Low-intensity pulsed ultrasound (LIPUS) can decrease the economic burden of fracture non-union

Abstract

Objectives:

Few studies have evaluated the economic burden of surgical and conservative treatment of fracture non-union. An analysis was undertaken of aggregated payer data to determine economic costs of non-unions treated with surgery only vs non-unions treated conservatively with low-intensity pulsed ultrasound (LIPUS) only.

Methods:

This study used administrative claims from a health plan database including nearly 80 million people. Patients with a claim for non-union surgery or LIPUS for non-union were identified, from April 2007 until April 2010. A retrospective cohort was formed by pairwise demographic matching among patients who received ‘Surgery Only’ or ‘LIPUS Only’. Date of the first non-union intervention (surgery or LIPUS) was defined as the index date. All medical costs were assessed over 12 months following the index date for the ‘Surgery Only’ and ‘LIPUS Only’ cohorts.

Results:

A total of 1158 matched patients were identified. ‘Surgery Only’ patients used significantly more healthcare services. In the year following intervention, ‘Surgery Only’ patients had total medical costs $6289 higher than ‘LIPUS Only’ patients (Mean = $11,276 vs $4986; p < 0.0001). Outpatient costs accounted for >68% of overall costs in both cohorts, and outpatient costs were significantly higher among the ‘Surgery Only’ cohort (Mean = $7682 vs $4196; p < 0.0001). Total inpatient costs were also significantly higher among the ‘Surgery Only’ cohort (Mean = $3302 vs $381; p < 0.0001).

Limitations:

Limitations of this work are typical of all studies based on administrative claims data: errors in the database are assumed to distribute randomly between cohorts, and some patients may have been miscoded as to treatment received or costs billed.

Conclusions:

‘Surgery Only’ patients used significantly and substantially more healthcare services in treatment of fracture non-union. Conservative treatment with ‘LIPUS only’ for fracture non-union could potentially result in cost savings projected to $4 billion dollars annually in the US.

Keywords::

• Health economics
• Aggregated payer data
• Surgery
• Ultrasound

View correction statement:

Corrigendum to: Low-intensity pulsed ultrasound (LIPUS) can decrease the economic burden of fracture non-union

Introduction

An estimated 6 million fractures occur each year in the US1. These fractures result in 14 million visits to an emergency room or other healthcare facility, at an estimated annual cost of $21 billion2. While most fractures heal, an estimated 5–10% result in non-union3–6 and symptomatic non-unions require secondary intervention3^,5^,7^,8. Non-unions can be devastating; a tibial non-union was judged by patients to be more disabling than congestive heart failure9.

Most symptomatic non-unions are treated surgically5. Risks and adverse events associated with surgery for non-union are well documented, and can include hardware failure, revision surgery, infection, post-traumatic arthritis, blood clots, post-operative pain, osteomyelitis, neurovascular injury, bleeding, and lengthy rehabilitation10–14. Several non-surgical options are available to treat non-unions, including low-intensity pulsed ultrasound (LIPUS).

LIPUS is FDA-approved for treating fracture non-unions (excluding skull and vertebrae) in which there are no progressive signs of healing15. LIPUS has been shown to stimulate healing of established non-unions, but it can also accelerate healing of fresh fractures and minimize delayed healing16. Level I clinical studies demonstrate the ability of a specific ultrasound signal (1.5 MHz ultrasound pulsed at 1 kHz, 20% duty cycle, 30 mW/cm² intensity) to accelerate fracture healing in the tibia and radius by up to 40%16. LIPUS accelerates all stages of the fracture repair process, including inflammation, soft callus formation, hard callus formation, and mineralization, probably through a mechanism of enhanced COX-2 production16. There are no known contraindications to LIPUS; advantages of LIPUS treatment could potentially include reduced risks associated with avoidance of surgery and a favorable cost/benefit ratio.

Few prior studies have evaluated the economic burden resulting from non-surgical management of non-union and none have done so with a large sample size. Here, we use an econometric approach to evaluate the cost of LIPUS or surgery for fracture non-union in a database of ∼80 million patients.

