What Is the Prevalence of Clinically Important Findings Among Incidentally Found Osseous Lesions?

Abstract

Background

Patients with incidentally found musculoskeletal lesions are regularly referred to orthopaedic oncology. Most orthopaedic oncologists understand that many incidental findings are nonaggressive and can be managed nonoperatively. However, the prevalence of clinically important lesions (defined as those indicated for biopsy or treatment, and those found to be malignant) remains unknown. Missing clinically important lesions can result in harm to patients, but needless surveillance may exacerbate patient anxiety about their diagnosis and accrue low-value costs to the payor.

Questions/purposes

(1) What percentage of patients with incidentally discovered osseous lesions referred to orthopaedic oncology had lesions that were clinically important, defined as those receiving biopsy or treatment or those found to be malignant? (2) Using standardized Medicare reimbursements as a surrogate for payor expense, what is the value of reimbursements accruing to the hospital system for the imaging of incidentally found osseous lesions performed during the initial workup period and during the surveillance period, if indicated?

Methods

This was a retrospective study of patients referred to orthopaedic oncology for incidentally found osseous lesions at two large academic hospital systems. Medical records were queried for the word “incidental,” and matches were confirmed by manual review. Patients evaluated at Indiana University Health between January 1, 2014, and December 31, 2020, and those evaluated at University Hospitals between January 1, 2017, and December 31, 2020, were included. All patients were evaluated and treated by the two senior authors of this study and no others were included. Our search identified 625 patients. Sixteen percent (97 of 625) of patients were excluded because their lesions were not incidentally found, and 12% (78 of 625) were excluded because the incidental findings were not bone lesions. Another 4% (24 of 625) were excluded because they had received workup or treatment by an outside orthopaedic oncologist, and 2% (10 of 625) were excluded for missing information. A total of 416 patients were available for preliminary analysis. Among these patients, 33% (136 of 416) were indicated for surveillance. The primary indication for surveillance included lesions with a benign appearance on imaging and low clinical suspicion of malignancy or fracture. A total of 33% (45 of 136) of these patients had less than 12 months of follow-up and were excluded from further analysis. No minimum follow-up criteria were applied to patients not indicated for surveillance because this would artificially inflate our estimated rate of clinically important findings. A total of 371 patients were included in the final study group. Notes from all clinical encounters with orthopaedic and nonorthopaedic providers were screened for our endpoints (biopsy, treatment, or malignancy). Indications for biopsy included lesions with aggressive features, lesions with nonspecific imaging characteristics and a clinical picture concerning for malignancy, and lesion changes seen on imaging during the surveillance period. Indications for treatment included lesions with increased risk of fracture or deformity, certain malignancies, and pathologic fracture. Diagnoses were determined using biopsy results if available or the documented opinion of the consulting orthopaedic oncologist. Imaging reimbursements were obtained from the Medicare Physician Fee Schedule for 2022. Because imaging charges vary across institutions and reimbursements vary across payors, this method was chosen to enhance the comparability of our findings across multiple health systems and studies.

Results

Seven percent (26 of 371) of incidental findings were determined to be clinically important, as previously defined. Five percent (20 of 371) of lesions underwent tissue biopsy, and 2% (eight of 371) received surgical intervention. Fewer than 2% (six of 371) of lesions were malignant. Serial imaging changed the treatment of 1% (two of 136) of the patients, corresponding to a rate of one in 47 person-years. Median reimbursements to work up the incidental findings analyzed was USD 219 (interquartile range USD 0 to 404), with a range of USD 0 to 890. Among patients indicated for surveillance, the median annual reimbursement was USD 78 (IQR USD 0 to 389), with a range of USD 0 to 2706.

Conclusion

The prevalence of clinically important findings among patients referred to orthopaedic oncology for incidentally found osseous lesions is modest. The likelihood of surveillance resulting in a change of management was low, but the median reimbursements associated with following these lesions was also low. We conclude that after appropriate risk stratification by orthopaedic oncology, incidental lesions are rarely clinically important, and judicious follow-up with serial imaging can be performed without incurring high costs.

Level of Evidence

Level III, therapeutic study.

