High healthcare utilization and associated costs in the year prior to both primary reverse and anatomic shoulder arthroplasty: a study of 2393 patients from a private insurance database

Abstract

Background

Reverse shoulder arthroplasty (rTSA) has recently become more prevalent than anatomic shoulder arthroplasty (aTSA) in primary settings. With a shift toward value-based healthcare models, it is essential to quantify the costs of nonoperative management to optimize resource utilization. The purpose of this investigation was to quantify the cost of non-operative interventions in the year prior to both aTSA and rTSA.

Methods

An observational cohort study was conducted using the IBM Watson Health MarketScan databases. Patients with shoulder arthritis who underwent unilateral, isolated primary aTSA or rTSA from January 1, 2018, to December 31, 2019, were included. The main outcome was the total costs of nonoperative procedures in the year before surgery. The nonoperative procedures examined were (1) physical therapy (PT); (2) bracing; (3) intra-articular injections: professional fee, hyaluronic acid, and corticosteroids; (4) medication: nonsteroidal anti-inflammatory drugs, opioids, and acetaminophen; and (5) shoulder-specific imaging.

Results

The study comprised 2393 patients undergoing aTSA and rTSA. The average cost of nonoperative management in the year preceding shoulder arthroplasty was $1416 ± 2271 for a total of nearly $2.6 million (USD). The total cost of nonoperative procedures was significantly higher for women compared to men ($1552 ± 2268 vs. $1323 ± 2270, P < .001). Patients undergoing rTSA incurred higher costs than those receiving aTSA ($1624 ± 2492 vs. $1092 ± 1827; P < .001), primarily due to increased spending on PT ($547 ± 1584 vs. $198 ± 1292; P < .001) and magnetic resonance imaging ($454 ± 790 vs. $242 ± 503; P < .001). For those awaiting surgery for 10 months or longer, significantly more was spent on nonoperative management ($2130.36 ± 45.6 vs. $1229.55 ± 409.18, P = .03), with PT contributing to this even in the later months.

Conclusion

There is high health care utilization and associated cost of nonoperative procedures in the year prior to rTSA and aTSA. rTSA patients had significantly higher preoperative utilization and costs, mainly due to PT and magnetic resonance imaging. Most notably, for those waiting more than 10 months for rTSA, nearly 30% of the spending occurred in the last 3 months preceding surgery. As shoulder arthroplasty volumes rise, especially with increasing rTSA, it is important to delineate the current usage. This will allow payors and surgeons to critically appraise nonoperative modalities and direct their use to optimize efficacy while providing value-based care.

Over the last few decades, shoulder replacement volume has increased exponentially.^2,11 In 2017, it was estimated that the prevalence of total shoulder arthroplasty (TSA) in the United States exceeded 823,000 patients.¹¹ Shoulder arthritis has been shown to affect close to 33% of adults over the age of 60 and can be a source of significant pain, morbidity, and disability.¹⁸ Anatomic shoulder arthroplasty (aTSA) has been shown to be highly effective in treating this pathology and is still considered to be the surgical gold standard.² Reverse total shoulder arthroplasty (rTSA) was developed as a treatment option for patients with shoulder arthritis who are rotator cuff-deficient.² Over time, the indications for rTSA have expanded beyond rotator cuff arthropathy (RCA) and now include irreparable rotator cuff tears (RCTs) and severe glenohumeral arthritis.^2,11,22 This broadening of indications has resulted in rTSA surpassing aTSA as the more frequently performed procedure in the primary setting in the United States, while contributing to the increasing popularity of shoulder arthroplasty as a whole.²

Despite being effective treatment modalities, aTSA and rTSA are typically only performed once the typical checkbox of conservative measures have been fully exhausted.¹⁸ As with other types of joint replacement surgery, the decision to proceed with shoulder arthroplasty is complex, and appropriate use criteria have been developed to guide treatment.¹⁵ Multiple stakeholders including the physician, the patient, and insurance payors are involved.²³ Appropriate surgical indications and patient preferences for nonsurgical treatment modalities must be respected. However, insurers frequently require patients to undergo nonoperative treatment before authorizing coverage for shoulder replacement surgery.¹⁸ This is driven, in part, by the upfront cost of surgical intervention.^6,23 Despite this, recent research has demonstrated that these conservative treatment modalities may impart a significant financial burden.¹⁸ Furthermore, up to 40% of the treatment modalities used were not recommended based on the current American Academy of Orthopaedic Surgeons (AAOS) clinical practice guidelines (CPGs) (ie opioid pain medication, steroid injections).¹⁸