Patients and methods

Data source

Data were used from the IMS Health LifeLink Health Plan Claims Database (Plymouth Meeting, PA), which contains fully adjudicated medical and pharmaceutical claims for ∼80 million patients in 80 healthcare plans across the US. Only healthcare plans submitting data for all members are included in the database, ensuring complete data capture and representative samples. Data are longitudinal, with an average member enrollment time of 2–3 years. Charged, allowed, and paid amounts are available for all services rendered. In addition: (1) demographic data (age, gender, geographic region); (2) health plan type (e.g., health-maintenance organization (HMO), point-of-service plan (POS), preferred-provider organization (PPO), consumer-directed health plan, indemnity, and unknown); (3) payer type (e.g., commercial, self-pay); and (4) provider specialty are also available. Records in the LifeLink database are accepted as representative of the national, commercially-insured population of the US.

In compliance with the Health Insurance Portability and Accountability Act (HIPAA), patient data included in this analysis were de-identified; therefore, this study was exempt from Institutional Review Board (IRB) review.

Patient selection

This retrospective cohort study includes all patients with at least one claim for LIPUS or non-union surgery between April 1, 2007 and March 31, 2010. The date of the first intervention (LIPUS or surgery) for non-union was defined as the index date. Patients were required to have a minimum of 12 months of continuous health plan enrollment before and after the index date.

Patients were included in the study if they had evidence of a fracture in any bone except the skull or vertebrae during the pre-index period (Figure 1). Evidence of non-union fracture could include: (1) International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) coded diagnosis of a non-union/malunion fracture; OR (2) ICD-9-CM coded diagnosis of a fracture >90 days prior to a relevant index intervention. General health of patients was quantified using the Charlson Comorbidity Index (CCI), a well-validated index that characterizes comorbid conditions which might alter the risk of mortality in longitudinal studies17.

Figure 1. CONSORT diagram showing how the cohorts of matched patients were obtained.

Patients were excluded from study if there was: (1) a skull or vertebral fracture; (2) any surgical or other intervention in the pre-index period; (3) a claim for treatment of non-union by electrical stimulation (EStim) during the index period; (4) evidence that the patient was aged 65 or more and not covered by Medicare Risk; (5) evidence of Medicare Cost coverage; OR (6) if there were problems with missing data or data quality (Figure 1).

Patients were classified into either a ‘LIPUS Only’ cohort (which had an index intervention limited to LIPUS, with no evidence of surgery until at least 12 weeks after the index date, to preclude concurrent surgery for non-union) or a ‘Surgery Only’ cohort (which had an index intervention limited to surgery, with no evidence of LIPUS or other non-invasive bone stimulation device until at least 4 weeks after the index date, to preclude concurrent LIPUS). The discrepancy in time to second intervention arose because it can take longer to assess failure following LIPUS than to assess failure following surgery.

Cohort matching

For each ‘LIPUS Only’ patient, a ‘Surgery Only’ patient was identified who was as close as possible to an exact match with respect to key demographic variables, established at the index date. Key demographic variables used were: (1) age class; (2) gender; (3) bone fractured (humerus, radius/ulna/wrist, femur, tibia/fibula/ankle, hand/fingers, foot, other/unknown); (4) health plan type; (5) pre-index 12-month total healthcare costs (medical and pharmacy) to control for comorbid diseases present before the fracture; (6) evidence of prior internal fixation (yes/no) to control for fracture severity; and (7) evidence of ankle fusion (yes/no) to control for fracture severity. If a ‘Surgery Only’ patient matched more than one ‘LIPUS Only’ patient, the ‘Surgery Only’ patient was randomly associated with one ‘LIPUS Only’ patient, to create a 1:1 ratio of pair-matched patients. Patients who could not provide a match were dropped from the analysis (Figure 1).

Calculation of direct medical costs

Direct medical costs are those costs associated directly with the medical care rendered. All-cause resource utilization was evaluated for each of the following services:

Pharmacy services

• Medication classes of interest (analgesics (NSAIDs and opioids), anticoagulants, anti-diabetics, anti-hyperlipidemics, anti-hypertensives, arrhythmia medications, β-blockers, bisphosphonates, calcium channel blockers, corticosteroids, diuretics, vasopressors).

• All other pharmacy services.

Outpatient services

• Physical therapy (non-union fracture-related only, including therapeutic exercises, physical or manipulative therapy, all services by a qualified physical therapist).