Introduction

The prevalence of incidentally found musculoskeletal lesions discovered on diagnostic imaging has been reported to be between 2% and 20% [5, 9, 10, 17, 19, 21, 24]. Many patients with such findings are referred to orthopaedic oncology for evaluation [16]. Most orthopaedic oncologists understand that most incidental findings are not aggressive and either call for no specific treatment or can be observed with serial imaging. Patients, on the other hand, often experience unnecessary distress about their diagnosis [6, 7, 12, 20]. They may also incur increased financial costs related to diagnostic tests and physician visits, lost days at work or school, and potential harm in the form of unnecessary tests and radiation [8]. These negative experiences may be exacerbated or prolonged among patients indicated for surveillance with serial imaging, which is a common recommendation for patients with benign-appearing lesions and a low risk of having a malignancy.

Despite the considerable measurable and unmeasurable costs associated with incidental findings, the frequency of clinically important lesions (defined as those receiving biopsy or treatment and those found to be malignant) among all incidentally discovered osseous findings remains unknown. We believe that quantifying the prevalence of clinically important lesions may reduce unnecessary stress and anxiety among patients. Furthermore, characterizing the financial burden associated with the evaluation and management of incidentally found osseous lesions, which is currently poorly understood, could enhance the delivery of value-based care to this patient population.

We therefore asked: (1) What percentage of patients with incidentally discovered osseous lesions referred to orthopaedic oncology had lesions that were clinically important, defined as those receiving biopsy or treatment or those found to be malignant? (2) Using standardized Medicare reimbursements as a surrogate for payor expense, what is the value of reimbursements accruing to the hospital system for the imaging of incidentally found osseous lesions performed during the initial workup period and during the surveillance period, if indicated?

Patients and Methods

Study Design and Setting

This was a retrospective study of patients referred to orthopaedic oncology for incidentally found osseous lesions at University Hospitals in Cleveland, Ohio, and Indiana University Health in Indianapolis, Indiana, which are large, urban, academic hospital systems. Both institutions are associated with major cancer centers that are referral centers in their respective states. The patients included in this study were the patients of the two senior authors of this paper (PJG and LDW), both of whom are board-certified orthopaedic surgeons with fellowship training in orthopaedic oncology and more than 20 years of clinical experience. No patients evaluated by other physicians for incidentally found lesions were included. Patients evaluated at Indiana University Health between January 1, 2014, and December 31, 2020, and those evaluated at University Hospitals between January 1, 2017, and December 31, 2020, were included. We used different time periods because records before 2017 were unobtainable at University Hospitals.

Patients

We queried the medical records of our respective institutions for the word “incidental.” It is the practice of our senior authors to write “incidentally found lesion” in the chief complaint of every note associated with patients evaluated for an incidentally found lesion. This documentation occurs only after the lesion is determined to be truly “incidental” by the attending physician. Both senior authors are meticulous with record keeping, and we believe this search strategy was unlikely to have missed a substantial number of patients. After identifying positive matches, we confirmed all incidental findings by a manual review of the medical record. Lesions were defined as “incidentally found” if they were discovered unintentionally on diagnostic imaging and the lesion was determined by an orthopaedic oncologist to be unrelated to the patient’s presenting complaint. Patient follow-up was established by the date of the most recent clinical encounter with orthopaedic oncology and the date of the most recent encounter with any physician in the hospital system. Patient status from all clinical encounters with orthopaedic and nonorthopaedic providers was screened manually. Follow-up with orthopaedic oncology and follow-up with any provider in the hospital system (“hospital follow-up”) are reported.

Variables

To contextualize the incidental findings analyzed, we extracted the following information from the medical record: patient age, gender, the patient’s presenting chief complaint, clinical setting of the incidental finding (outpatient, inpatient, or emergency), imaging modality responsible for the incidental finding, anatomic location of the incidental finding, final diagnosis, and the patient’s cancer history. The final diagnoses were determined using biopsy results, if available, or the documented opinion of the consulting orthopaedic oncologist.

Description of Study Population

Our search identified 625 patients to be reviewed for potential inclusion in the study. A total of 16% (97 of 625) of patients were excluded because their lesions were not incidentally found. Another 12% (78 of 625) were excluded because the incidental findings were soft tissue lesions and not lesions of bone. Four percent (24 of 625) of patients were excluded because they had received workup or treatment for their lesions by an outside orthopaedic oncologist, and 2% (10 of 625) were excluded for missing information.