Considering the trend toward value-based health care models, along with rTSA now representing the majority of shoulder arthroplasty procedures, it is important to accurately quantify the costs of nonsurgical management in patients prior to undergoing aTSA and rTSA. The objective of this investigation was to describe the costs associated with nonoperative procedures for patients in the 1 year period leading up to primary aTSA and rTSA. We hypothesize that, due to broader indications and lack of familiarity with rTSA from primary care physicians and general orthopedists prior to referral to shoulder specialist surgeons, the cost of nonoperative management will be higher for rTSA; the clinical significance of which will ultimately provide a more accurate depiction of the financial burden associated with these procedures in the year proceeding shoulder arthroplasty.

Materials and methods

Data source

An observational cohort study was conducted using data from the IBM Watson Health MarketScan Commercial Claims and Encounters and MarketScan Medicare Supplemental and COB databases (IBM Corp., Armonk, NY, USA) from January 1, 2018, to December 31, 2019. The dataset allows the tracking of patients across providers and hospitals. The Commercial Claims and Encounters database comprises medical and drug data from employers and health plans for >203 million individuals annually, including employees, their spouses, and dependents who are covered by employer-sponsored private health insurance in the United States¹ The Medicare Supplemental and COB database consists of the Medicare-covered portion of payment (represented as the Coordination of Benefits, or COB, amount), the employer paid portion, and out-of-pocket patient expenses.¹

Inclusion and exclusion criteria

The study population included patients with shoulder arthritis (including osteoarthritis, RCA, rheumatoid arthritis, and ‘other’ arthropathies) who underwent unilateral, isolated primary aTSA or rTSA during the period of January 1, 2018, to December 31, 2019. Current Procedural Terminology (CPT) and International Classification of Diseases Tenth Revision (ICD-10) codes were used to identify procedures and diagnoses. Shoulder arthroplasty patients were initially identified using CPT code 23472. Since there is no specific CPT code to differentiate aTSA vs. rTSA, ICD-10 procedure codes specific to each were used to differentiate the procedures (0RRJ0JZ, 0RRK0JZ; 0RRJ00Z, 0RRK00Z). The ICD-10 diagnosis codes used are listed in Supplementary Appendix S1.

Patients with other potential confounding diagnoses, such as metastatic cancers, lymphoma, leukemia, fibromyalgia, regional pain (except in the shoulder), and pathologies of joints other than the shoulder that may have resulted in overlapping treatments were excluded from the analysis (see Supplementary Appendix S2). Patients undergoing arthroplasty for fracture or concurrent shoulder fracture fixation and shoulder hemiarthroplasty aTSA with concurrent diagnosis of RCT were also excluded (see Supplementary Appendix S2).

Primary outcome

The primary outcome measure was the cost of identified nonoperative procedures in the 1-year period before primary aTSA or rTSA. The total cost was based on the gross covered payment (PAY) in the database, which includes deductibles, coinsurance, and reimbursements made by the insurer. All dollar values were inflation-adjusted to July 1, 2020, U.S. dollars.

On the basis of treatment guidelines reviewed by the Arthritis Foundation in collaboration with the American College of Rheumatology and the AAOS CPG, the nonoperative procedures for shoulder arthritis and irreparable RCT included in this study were physical therapy (PT), intra-articular injections: professional fee, hyaluronic acid (IA-HA), and corticosteroids (IA-CS); medication: nonsteroidal anti-inflammatory drugs, opioids, and acetaminophen; and shoulder-specific imaging.²⁵ PT, intra-articular injections, and imaging procedures were identified using CPT codes associated with each encounter (see Supplementary Appendix S2). Because of the structure of the database, professional fees for injections were analyzed separately, because there were instances in which same-day reimbursements made for injectable drugs and professional fees could not be correspondingly matched (eg, 2 injectable drug codes with 1 professional fee on the same day). Furthermore, the CPT code for professional fees includes procedures such as aspiration, which may be an independent procedure. The drug therapeutic class variable was used to identify types of medication.