• Diagnostic tests of interest (non-union fracture-related only, including X-ray, CT, MRI).

• Emergency room visits.

• Physician office visits.

• Laboratory and pathology tests.

• Radiology services.

• Surgical services.

• Ancillary services.

Inpatient hospitalizations

• Hospital admissions.

• Total days in hospital.

• Average length of stay.

All costs are calculated from administrative claims for medical services. Because claims are fully adjudicated by the insurance company, and because making false claims can incur a legal penalty, the dollar amounts charged, allowed, and paid are thought to be reliable for patients in a covered population. This enabled us to calculate pre-index medical and pharmacy costs, as well as post-index medical and pharmacy costs.

Statistical analyses

P-values using McNemar’s test for categorical variables and the Wilcoxon signed-rank test for continuous variables were generated to evaluate differences in outcome between the matched cohorts. Non-parametric testing was used because data were likely non-normally distributed; we did not test for normality of variables in this study because such tests typically have low power to detect departures from normality. A p-value of <0.01 was considered statistically significant, and all p-values are nominal, unadjusted for multiple testing. All analyses were based on observed, not projected, data and analyses used SAS version 9.2 software (SAS, Cary, NC).

Results

Patient demographic and clinical characteristics

Cohorts were indistinguishable in the demographic variables used for matching (Table 1). Roughly 59.4% of all patients were females. The most common fracture was of the foot (37.7% of all patients), with fractures of the leg next most common (29.9%). The average age of LIPUS patients was 43.7 years old (±15.4 SD), which does not differ significantly from the average age of surgery patients (43.6 ± 15.8). While we know age to have an important impact on fracture healing by LIPUS18, those effects were balanced in this study. Roughly 77% of patients were enrolled in PPOs, with a significant presence in the southern US. Pre-index total healthcare costs were virtually identical between the cohorts (Table 1), suggesting that cohorts were well-matched for comorbid illness at baseline.

Table 1. Demographic characteristics used to match ‘LIPUS Only’ and ‘Surgery Only’ cohorts.

Characteristic	LIPUS Only (n = 579)		Surgery Only (n = 579)
Age group: (n, %)
<18 years	40	6.9%	40	6.9%
18–34 years	125	21.6%	125	21.6%
35–44 years	84	14.5%	84	14.5%
45–54 years	147	25.4%	147	25.4%
55–64 years	181	31.3%	181	31.3%
65+years	2	0.3%	2	0.3%
Gender: (n, %)
Female	344	59.4%	344	59.4%
Fracture type: (n, %)
Foot	218	37.7%	218	37.7%
Tibia, fibula, ankle	173	29.9%	173	29.9%
Radius, ulna, wrist	92	15.9%	92	15.9%
Other/Unknown	50	8.6%	50	8.6%
Femur	19	3.3%	19	3.3%
Humerus	18	3.1%	18	3.1%
Hand, fingers	9	1.6%	9	1.6%
Health plan type: (n, %)
PPO	446	77.0%	446	77.0%
HMO	71	12.3%	71	12.3%
POS	49	8.5%	49	8.5%
Indemnity	7	1.2%	2	0.3%
Consumer-directed	4	0.7%	3	0.5%
Unknown	2	0.3%	8	1.4%
Pre-index medical cost:
Mean (SD)	$19,054 ($36,769)		$19,296 ($35,745)
Median	$7,417		$7,367
Pre-index pharmacy cost:
Mean (SD)	$2,191 ($5,276)		$2,322 ($5,136)
Median	$469		$488
Prior internal dixation: (n, %)	138	23.8%	138	23.8%
Prior ankle fusion: (n, %)	0	0.0%	0	0.0%

Orthopedic surgeons were more likely to treat patients using surgery only (Table 2), while other physician specialties were more likely to treat using LIPUS (p < 0.0001). Most patients had a Charlson Comorbidity Index (CCI) score of 0 (Table 2), indicating a healthy cohort of patients with few co-morbidities that could affect outcome, although diabetes was somewhat more prevalent among LIPUS Only patients (Table 2).

Table 2. Patient pre-treatment characteristics for the matched cohorts.

