ≥ 3. By identifying high-risk patients, surgeons can both better counsel and optimize management to prevent secondary complications from blood transfusions.