Other outcomes

Nonoperative cost per patient across patient sex (male or female), region (Northeast, Midwest, South, or West), and time from diagnosis to aTSA or rTSA (in months, within the 1-year follow-up period) were recorded.

Statistical analysis

Descriptive analyses were performed to compare the aggregate costs across all nonoperative procedures that took place within 1 year before primary aTSA or rTSA. Differences between categorical exposure and continuous outcome measures were compared using 2-tailed t tests and a 1-way analysis of variance. Bonferroni corrected post hoc tests were performed as needed.

Results

A total of 2393 patients (38.9% aTSA, 61.1% rTSA; 40.8% female) were included in the analysis (Fig. 1). The mean age and standard deviation of the cohort was 63.5 ± 9.1 years (60.4. ± 8.1 years for aTSA and 65.6 ± 9.1 years for rTSA, P < .001). Detailed patient demographics are presented in Table I.

Figure 1.

Flowchart of inclusion and exclusion criteria. A total of 2393 patients (38.9% aTSA, 61.1% rTSA; 40.8% female) were included in the analysis. aTSA, anatomic shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty; TSA, total shoulder arthroplasty; OA, osteoarthritis.

Table I.

Characteristics of patients undergoing anatomic or rTSA.

	Total	Anatomic	Reverse
N	1828	712 (38.9%)	1116 (61.1%)
Age	63.5 ± 9.1	60.4 ± 8.1	65.5 ± 9.1
Gender
Female	745 (40.8%)	236 (33.1%)	509 (45.6%)
Male	1083 (59.2%)	476 (66.9%)	607 (54.4%)
Region
Northeast	253 (13.9%)	122 (17.2%)	131 (11.7%)
Midwest	627 (34.3%)	211 (29.7%)	416 (37.3%)
South	675 (37.0%)	269 (37.9%)	406 (36.4%)
West	271 (14.8%)	108 (15.2%)	163 (14.6%)
Time to surgery [d]	168.7 ± 109.1	170.1 ± 106.6	167.7 ± 110.9

rTSA, total shoulder arthroplasty.

The average cost of nonoperative management per patient in the year preceding shoulder arthroplasty was $1416 ± 2271 (1092 ± 1827 for aTSA and 1624 ± 2492, P < .001) for a total of nearly $2.6 million in this national sample. PT and magnetic resonance imaging (MRI) were the 2 largest drivers of cost (29% and 26.2%, respectively). (Table II). When comparing expenditures between aTSA and rTSA, the rTSA cohort spent significantly more on PT per patient ($547 ± 1584 for rTSA vs. $198 ± 1292 for aTSA; P < .001) and MRI ($454 ± 790 for rTSA vs. $242 ± 503 for aTSA, P < .001), while aTSA was associated with increased spending on computed tomography (CT) scans per patient ($248 ± 425 for aTSA vs. $176 ± 392 for rTSA, P < .001). There were no significant differences in spending for preoperative injections (IA-HA and IA-CS) or medications (nonsteroidal anti-inflammatory drugs, opioids, and acetaminophen) per patient between the aTSA and rTSA cohorts (Table III).

Table II.

Breakdown of costs of nonoperative procedures and corresponding American Academy of Orthopaedic surgeons clinical practice guideline (AAOS CPG) recommendations.

Procedure	Number of patients	Percentage of total patients [%]	Total expenditure within cohort [$]	Percentage of total expenditure [%]	Mean cost ± SD per patient ($)	Median cost (IQR) per patient ($)	AAOS CPG recommendation	Strength of recommendation
All	1828	100.0	2,589,269	−	1416 ± 2271	782 (318-1583)	-	-
PT	507	27.7	751,386	29.0	1482 ± 2527	685 (235-1720)	”In the absence of reliable evidence, it is the opinion of the work group that physical therapy may benefit select patients with glenohumeral arthritis”	Consensus (1/4 stars)
Injection
Professional fee	1002	54.8	279,302	10.8	279 ± 668	150 (87-288)	-	-
IA-HA	17	0.9	15,590	0.6	917 ± 694	597 (543-971)	“…there is no benefit to the use of HA in the treatment of glenohumeral joint arthritis”	Strong (4/4 stars)
IA-CS	915	50.1	23,498	0.9	26 ± 37	15 (7-29)	-	-
Medication
NSAIDs	736	40.3	147,915	5.7	201 ± 1145	24 (6-96)	-	-
Opioids	699	38.2	88,063	3.4	126 ± 987	10 (5-23)	“In the absence of reliable evidence, it is the opinion of the work group that opioids not be prescribed as routine and long-term pain management of glenohumeral arthritis”	Consensus (1/4 stars)
Acetaminophen	287	15.7	14,972	0.6	52 ± 312	4 (2-13)	-	-
Imaging
X-ray	1624	88.8	217,024	8.4	134 ± 270	68 (39-134)	“In the absence of reliable evidence, it is the opinion of the work group that patients with glenohumeral arthritis undergoing arthroplasty should be imaged with axillary and true AP (Grashey view) radiographs, with advanced imaging performed at the discretion of the clinician”	Consensus (1/4 stars)
CT	738	40.4	328,815	14.4	505 ± 508	321 (170-599)	-	-
MRI	797	43.6	678,718	26.2	852 ± 845	548 (307-1049)	-	-