A total of 416 patients were available for preliminary analysis. Among these patients, 67% (280 of 416) were instructed to follow up as needed and 33% (136 of 416) were indicated for scheduled surveillance with imaging. The primary indication for surveillance included lesions with a benign appearance on imaging and low clinical suspicion for malignancy or fracture. A total of 33% (45 of 136) of patients indicated for surveillance had less than 12 months of follow-up with any provider at their respective institutions and were excluded from further analysis. No minimum follow-up criteria were applied to patients not indicated for surveillance because this would artificially inflate our estimated rate of clinically important findings. As a result, a total of 371 patients were included in the final study group.

Patients indicated for surveillance had a mean of 12.3 ± 10.3 months of follow-up with orthopaedic oncology and 33.0 ± 15.3 months of follow-up with any provider in their hospital system. A total of 53% (48 of 91) of patients had less than 12 months of follow-up with orthopaedic oncology, but all of these individuals continued to be followed by other providers at their respective institutions for reasons not specifically related to their incidental findings. Notes from these clinical encounters were screened for our primary endpoints. The mean hospital follow-up for the entire study group was 22.4 ± 21.9 months.

There was a bimodal age distribution with peaks at 15 years and 51 years. The median age of the study group was 47 years (interquartile range 24 to 60 years). A total of 59% (217 of 371) of patients were female, and 8% (31 of 371) had a history of cancer. Most incidental findings were discovered in the outpatient (82% [304 of 371]) and emergency (17% [64 of 371]) settings. A total of 61% (227 of 371) of lesions were identified on plain radiographs, 22% (82 of 371) on MRI, and 15% (57 of 371) on CT. The most common presenting chief complaint was ipsilateral or bilateral extremity pain (54% [201 of 371]) (Table 1), and the most common anatomic location of incidental findings was the femur (36% [134 of 371]) (Table 2).

Table 1.

Presenting chief complaint of patients

Chief complaint	Lesions that were not clinically important (n = 345)	Lesions that were clinically important (n = 26)	p value
Ipsilateral or bilateral extremity pain	56 (193)	31 (8)	0.01
Other musculoskeletal pain	9 (32)	23 (6)	0.03
Trauma	15 (52)	23 (6)	0.28
Medical symptoms/other	20 (68)	23 (6)	0.68

Data presented as % (n).

Table 2.

Anatomic location of incidental findings

Anatomic location	Lesions that were not clinically important (n = 345)	Lesions that were clinically important (n = 26)	p value
Femur	38 (131)	12 (3)	0.007
Humerus	17 (58)	19 (5)	0.75
Innominate	15 (53)	27 (7)	0.12
Tibia	14 (49)	8 (2)	0.35
Fibula	3 (12)	0 (0)	0.33
Spine	3 (11)	8 (2)	0.23
Calcaneus	3 (9)	0 (0)	0.40
Ulna	1 (5)	0 (0)	0.54
Radius	1 (3)	4 (1)	0.16
Sacrum	1 (2)	0 (0)	0.70
Multiple	1 (2)	8 (2)	< 0.001
Othera	3 (10)	15 (4)	< 0.001

Data presented as % (n).

^aClinically important lesions were in the sternum, coracoid, acromion, and metacarpal.

Potential Sources of Bias

This was a study of patients referred to and seen by orthopaedic oncology. This is a specific patient population who may have a greater prevalence of clinically important findings than the broader population of patients with incidentally found lesions. Our findings might not be generalizable to all patients with incidentally found osseous lesions, and may only be applicable to the subgroup of patients presenting for follow-up with orthopaedic oncology.

A substantial proportion of patients who were indicated for imaging surveillance had less than 12 months of follow-up with any provider in the hospital system. These patients had a mean of 3.2 ± 3.3 months of follow-up with orthopaedic oncology and 4.9 ± 3.6 months of hospital follow-up. None of these patients’ incidental findings had aggressive characteristics on imaging or a clinical picture concerning for malignancy. When serial imaging is performed, follow-up is complicated by many factors, including a lack of symptoms and patients’ perceptions that follow-up is not necessary [22]. We believe that if a patient lost to follow-up were to experience symptoms, he or she would be more likely to return for evaluation. In addition, because both of our hospital systems are associated with major cancer centers that are referral centers in their respective states, we believe it is unlikely that patients would seek care elsewhere.