Characteristic	LIPUS Only (n = 579)		Surgery Only (n = 579)		p Value
Payer type: (n, %)
Commercial	510	88.1%	505	87.2%	0.06
Medicaid	3	0.5%	16	2.8%
Medicare Risk	3	0.5%	6	1.0%
Self-insured	61	10.5%	47	8.1%
Unknown	2	0.3%	5	0.9%
Physician specialty: (n, %)
Generalist	59	10.2%	79	13.6%	<0.0001
Podiatrist	114	19.7%	35	6.0%
Orthopedist	12	2.1%	23	4.0%
Orthopedic surgeon	223	38.5%	354	61.1%
Other surgeon	22	3.8%	37	6.4%
Physical/occupational therapist	23	4.0%	10	1.7%
Other physician specialty	117	20.2%	36	6.2%
Unknown physician specialty	9	1.6%	5	0.9%
Charlson Comorbidity Index: (n, %)
Score 0	395	68.2%	391	67.5%	0.16
Score 1	79	13.6%	100	17.3%
Score 2–3	81	14.0%	64	11.1%
Score ≥4	24	4.1%	24	4.1%
Mean (SD)	0.7 (1.3)		0.7 (1.4)
Pre-Index Comorbidities: (n, %)
Diabetes	77	13.3%	59	10.2%	0.08
Renal failure	10	1.7%	9	1.6%	0.80
Chronic kidney disease	13	2.2%	6	1.0%	0.11
Osteoporosis	19	3.3%	29	5.0%	0.14
Metabolic bone disease	20	3.5%	31	5.4%	0.13
Osteomyelitis	19	3.3%	22	3.8%	0.63
Stress fractures	36	6.2%	30	5.2%	0.46

Healthcare resource utilization

Important differences between cohorts emerged in the post-index period (Table 3). Surgery Only patients had more pharmacy prescriptions (p = 0.0006), and more opioid analgesic prescriptions (p < 0.0001) in the post-index period. In addition, Surgery Only patients had more diagnostic testing (p < 0.0001), and used more ancillary services (p < 0.0001) than LIPUS patients.

Table 3. Non-union-related healthcare resource utilization in the 12-month post-index period. Patients were excluded from this analysis if their index fracture was coded as ‘Other’ or ‘Multiple’ or if they had evidence of a post-index fracture in a non-index location. The p-value tested mean values.

	LIPUS Only (n = 408)^a		Surgery Only (n = 408)^a		p Value*
	Mean (SD)	Median	Mean (SD)	Median
Pharmacy services (total scripts)	4.2 (8.7)	1	5.5 (8.8)	2	0.0006
Analgesics—Non-narcotic	0.1 (0.08)	0	0.1 (0.6)	0	0.67
Analgesics—Opioid	2.8 (7.3)	0	4.2 (7.3)	2	<0.0001
Analgesics—Anti-inflammatory	0.7 (2.0)	0	0.7 (2.0)	0	0.77
NSAIDS	0.6 (1.7)	0	0.7 (1.9)	0	0.62
Bisphosphonates	0.4 (1.6)	0	0.2 (1.1)	0	0.28
Outpatient services
Physician Office Visits	3.7 (6.7)	2	3.9 (7.7)	1	0.67
Laboratory and Pathology Claims	0.2 (0.9)	0	0.2 (1.5)	0	0.60
Radiology Examinations	0.3 (0.9)	0	0.2 (0.8)	0	0.06
Physical Therapy	0.7 (3.5)	0	1.3 (5.4)	0	0.02
Diagnostic Testing	3.3 (2.8)	3	4.9 (3.3)	4	<0.0001
Ancillary Services	1.7 (5.4)	0	3.0 (7.8)	1	<0.0001

Direct medical costs

Non-union-related costs were significantly higher for the Surgery Only cohort ($11,276 vs $4986; p < 0.0001) compared to the LIPUS Only cohort (Table 4). The mean difference in the cost of the intervention itself was a significant contributor to this observed difference (Surgery Only = $9178 vs LIPUS Only = $3096; p < 0.0001). Significant differences were also observed between the cohorts related to diagnostic testing, surgical services, and ancillary services (all p < 0.0001).

Table 4. Total non-union-related direct medical costs in the 12 month post-index period. Patients were excluded from this analysis if their index fracture was coded as ‘Other’ or ‘Multiple’ or if they had evidence of a post-index fracture in a non-index location. The p-value tested mean values.