SD, standard deviation; IQR, interquartile range; HA, hyaluronic acid; CT, computed tomography; MRI, magnetic resonance imaging; IA-HA, intra-articular injections: hyaluronic acid; IA-CS, intra-articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; PT, physical therapy; AP, anterior-posterior.

Table III.

Breakdown of per patient costs for each procedure by TSA type.

Procedure	Mean cost ± SD per patient∗ ($)
	Anatomic	Reverse	P value
All	1092 ± 1827	1624 ± 2492	<.001
PT	198 ± 1292	547 ± 1584	<.001
Injection
Professional fee	140 ± 293	161 ± 615	.391
IA-HA	7 ± 89	10 ± 120	.534
IA-CS	13 ± 30	13 ± 29	.610
Medication
NSAIDs	74 ± 571	85 ± 820	.758
Opioids	50 ± 710	47 ± 543	.935
Acetaminophen	13 ± 189	5 ± 52	.222
Imaging
X-ray	107 ± 189	126 ± 293	.106
CT	248 ± 425	176 ± 392	<.001
MRI	242 ± 503	454 ± 790	<.001

CT, computed tomography; MRI, magnetic resonance imaging; IA-HA, intra-articular injections: hyaluronic acid; IA-CS, intra-articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; PT, physical therapy; TSA, total shoulder arthroplasty; SD, standard deviation.

^∗Denotes P value statistical significance <.05.

The cost prior to aTSA and rTSA was also stratified by diagnosis (Table IV). There was no significant difference in spending for patients undergoing either aTSA or rTSA for osteoarthritis (1055 ± 1429 vs. 967 ± 1808; P = .48) or rheumatoid arthritis (670 ± 166 vs. 424). Of note, only 1 patient undergoing rTSA was identified as having a diagnosis of rheumatoid arthritis. When looking at ‘other arthritis and arthropathies,’ significantly more was spent prior to aTSA per patient (2251 ± 6550 vs. 851 ± 1101; P = .082).

Table IV.

Breakdown of per patient cost for each diagnosis by TSA type.

Diagnosis	Mean cost ± SD per patient∗ ($)
	Total	Anatomic	Reverse	P value
Shoulder osteoarthritis (OA)	1036 ± 1518	1055 ± 1429	967 ± 1808	.48
Rotator cuff tears (RCT) or arthropathy (RCA)	1836 ± 2666	-	1837 ± 2666	-
Shoulder rheumatoid arthritis	608 ± 183	670 ± 166	424	.33
Other arthritis and arthropathies	1190 ± 3364	2251 ± 6550	851 ± 1101	.082

TSA, total shoulder arthroplasty; SD, standard deviation.

^∗Denotes P value statistical significance <.05.

The total cost of nonoperative procedures prior to shoulder arthroplasty was significantly higher for female patients compared to male patients ($1552 ± 2268 for women vs. $1323 ± 2270 for men, P < .001). When looking at each procedure independently, significantly more was spent on average for each female patient prior to rTSA ($1809 ± 115 for women vs. $1468.25 ± 97.14 for men) compared to male patients; however, there were no significant sex-related differences for patients who underwent aTSA ($998 ± 73 for women vs. $1137 ± 96 for men). Within the rTSA cohort, PT was a significant contributor to the sex-related differences in expenditures (717 ± 1847 for women vs. 403 ± 1309) (Fig. 2).