Our data were collected over two time periods at two separate institutions. During the study period, all patients were evaluated and treated by the two senior authors of this study (PJG and LDW) with no specific changes in their approaches to surveillance or treatment. No national, local, or institution-specific guidelines were published during the study period that could have changed the practice patterns of these treating physicians. In addition, our patient group was evaluated and managed by only two physicians, thereby limiting variation resulting from multiple providers with different thresholds for surveillance, biopsy, and treatment. As a result, we have no reason to suspect that our methodology systematically invalidates or biases our results.

Our study period includes the first year and a half of the Coronavirus-19 pandemic, during which time outpatient medicine underwent meaningful but temporary changes. A total of 19% (79 of 416) of patients had their first appointment with orthopaedic oncology after January 1, 2020. Throughout the pandemic, however, our providers continued treating patients in person and through virtual clinic visits. We have no evidence that the referral patterns at our hospitals were systematically altered during this time. Moreover, the clinical treatment of patients with incidentally found lesions by orthopaedic oncology was unchanged during this time.

We estimated the financial burden of imaging patients with incidentally found lesions using reimbursement amounts obtained from the Medicare Physician Fee Schedule. Although the use of Medicare reimbursements is a consistent and reproducible method of measuring resource use, this approach may underestimate the total financial burden of imaging because it does not adjust for the higher payment amounts paid by commercial insurers. Nevertheless, Medicare reimbursements are a reliable proxy for quantifying average hospital reimbursements, and we do not believe this strategy invalidates our results.

Primary and Secondary Study Outcomes

Our primary outcome of interest was the percentage of patients with clinically important findings among those with incidentally found osseous lesions. For this study, clinically important findings were defined as lesions that were biopsied, treated medically or surgically, or determined to be malignant. Notes from all clinical encounters with orthopaedic and nonorthopaedic providers were screened for these endpoints. Indications for biopsy included lesions with aggressive features on imaging, lesions with nonspecific imaging characteristics and a clinical picture concerning for malignancy, and lesion changes seen on imaging during the surveillance period. Indications for treatment included lesions with an increased risk of fracture or deformity, certain malignancies, and pathologic fracture. All patients were indicated for biopsy or treatment by the two senior authors of this study (PJG and LDW) in consultation with the musculoskeletal radiologists and tumor boards at their respective institutions. The musculoskeletal radiologists at our institutions stratify the risk of aggressive behavior of osteolytic lesions by using the Lodwick or modified Lodwick-Madewell classification on musculoskeletal radiographs [2]. Generally, well-circumscribed lesions with sclerotic borders behave in a nonaggressive fashion while more poorly defined lesions without sclerotic boarders tend to behave more aggressively. Musculoskeletal radiologists may recommend repeat radiographs, MRI, or biopsy based on the appearance of a lesion using this classification in conjunction with the orthopaedic oncologist.

Our secondary objective was to estimate the value of reimbursements accruing to the hospital system as a result of incidentally found osseus lesions on imaging. The dates and types of imaging performed during the initial workup and subsequent surveillance period, if indicated, were obtained from the medical record. Imaging that was not expressly performed for these purposes, even if it captured the lesion in question, was excluded from the analysis. All imaging studies were matched to the baseline reimbursement paid by Medicare Part B for diagnostic imaging performed in 2022. These values reflect the national payment amounts obtained from the Medicare Physician Fee Schedule Look Up tool [3], which are not adjusted for regional cost variations. Each reimbursement is comprised of a technical component covering nonprovider-related expenses and a professional component covering the procedure, provider fees, and malpractice insurance. Of the amount collected from the Medicare Physician Fee Schedule, 80% is paid by Medicare and remaining 20% is expected to be collected from beneficiaries. As a result, the expense of imaging is evaluated from the perspective of the payor, including the patient and his or her insurer. We chose this methodology because reimbursements for imaging vary across institutions and payors. Use of Medicare reimbursements, on the other hand, is a consistent and reproducible method of measuring resource use because it is objective, relatively consistent across the United States, and readily available to the public. Our method, which has been used in medical research, allows for greater comparability across multiple health systems and research studies [1, 11, 15, 23, 25]. Inflation adjustments were not performed because all costs are expressed in 2022 dollars.

We calculated two financial metrics: reimbursements accrued to the hospital system for imaging performed during the initial workup of the lesion and the annual value of reimbursements for imaging performed for surveillance. The index imaging study, that is, the one that revealed the incidental finding triggering inclusion in our study, was excluded from the analysis. Annual reimbursements were calculated by dividing the total value of reimbursements for imaging used for surveillance by the length of follow-up in years.