	LIPUS Only (n = 408)^a		Surgery Only (n = 408)^a		p Value*
	Mean (SD)	Median	Mean (SD)	Median
Overall costs for patients	$4,986 ($5,051)	$3,829	$11,276 ($12,213)	$7,833	<0.0001
Pharmacy costs	$409 ($2,023)	$4	$292 ($1,169)	$22	0.0002
Analgesics—Non-narcotic	$4 ($44)	$0	$1 ($7)	$0	0.98
Analgesics—Opioid	$165 ($1,005)	$0	$167 ($660)	$16	<0.0001
Analgesics—Anti-inflammatory	$201 ($1,716)	$0	$99 ($944)	$0	0.77
NSAIDS	$41 ($246)	$0	$49 ($218)	$0	0.73
Outpatient costs	$4,196 ($2,910)	$3,670	$7,682 ($6,643)	$5,918	<0.0001
Physical Therapy	$35 ($205)	$0	$83 ($400)	$0	0.02
Diagnostic Testing	$246 ($395)	$118	$302 ($403)	$185	<0.0001
Emergency Room Visits	$15 ($117)	$0	$40 ($420)	$0	0.50
Physician Office Visits	$388 ($949)	$102	$408 ($937)	$66	0.41
Laboratory and Pathology	$13 ($156)	$0	$5 ($47)	$0	0.21
Radiology Examinations	$27 ($152)	$0	$25 ($151)	$0	0.64
Surgical Services	$225 ($825)	$0	$329 ($857)	$0	<0.0001
Ancillary Services	$328 ($1,236)	$0	$340 ($1,485)	$33	<0.0001
Inpatient costs	$381 ($3,296)	$0	$3,302 ($11,343)	$0	<0.0001

Current estimates suggest that there are up to 600,000 non-unions per year in the US8. Based on the increased cost of surgery of $6290 per patient (Table 4; overall cost of surgery of $11,276 – overall cost of LIPUS of $4986), the additional cost from surgery would be $3.77 billion annually in the US alone.

Discussion

Patients who received surgery or LIPUS were well matched at the index intervention, both in terms of demographic characteristics (Table 1) and in terms of pre-treatment co-morbidities (Table 2). However, fracture-related healthcare utilization was significantly and substantially higher among surgery patients after intervention (Table 3). This resulted in overall medical costs that were more than 2-fold higher for patients who received surgery only (Table 4).

In the current era of medical cost containment, objective data such as we provide are compelling, but largely lacking in the literature19. Much of the recent econometric literature focuses either on the cost of treating long bone non-union using bone morphogenetic proteins as an adjunct to surgery20–22 or of contrasting various surgical techniques23–25.

Some prior work has assessed the economic cost of LIPUS compared to surgery, but that research was limited to tibial fractures and was done long ago, so that absolute surgical costs are no longer relevant and surgical treatments have evolved26. Nevertheless, a substantial cost savings was reported from using LIPUS as an adjunct to conservative or surgical treatment of delayed union in tibial fracture26. In a randomized clinical trial of closed or grade-I open tibial shaft fractures, LIPUS was found to accelerate healing by ∼4 weeks, compared to patients who received similar treatment without LIPUS27. Acceleration of tibial union was associated with a substantial cost reduction because fewer patients required surgical revision and because there were smaller payouts for worker’s compensation26. Rapid healing of tibial fracture was associated with less loss of patient productivity23 and may also have reduced the risk of fracture non-union. Economic costs of tibial non-union are high, especially if indirect costs and the monetized costs of reduced quality-of-life are included in the calculation19. Effective treatment for tibial fracture can substantially reduce individual patient costs24 and small improvements in individual patient costs can result in large governmental savings25. Strategies that reduce fracture healing time, so that patients can return to work more quickly, are likely to improve medical outcome for the patient while reducing the financial burden to society28.

Direct medical costs of non-unions are substantial (Table 4), and indirect costs (e.g., loss of work, use of physical therapy, follow-up medical intervention) further increase total costs19. Our findings indicate that it may be more economical to use LIPUS in lieu of surgery in treating patients with non-union. Current estimates suggest that there are up to 600,000 non-unions per year in the US8. Based on the increased cost of surgery of $6290 per patient (Table 4; overall cost of surgery of $11,276 – overall cost of LIPUS of $4986), the additional cost from surgery would be roughly $3.77 billion annually in the US alone.