Figure 2.

Bar chart showing total cost of nonoperative procedures 1-year prior to aTSA and rTSA for female and male patients. aTSA, anatomic shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty; IA-AH, intra-articular injections: hyaluronic acid; IA-CS, intra articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; PT, physical therapy; CT, computed tomography; MRI, magnetic resonance imaging.

In terms of the overall spending patterns between the 4 geographic regions of the United States (Northeast, Midwest, South, and West), no significant differences were found when looking at both types of shoulder arthroplasty combined (P = .533). Furthermore, no regional differences were found when looking at aTSA (P = 1) and rTSA (P = .16). When looking specifically at opioid prescribing, a non-significant trend (P = .23) toward increased spending was observed in the Midwest and West ($86.34/patient and $49.19/patient, respectively) compared to the South and Northeast ($28.17/patient and $6.24/patient, respectively). However, the average number of opioid prescriptions per patient in the West (1.6 ± 3.6) was found to be significantly higher than that in the South (1.0 ± 2.3) and Northeast (0.7 ± 1.5) (P = .002) (Fig. 3).

Figure 3.

Bar chart showing total cost of nonoperative procedures 1-year prior to aTSA and rTSA by region (Northeast, Midwest, South, and West). aTSA, anatomic shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty; IA-AH, intra-articular injections: hyaluronic acid; IA-CS, intra articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; CT, computed tomography; MRI, magnetic resonance imaging.

The cost of nonoperative procedures per patient according to the duration (in months, rounded to the nearest integer) between diagnosis and surgery is presented in Figure 4. The average duration between diagnosis and surgery was 5.6 months ± 3.6 days. The total cost of nonoperative procedures per patient generally increased with an increasing number of months between diagnosis and surgery, with the average spending per patient on those waiting for 10 months or more being significantly higher ($2130.36 ± 45.6 for >10 months vs. $1229.55 ± 409.18 <10 months, P = .03). Again, PT was a significant driver of this cost difference for both aTSA ($147 ± 548 for <10 months vs. $383 ± 2557 for >10 months; P = .043) and rTSA ($382 ± 1188 for <10 months vs. $1041 ± 2343 for >10 months; P < .001) (Table V).

Figure 4.

Bar chart showing total cost of nonoperative procedures 1-year prior to aTSA and rTSA by duration between diagnosis and surgery. The 12-month group includes patients with a duration of ≥12 months. aTSA, anatomic shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty; IA-AH, intra-articular injections: hyaluronic acid; IA-CS, intra articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; PT, physical therapy.

Table V.

Breakdown of per patient costs for each procedure by TSA type for patients waiting for 0-9 months vs. 10-12 months.

	Anatomic			Reverse
	0-9 mo (n = 556)	10-12 mo (n = 156)	P value	0-9 mo (n = 837)	10-12 mo (n = 279)	P value
PT	$147 ± 548	$383 ± 2557	.043	$382 ± 1188	$1041 ± 2343	<.001
Injection
Professional fee	$106 ± 185	$261 ± 502	<.001	$142 ± 686	$217 ± 312	.081
IA-HA	$3 ± 45	$20 ± 171	.030	$48 ± 97	$15 ± 172	.40
IA-CS	$411 ± 24	$22 ± 45	<.001	$10 ± 23	$20 ± 40	<.001
Medication
NSAIDs	$65 ± 486	$109 ± 806	.39	$69 ± 848	$132 ± 731	.27
Opioids	$61 ± 803	$9 ± 30	.42	$27 ± 215	$108 ± 1018	.031
Acetaminophen	$15 ± 214	$3 ± 21	.46	$5 ± 53	$7 ± 49	.63
Imaging
X-ray	$107 ± 406	$104 ± 127	.83	$121 ± 307	$143 ± 247	.27
CT	$242 ± 406	$272 ± 487	.44	$172 ± 395	$186 ± 383	.61
MRI	$241 ± 490	$247 ± 547	.90	434 ± 708	$512 ± 994	.16

TSA, total shoulder arthroplasty; CT, computed tomography; MRI, magnetic resonance imaging; IA-HA, intra-articular injections: hyaluronic acid; IA-CS, intra-articular injections: corticosteroids; NSAIDs, nonsteroidal anti-inflammatory drug; PT, physical therapy.