Statistical Analysis

Patients’ presenting chief complaints, the anatomic locations of findings, and the final diagnoses of lesions were analyzed using descriptive statistics. Comparisons were performed using Pearson chi-square tests. Binary logistic regression was performed to determine whether the odds of a lesion being clinically important varied with patient age or history of cancer. A subgroup analysis was performed to evaluate whether patient age or history of cancer influenced the odds of undergoing biopsy or treatment or the odds of having a diagnosis of malignancy. The threshold for statistical significance was set at an a priori alpha level of 0.05. Results are presented as odds ratios and 95% confidence intervals.

Reimbursements for the initial workup of a lesion, as well as the annual reimbursements for imaging performed for scheduled surveillance, are summarized using descriptive statistics. Results are presented as median values and IQR. The median annual expense of surveillance was calculated on an intention-to-treat basis, and all patients indicated for scheduled surveillance were included regardless of follow-up status. All analyses were performed in SPSS Statistics 28 (IBM Corp).

Ethical Approval

Ethical approval for this study was obtained from the institutional review boards of University Hospitals (#20201557) and Indiana Health University (#2011892263).

Results

Clinically Important Findings

Seven percent (26 of 371) of incidental findings were determined to be clinically important, which was defined as lesions receiving biopsy or treatment and those found to be malignant. Five percent (20 of 371) of lesions underwent tissue biopsy, and 2% (eight of 371) received surgical intervention. One patient with a known history of nonmetastatic breast cancer underwent open biopsy and intramedullary nailing for a pathologic femur fracture after being lost to follow-up after her initial incidental finding. This patient was referred for CT-guided biopsy after initial evaluation by orthopaedic oncology, but the patient did not present for follow-up. She ultimately received a diagnosis of metastatic breast cancer. All others who received treatment for their incidentally found lesions underwent elective curettage and bone graft or intraosseous steroid injections for benign lesions of bone (Table 3).

Table 3.

Diagnoses, treatments, and indications for all treated lesions

Diagnosis	Treatment and indication
Metastatic breast cancer	84-year-old woman with nonmetastatic breast cancer and an asymptomatic lytic lesion of the proximal femur. The patient was initially referred for CT-guided biopsy, but she did not follow up. She returned several months later with a pathologic femur fracture. Patient underwent open biopsy and intramedullary nailing. Intraoperative pathology confirmed metastatic breast cancer.
Enchondroma	49-year-old woman referred from the Hand Service for an enchondroma of the proximal humerus discovered incidentally on MRI. Observation with serial imaging was recommended by orthopaedic oncology. Patient was referred to the Hand Service after PET CT demonstrated no uptake concerning for malignancy. The Hand Service was concerned about an occult rotator cuff tear that was not visible on MRI, and the patient was taken to surgery by the Hand Service for subacromial decompression with possible rotator cuff repair along with curettage and bone grafting of the bone lesion.
Enchondroma	44-year-old woman with an enchondroma of the proximal phalanx of the ring finger. The patient was taken to surgery for curettage and bone graft because of multiple areas of thin cortices with increased risk of fracture.
Schwannoma	34-year-old woman with a sacral lesion noted on workup for abdominal pain. Subsequent MRI was concerning for contrast enhancement; needle biopsy was performed. Pathology confirmed a schwannoma. Patient developed radicular pain after the biopsy, which was determined by the Spine Service to be caused by irritation from the biopsy. The lesion was excised by the Spine Service.
Nonossifying fibroma	13-year-old girl with a nonossifying fibroma of the proximal tibia. MRI revealed the lesion to be half the diameter of the tibia with associated edema. Patient underwent curettage and bone grafting.
Unicameral bone cyst	12-year-old boy with an asymptomatic unicameral bone cyst of the proximal femur. Patient underwent injection of demineralized bone matrix to consolidate the lesion because of parental preference and increased risk of fracture.
Chronic inflammation	11-year-old boy with a nonspecific lesion of the radius. Needle biopsy suggested possible eosinophilic granuloma, but neoplasm was not excluded. The tumor board recommended open biopsy and curettage, and pathology revealed chronic inflammation with no evidence of neoplasm.
Fibrous histiocytoma	9-year-old girl with a benign fibrous histiocytoma of the femoral neck. Patient underwent curettage and bone grafting based on parental preference and increased risk of fracture.