Nevertheless, LIPUS will not be an attractive treatment option if the heal rate is substantially lower than the heal rate that can be attained with surgery, even if surgery is more expensive. Overall heal rate of non-union surgery was characterized in a meta-analysis3. Surgical treatment of non-union was reported in 23 studies published between 1961–1997, and the overall heal rate for 2041 patients was 86%3. An 87% heal rate was reported for LIPUS in eight studies published between 2000–2007, which evaluated 594 patients15. While current clinical evidence in support of LIPUS rarely takes the form of a randomized clinical trial29, the National Institute for Health and Clinical Excellence (NICE) Committee in the UK recently recognized the clinical efficacy and health economic value of LIPUS for long bone non-union30.

The strength of our work is that it directly addresses a clinical need. NICE evaluated the clinical literature supporting fracture healing with LIPUS and judged it sufficient to endorse LIPUS as a cost-effective treatment compared with competing treatments for non-union30. The National Institute for Health and Care Excellence (NICE) Committee noted that any non-union treatment, ‘that avoids or reduces the need for surgery is of significant benefit to patients and also has potential advantages to the [National Health Service] in terms of resource use’ (p. 12). However, NICE was unable to identify any prior LIPUS studies that used surgery avoidance as an outcome measure30. Our work goes some way towards filling that gap, as we contrast patients who received LIPUS only with patients who received surgery only.

Limitations of our work are typical of all studies based upon administrative claims data31. Errors in the database certainly exist, but are assumed to distribute randomly between cohorts. Some patients may have been miscoded as to treatment received, and the costs of treatment might be improperly allocated between treatments of interest. Patients also could have been misclassified as to diagnosis. Because of limited use of the specific ICD-9 code for fracture non-union, non-unions were also identified using proxy measures, including evidence of a prior fracture or osteotomy procedure >90 days prior to the index intervention. Therefore, we cannot be certain that an intervention was actually for a non-union. If procedure codes were missing, there is a potential for under-reporting utilization of fracture-related services. However, due to the large size of the healthcare plans in the database, and to the fact that costs are adjudicated carefully for pecuniary reasons, these factors should not have impacted our study results any more than in other administrative claims studies. This study was not designed to assess treatment outcomes, so we cannot determine whether the competing treatments yielded equivalent outcomes. Finally, we note that our results are not generalizable to countries other than the US, because medical costs and allocation of services differ elsewhere.

Our study provides unique information regarding the economic advantage of treating non-union by LIPUS rather than surgery. Additional research is needed to evaluate our conclusions in other claims databases, as well as to characterize the clinical outcomes related to each type of treatment.

Conclusions

Patients with fracture non-union who were treated with either surgery or LIPUS were well-matched at baseline, both in terms of demographic characteristics and in terms of pre-treatment co-morbidities. However, fracture-related healthcare utilization was significantly and substantially higher among patients who received surgical intervention. This resulted in overall medical costs that were more than 2-fold higher for patients who received surgery only. Conservative treatment with ‘LIPUS only’ for fracture non-union could potentially result in cost savings projected to $4 billion dollars annually in the US.

Abbreviations
LIPUS	=	low-intensity pulsed ultrasound
HMO	=	health-maintenance organization
POS	=	point-of-service plan
PPO	=	preferred-provider organization
HIPAA	=	Health Insurance Portability and Accountability Act
IRB	=	Institutional Review Board
ICD-9-CM	=	International Classification of Diseases, Ninth Revision, Clinical Modification
CCI	=	Charlson Comorbidity Index
NICE	=	National Institute for Health and Clinical Excellence

Transparency

Declaration of funding

This study was funded by Bioventus, LLC.

Declaration of financial/other relationships

SM is a consultant for Bioventus, which makes a LIPUS device, and he has received speaker fees and travel reimbursement. MD and ES are IMS Health employees, but IMS was contracted by Bioventus to produce the analysis upon which this paper is based. KL and RGS are both Bioventus employees. None of the authors have any non-financial competing interests.

Acknowledgments

We would like to thank Dr Peter Heeckt of Bioventus for medical advice and Katharine Coyle of IMS Health for writing assistance.