When looking at all patients, those who waited for 10 months or more to undergo rTSA had 24.9% of their spending occur during the last 3 months before surgery, with 10% being attributed to PT in the last 3 months (Table VI). Stratifying by diagnosis, those with rheumatoid or ‘other’ arthropathies spent the most on PT during this period (47.4%), followed by those with RCT or RCA (9.1%) and OA (1.7%).

Table VI.

Cost of total nonoperative treatments and physical therapy (PT) for those undergoing reverse total shoulder arthroplasty (rTSA) 10 months or more after diagnosis.

	rTSA >10 mo (n = 279)
	Late costs (3-mo period before surgery)	Percentage of total cost∗ [%]
All diagnoses
Total	$593 ± 1104	24.9
PT	$112 ± 569	10.8
Osteoarthritis (OA)
Total	$449 ± 968	33.3
PT	$5 ± 34	1.7
Rotator cuff tears (RCT)/arthropathy (RCA)
Total	$591 ± 911	21.9
PT	$116 ± 429	9.1
Rheumatoid and other arthritis or arthropathies
Total	$969 ± 2363	48.3
PT	$351 ± 1558	47.4

^∗Total cost of treatment from diagnosis to surgery.

Discussion

In the year preceding aTSA and rTSA, there is high healthcare utilization and associated cost of nonoperative procedures. Patients undergoing rTSA had significantly higher preoperative utilization and costs versus aTSA, which can be attributed to PT and MRI. Notably, for those awaiting rTSA for longer than 10 months, nearly one-quarter of the spending occurred in the 3 months preceding surgery, with a large percentage still being directed to PT. Interestingly, this pattern was not limited to only those with the diagnosis of a previous RCT.

Extrapolating the results of this study to the total volume of shoulder replacements being performed annually in the United States (approximately 100,000/year) reveals that the total cost of nonoperative treatment approaches half a billion dollars over a 3-year period.² Over the last decade, the volume of shoulder arthroplasty cases has increased exponentially, with a significant trend toward more rTSA.^2,13 Presently in the United States, up to 70% of all shoulder arthroplasty cases are rTSA, a procedure that has been shown to be associated with higher upfront cost.^2,6 Thus, as we move toward value-based models for health care delivery, accurate quantification of the costs preceding these increasingly popular procedures is critical. Orthopedic providers and insurance payors are urged to consider prudent use of such diagnostic and interventional options, particularly in patients likely indicated for surgical intervention for severe pathology in a shorter time reference. This may allow a reduction in the costs associated with the overall diagnostic episode of care.

Our investigation is unique compared to prior literature, as our study analyzed rTSA in addition to aTSA. A prior cost analysis of the nonoperative interventions performed by Malik et al only investigated patients undergoing aTSA. Further, their investigation relied primarily on data from patients on Medicare versus our study that utilized a large proportion of a private insurance data set, which may be more generalizable nationwide. In the prior Malik et al study, the mean amount spent was over $800 dollars per patient.¹⁹ In our investigation, looking at both aTSA and rTSA, we found an even higher utilization of resources. Based on data from private insurers from the MarketScan database, the mean cost of nonoperative procedures per patient in the year preceding aTSA and rTSA was $1092 ± 1827 and $1624 ± 2492 (P < .001), respectively. Most notably, significantly more was spent on nonoperative interventions for those waiting for 10 months or more for either aTSA or rTSA. When examining the cost breakdown, PT was the major contributor to this increase in expenditure over time. Considering the fact many patients with shoulder arthritis are initially treated with a course of conservative treatment involving 6-12 weeks of PT, which is reinforced by the fact that most insurance providers mandate documentation of this, these findings do not come as a surprise.¹⁹ To that matter, there are certain populations for which initial nonoperative management may be suited, in particular, younger patients.²⁷ In these populations, PT may provide a reduction in symptomatology and help delay operative interventions, although it has been well-studied that shoulder arthroplasty is a cost-effective intervention, even in younger populations.^4,9 However, the results of our study do shed light on the fact that a large amount of resources is being directed to an intervention for which there is limited supporting evidence in the treatment of shoulder arthritis.²⁵ Due to this paucity of evidence, The AAOS CPG for the management of glenohumeral arthritis only provides a 1 out 4-star recommendation for the use of preoperative PT.²⁵ In order to make better use of these resources, more research needs to be done to identify patient factors associated with improved outcomes following preoperative PT.¹⁹ One possible target might be patients under the age of 50. A recent study by Mandalia et al demonstrated that the rates of aTSA and rTSA are increasing in this younger age bracket.²⁰ However, they also found that rates of revision were also higher for these patients.²⁰ Thus, PT might be an effective way of delaying shoulder arthroplasty in this higher risk patient population.