Fewer than 2% (six of 371) of lesions were malignant (Table 4). Four of these six patients had a history of cancer. One percent (two of 371) received new cancer diagnoses, 0.3% (one of 371) with a previous cancer diagnosis had a new diagnosis of metastatic disease, and 1% (three of 371) with known metastatic cancer had diagnoses of new osseus metastases. No sarcomas were identified. All malignancies were diagnosed within 6 months of the incidental finding. One of these patients was initially indicated for scheduled surveillance. This patient had a history of renal cell carcinoma and an incidentally found blastic lesion of the pubis. The lesion had no aggressive characteristics and was not consistent with metastatic renal cell carcinoma on imaging, so it was monitored closely with imaging. Six months later, the lesion demonstrated changes on CT prompting biopsy, and the patient was found to have metastatic prostate cancer. No other malignancies were identified during the surveillance period. Excluding patients who were worked up and released to follow-up as needed, imaging surveillance resulted in changes in management in 1% (two of 136), corresponding to a rate of 1 in 47 person-years.

Table 4.

Diagnoses and outcomes of incidental findings

Diagnosis	Total patients (n = 371)	Patients followed (n = 92)	Patients followed vs total patients, p value	Patients biopsied (n = 20)	Patients biopsied vs total patients, p value	Patients treated (n = 8)	Patients treated vs total patients, p value
Benign lesions
Enchondroma	23 (87)	28 (26)	0.18	0 (0)	0.01	25 (2)	0.92
Nonossifying fibroma	15 (57)	11 (10)	0.18	5 (1)	0.19	13 (1)	0.82
Fibrous dysplasia	10 (38)	2 (2)	0.004	0 (0)	0.12	0 (0)	0.33
Enostosis	6 (21)	7 (6)	0.66	0 (0)	0.26	0 (0)	0.48
Lipoma	4 (16)	5 (5)	0.52	0 (0)	0.33	0 (0)	0.54
Osteochondroma	2 (9)	2 (2)	0.87	0 (0)	0.47	0 (0)	0.65
Unicameral bone cyst	2 (8)	2 (2)	0.96	0 (0)	0.50	13 (1)	0.04
Hemangioma	2 (7)	1 (1)	0.53	15 (3)	< 0.001	0 (0)	0.69
Simple cyst	2 (7)	0 (0)	0.13	0 (0)	0.52	0 (0)	0.69
Paget disease	2 (6)	0 (0)	0.16	0 (0)	0.56	0 (0)	0.71
Bone infarct	1 (5)	0 (0)	0.20	0 (0)	0.60	0 (0)	0.74
Liposclerosing myxofibrous tumor	1 (3)	3 (3)	0.002	0 (0)	0.68	0 (0)	0.80
Other	8 (29)	4 (4)	0.16	30 (6)	< 0.001	38 (3)	0.002
Unknown	19 (72)	33 (30)	< 0.001	25 (5)	0.52	0 (0)	0.16
Total benign lesions	98 (365)	99 (91)	0.16	75 (15)	< 0.001	88 (7)	0.01
Malignant lesions
Metastatic breast cancer	1 (3)	0 (0)	0.32	10 (2)	< 0.001	13 (1)	< 0.001
Metastatic lung cancer	0 (1)	0 (0)	0.57	5 (1)	< 0.001	0 (0)	0.88
Metastatic prostate cancer	0 (1)	1 (1)	0.57	5 (1)	< 0.001	0 (0)	0.88
Plasmacytoma	0 (1)	0 (0)	0.57	5 (1)	< 0.001	0 (0)	0.88
Total malignant lesions	2 (6)	1 (1)	0.16	25 (5)	< 0.001	13 (1)	0.01

After controlling for patient age, we found that patients with a history of cancer had greater odds of their incidental findings being clinically important (OR 5.1 [95% CI 1.7 to 14.7]; p = 0.003), undergoing biopsy (OR 4.5 [95% CI 1.4 to 13.9]; p = 0.01), or being malignant (OR 10.7 [95% CI 1.8 to 65.7]; p = 0.01). Receiving treatment was not associated with a history of cancer (OR 4.0 [95% CI 0.4 to 46.1]; p = 0.26). When controlling for cancer history, patient age was associated with increased odds of malignancy (OR 1.1 [95% CI 1.0 to 1.1]; p = 0.045), but not clinical importance (OR 1.0 [95% CI 1.0 to 1.0]; p = 0.91), biopsy (OR 1.0 [95% CI 1.0 to 1.4]; p = 0.37), or treatment (OR 1.0 [95% CI 0.9 to 1.0]; p = 0.08).