When comparing the expenditures prior to aTSA versus rTSA, the significantly higher cost of nonoperative treatments in the year preceding rTSA was primarily attributed to the financial burden associated with increased use of PT and MRI. Although the indications are growing to include non-RCA, the majority of rTSAs included in this study (76.4%) were still associated with rotator cuff pathology. Thus, as PT continues to be the cornerstone of the nonoperative management of RCTs, it was expected that more would be spent on this in the rTSA cohort.⁸ However, for those waiting for 10 months or longer for surgery, total spending on PT was not only significantly higher (P < .001) but also accounted for nearly 10% of the spending in the 3 months preceding surgery (Table VI). Interestingly, this spending pattern was not limited to patients with rotator cuff pathology. PT also accounted for a large percentage of the spending on nonoperative treatment for those with rheumatoid and ‘other’ forms of arthritis during the same time period (Table VI). These findings highlight the fact that, for a wide variety of arthritides, there continues to be a significant amount of spending on PT well after 12 weeks following diagnosis. Coupled with the fact that the presence of arthritis has been shown to be a negative prognostic factor for successful outcomes post-rTSA, the cost-to-benefit ratio of prolonged PT for patients with concomitant end-stage shoulder arthritis needs to be critically appraised.^8,26

With regards to the cost of imaging, approximately $150 more was spent per patient undergoing rTSA. As hypothesized, this difference was driven by the greater costs associated with MRI in the rTSA cohort (P < .001), which is due to the fact that diagnosis of an RCT would likely influence the decision to proceed with rTSA vs. aTSA. However, it was also found that significantly less was spent on CT scans for these patients compared to those undergoing aTSA (P < .02). The use of advanced imaging prior to rTSA has been recently evaluated by Cancienne et al.⁵ In their investigation, it was found that the use of preoperative MRIs was associated with decreased rates of periprosthetic fracture, revision rTSA, and periprosthetic dislocations.⁵ However, significantly lower complications were also noted for patients receiving preoperative CT scans. Furthermore, over the 8-year study period, it was found that the rates of CT scan use increased significantly, while the preoperative MRIs trended in a downward direction.⁵

Since impingement and implant stability are reliant on glenosphere placement, the authors suggest that the observed decrease in the complication rate may be partially attributed to more accurate preoperative planning allowed for by these advanced imaging modalities. Although MRIs are needed for certain patients in order to indicate them for the appropriate type of shoulder arthroplasty, in those who can be indicated for shoulder arthroplasty without MRI, CT scans may be a cost-effective alternative. However, the increased use of CT should be approached with caution. As 3D planning software becomes increasingly popular, not only are the volume of scans increasing, but so too is scan resolution.^7,16 These higher resolution scans expose patients to increased doses of harmful ionizing radiation, while also being associated with more incidental findings.^7,16 Interestingly, despite resulting in more incidental findings, it has been shown the number of potential pathological findings is not different when comparing higher and lower resolution protocols.²⁰ Furthermore, Lorenzana et al have shown that lower resolution scans afford sufficient detail for accurate preoperative planning.¹⁶ Thus, when opting for CT scans for preoperative planning, low-dose protocols should be considered if available.