Value of Imaging Reimbursements

The median value of reimbursements accrued to the hospital system for imaging performed during the initial workup of the incidental findings analyzed was USD 219 (IQR USD 0 to 404), with a range of USD 0 to 890. For patients indicated for scheduled surveillance, the median annual value of reimbursements incurred during the surveillance period was USD 78 (IQR USD 0 to 389), with a range of USD 0 to 2706 (Fig. 1). High-cost outliers were predominantly the result of serial MRIs.

Fig. 1.

This figure shows the distribution of reimbursement incurred for imaging the incidentally found osseus lesions we analyzed. Reimbursements reflect those related to the initial workup of the lesion and the annual reimbursement incurred during the surveillance period.

Discussion

Incidentally found musculoskeletal lesions are frequently encountered on diagnostic imaging and referred to orthopaedic oncology for evaluation [5, 13, 17, 19, 21, 24]. Many orthopaedic oncologists understand that most incidental findings are nonaggressive and either call for no specific treatment or can be observed with serial imaging. Yet, the prevalence of clinically important lesions among incidentally found lesions remains unknown. The results of this study confirm that the proportion of patients with clinically important findings among patients referred to orthopaedic oncology for incidentally found osseous lesions is relatively modest. Our findings further suggest that after appropriate risk stratification, low-risk, benign-appearing lesions may be judiciously followed with serial imaging without incurring an exorbitant cost to the payor. Quantifying and documenting the low frequency of clinically important findings among incidentally found osseous lesions may reduce unnecessary stress and anxiety among patients. Moreover, because guidelines for the management of incidental lesions are not well established, our findings may support the clinical practice of orthopaedic oncologists [4].

Limitations

Although we evaluated more than 400 patients seen by orthopaedic oncology at two large hospital systems, this series had only six malignancies. As a result, the study was underpowered to detect primary malignancies of bone, such as osteosarcoma, which has an incidence of 5 per million person years [18]. In addition, our findings are limited by wide CIs because of low event frequencies. Larger multicenter studies are warranted to further refine the results presented here.

A substantial portion (33% [45 of 136]) of patients indicated for imaging surveillance had less than 12 months of follow-up. These patients had a mean follow-up of 3.2 ± 3.3 months with orthopaedic oncology and 4.9 ± 3.6 months with any provider in their hospital system. The medical records from all clinical encounters with orthopaedic and nonorthopaedic providers were screened for our primary endpoints, and no evidence of new orthopaedic diagnoses, biopsies, or treatments were identified. These patients were excluded from the analysis. If additional follow-up would have resulted in any biopsies, treatments, or new cancer diagnoses, our estimated rate of clinically important lesions could have been meaningfully increased. On the other hand, if additional follow-up would have not changed management in these patients, our results would have been lower than what is reported.

Our patient population was evaluated and treated by the two senior authors of this study (PJG and LDW). Both physicians are board-certified orthopaedic surgeons with more than 20 years of clinical experience. Nevertheless, these surgeons have different thresholds for indicating patients for surveillance, biopsy, and treatment. We cannot guarantee that all patients would have been treated identically by both physicians. Moreover, patients indicated for surveillance were not necessarily followed at consistent intervals at our institutions, nor were strict surveillance protocols used at either hospital system to correlate the frequency of imaging to patient pathology. All surveillance recommendations were made by the attending orthopaedic oncologist based on their clinical experience in consultation with the musculoskeletal radiologists at their respective institutions. As a result, our findings may have been affected by variations in the treatment of our patient population.

Our estimates of the financial burden incurred by imaging incidentally found lesions were estimated using data from the Medicare physician fee schedule. This method was chosen because it is a consistent and reproducible method of estimating reimbursements. However, not all patients in our study group had Medicare insurance, and this method likely underestimates true costs. Our estimates also do not include the opportunity costs created by patient anxiety or taking time away from work for medical appointments, nor do they include costs related to physician visits. As a result, the fully loaded cost of evaluating and managing incidentally found lesions is likely higher than what is reported here.