Preoperative opioid use is now well-established as a strong risk factor for prolonged postoperative use and dependence following shoulder arthroplasty.^2,10,14,24 Despite increasing awareness and efforts to curb opioid prescribing, close to 40% of the patients in this investigation received prescriptions for narcotics in the year preceding surgery. Furthermore, for patients waiting for 10 or more months for rTSA, significantly more was spent on opioid prescriptions (P = .031) when compared to those waiting for a shorter period of time, a trend that was not observed for aTSA. These patients also spent more on IA-CS (P < .001). These findings may allude to rTSA patients experiencing more pain, as the duration between diagnosis and surgery increases, which may highlight the need to indicate these patients for surgery earlier on. Increased use of steroid injections is also a concern due to the fact that intraarticular steroid has been shown to increase the risk of periprosthetic infection.¹⁷ Furthermore, intra-articular corticosteroid injections have been shown to interfere with collagen synthesis, leading to tendon attenuation and decreased tensile strength, potentially increasing the risk of tendon tear progression.^12,21 Given that irreparable RCTs and cuff tear arthropathy are 2 common indications for rTSA, it is understandable that orthopedic surgeons may be less likely to offer CS injections early in disease treatment. This may explain our finding that intra-articular CS expenditure was greater in the aTSA cohort versus the rTSA patients, specifically in the first 9 months. Regarding other injectables, HA was used at a relatively low frequency per patient, as discerned by the cost of procedure and patient cost. This may be best explained by recent literature, particularly the AAOS CPG, which states “there is no benefit to the use of HA in the treatment of glenohumeral joint arthritis” as a strong recommendation (Table II).²⁵

The geographical variations in opioid prescribing found in this study were also notable. Patients in the Western US received significantly more prescriptions for opioids compared to those living in the Northeast and the South (P = .002). A previous MarketScan database study conducted by Best et al identified preoperative opioid prescribing and living in the West as independent risk factors for long-term opioid use following shoulder arthroplasty.³ Although outside the scope of this study, our findings may provide insight into this relationship and serve as target for future education and research directed at narcotic stewardship within the orthopedic community.

Limitations

The main limitation of this investigation stems from the fact that, presently, aTSA and rTSA do not have separate CPT codes. However, the 2 procedures do have separate ICD-10 procedure codes. Thus, in order to capture the largest number of patients in the most accurate way possible, CPT codes were used to identify all shoulder arthroplasty cases in the MarketScan database. ICD procedure and diagnosis codes were then used to determine which cases met the inclusion criteria for aTSA and rTSA. Another limitation of using MarketScan data, as with other databases, is that it relies on accurate coding of procedures and diagnoses. In an effort to mitigate coding errors, strict inclusion and exclusion criteria were employed. Specifically, TSA cases with an unspecified diagnosis code or aTSA with a concomitant rotator cuff diagnosis were excluded to avoid incorrectly coded rTSA. CPT codes were also used to exclude patients undergoing arthroplasty for fracture and hemiarthroplasty cases, while ICD-10 diagnosis codes were used to eliminate cases with potentially confounding indications for arthroplasty. It should also be noted that the MarketScan database is not inclusive of all Medicare beneficiaries. Although this excludes a subset of elderly patients, other studies looking specifically at the costs for Medicare patients in the year preceding aTSA demonstrate similar patterns of resource allocation.¹⁸ The strength of this current study is that it presents data on both aTSA and rTSA; however, further studies focusing on Medicare patients receiving rTSA are warranted. Another limitation is the fact that only 1 patient undergoing rTSA was coded as having rheumatoid arthritis as their diagnosis. Using data from 1 case limits accurate comparison, while also highlighting the limitations associated with using ICD-10 diagnosis codes, as it is highly likely that more than 1 patient undergoing rTSA has rheumatoid arthritis. Finally, the data does not include treatments that are not covered by insurance, such as platelet-rich plasma and other biologics, which can be associated with considerable cost. Ultimately, these limitations suggest that the estimates for health care utilization in the year preceding shoulder arthroplasty presented in this study are conservative and that the actual cost of nonoperative treatment is likely considerably higher.

Conclusion

There is high health care utilization and associated cost of nonoperative procedures in the year prior to rTSA and aTSA. Patients undergoing rTSA had significantly higher preoperative utilization and costs, which can be attributed mainly to PT and MRI. Most notably, for those waiting for more than 10 months for rTSA, nearly 30% of the spending occurred in the last 3 months preceding surgery. As shoulder arthroplasty volumes increase, with the ever-increasing trend toward more rTSA, it is important to delineate the current usage. Understanding these costs will allow payors and surgeons to critically appraise these nonoperative modalities and direct their use in a manner that optimizes efficacy while providing value-based care.

Disclaimers

Funding: The authors did not receive any funding or support from any organization for this study.

Conflicts of interest: Dr. Shah is a paid consultant for Exactech, Inc., which is related to the subject of this work. Dr. Smith is a paid consultant for DePuy and a paid presenter for Pacira, which is related to the subject of this work. The other authors, their immediate families, and any research foundation with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.