Our study period includes the first year and a half of the Coronavirus-19 pandemic. Throughout the pandemic, our providers continued seeing patients in person and through virtual clinic visits. We have no evidence that the referral patterns at our hospitals were systematically altered during this time, but it is possible that Coronavirus-19 precautions and social distancing deterred physicians from referring patients with incidental findings to our clinic. It is unclear what impact this possibility would have had on our findings, but the addition of one or two clinically important findings would meaningfully increase our estimates of prevalence owing to low frequency rates.

Finally, because our data are limited to patients presenting to orthopaedic oncology, our findings cannot be generalized to all patients with incidental osseous findings. As a result, the results of this study are only applicable to the population of patients being evaluated in the outpatient setting by practicing orthopaedic oncologists and consulting radiologists for incidentally found lesions

Clinically Important Findings

We found that a minority of incidentally found lesions are clinically important and a few result in treatment or are found to be malignant. Studies from outside orthopaedics have correlated the identification of incidental findings with increased patient distress [7, 12, 14]. Quantifying and documenting incidentally found osseous lesions can help orthopaedic oncologists counsel their patients and may alleviate unnecessary stress and anxiety among individuals with incidentally found lesions.

In the present series, most clinically important lesions were identified during the initial workup by orthopaedic oncology. One malignancy was diagnosed during the surveillance period, and serial imaging resulted in a change in management only twice. These findings suggest that incidentally found lesions warrant appropriate referral to orthopaedic oncology for risk stratification, after which patients believed to be at low risk of having malignancy may be judiciously followed with serial imaging. Although the practice of many orthopaedic oncologists is to follow low-risk lesions as described, there are no accepted guidelines for the management of incidental musculoskeletal findings, and the practice of orthopaedic oncologists is variable. The Society for Skeletal Radiology recently published a consensus statement introducing a proposed algorithm for the diagnostic management of incidentally found solitary lesions of bone encountered on CT and MRI [4]. The group devised four categories (called Bone-RADS) of incidental lesions, each of which was associated with a corresponding recommendation ranging from “leave alone” to “follow-up imaging” at 6 months after the index appointment, 6 months after the second appointment, and 12 months after the third appointment to “biopsy and/or referral to orthopaedic oncology” [4]. These recommendations were based on the consensus opinion of 12 radiologists and one orthopaedic oncologist, but they have not been validated clinically. Further research is warranted to characterize the outcomes of patients with incidentally found lesions and risk-stratify patients further according to evidence-based guidelines.

Value of Imaging Reimbursements

Our findings using a surrogate metric for the expense of surveillance revealed a low median value, but a highly skewed distribution and a long tail of high-cost outliers. This distribution was shaped by the relatively large number of patients who were lost to follow-up and a large difference between reimbursement for radiography and that for advanced imaging such as MRI. The annual reimbursements for surveillance for any patient could vary greatly, depending on the imaging modality used, but the average expense to the payor is relatively modest. Because of our use of Medicare reimbursements as a surrogate for payor expenses, however, these results most likely underestimate the true value of reimbursements. To our knowledge, no other studies have investigated the cost of imaging incidentally found osseous lesions. As a result, our findings can reassure the orthopaedic oncologist who routinely follows such lesions with serial imaging. Additional research is warranted to verify the cost-effectiveness of serial imaging at different intervals.

Conclusion

Among patients referred to orthopaedic oncology for incidentally found osseous lesions, the prevalence of clinically important findings (defined as those receiving biopsy or treatment and those found to be malignant) was approximately 7%. The proportion of lesions undergoing biopsy or treatment was 5% and 2%, respectively, and the percentage of lesions found to be malignant was less than 2%. For most patients, the clinical importance of their lesions can be recognized during the initial evaluation by orthopaedic oncology. Imaging surveillance may result in few changes in management or new diagnoses, but is associated with a low average annual expense. Therefore, we conclude that appropriate referral to orthopaedic oncology is warranted for the initial risk stratification of incidentally found lesions. Among patients deemed at low risk for malignancy but nonetheless who were indicated for surveillance, the prevalence of clinically important findings was low, and these patients might be judiciously followed with serial imaging without incurring an exorbitant cost to the payor. These results may support the clinical practice of orthopaedic oncologists and help reduce unnecessary stress and anxiety among their patients. Larger multicenter studies are warranted to refine our results and better risk-stratify patients to develop evidence-based guidelines for the surveillance of incidental musculoskeletal lesions.