Free Flap Reconstruction in the Era of Commercial Price Transparency – What are We Paying For?

Abstract

Background

Commercial rates for free flap reconstruction were not known publicly prior to the 2021 Hospital Price Transparency Final Rule. The purpose of this study was to examine commercial facility payments to characterize nationwide variation for microsurgical operations and identify opportunities to improve market effectiveness.

Methods

A cross-sectional study was performed using 2022 commercial insurance pricing merged with hospital performance data. Facility payment rates were extracted for nine CPT codes for free flap operations. Price variation was quantified via across-hospital ratios (AHRs) and within-hospital ratios (WHRs). Mixed effects linear models evaluated commercial rates relative to value, outcomes, and equity performance metrics, in addition to facility-level factors that included healthcare market concentration.

Results

20,528 commercial rates across 675 hospitals were compiled. AHRs ranged from 5.85–7.95, while WHRs ranged from 1.00–1.71. Compared to the lowest scoring hospitals (grade D), hospitals with an outcome grade of A and equity grades of B or C were associated with higher commercial rates (p<0.04); there were no significant differences in rate based on value. Higher commercial rates were also associated with nonprofit status and more concentrated markets (p<0.006). Lower commercial rates were correlated with safety-net and teaching hospitals (p<0.001).

Conclusion

Commercial rates for free flaps varied substantially both across and within hospitals. Associations of higher commercial rates with less competitive markets, and the lack of consistent association with value and equity, identify pricing failures. Additional work is needed to improve market efficiency for free flap operations.

Introduction

Commercial prices in healthcare have historically been hidden from public view, creating information asymmetry that prevents supply-side and demand-side forces from functioning according to classic economic principles. Health economist Uwe Reinhardt once likened buying healthcare in the US to shopping blindfolded in Macy’s and then receiving a bill months after leaving the store.^1,2 Indeed, without knowledge of prices, American healthcare consumers do not know what exactly they are purchasing and how much their selection costs the healthcare system. This is particularly problematic given evidence of substantial price variation throughout US healthcare, where high prices may not correlate with favorable endpoints such as high quality.³ Instead, high prices may be more reflective of monopolistic pricing behavior in consolidated markets,⁴ without any correlation with the value or equity of care provided.

The potential consequences of such pricing failure are especially salient for microsurgical reconstruction, which requires a sizable investment of time and resources from both the patient and provider. Prior studies utilizing nationwide databases have evaluated publicly-available charges for microsurgical operations.^5,6 Charges, also known as list prices or chargemaster rates, represent the amount billed for procedures rather than the actual amount paid; these amounts have been shown to be grossly discordant.⁷ Additionally, for microsurgical breast reconstruction, costs derived from cost-to-charge ratios demonstrate substantial nationwide variation, with higher cost associated with higher volume centers and fewer complications.⁸ Cost-to-charge ratios may yield estimates closer to actual payments; however, they are still highly inaccurate because ratios are estimated parameters and are not unique to any individual facility. Variation in the actual prices of free flap procedures and correlations with performance remain unexplored.

The objective of this study was to leverage data made public by the Hospital Price Transparency Final Rule of January 2021 to explore commercial payer-negotiated rates for microsurgical procedures. The Rule, which requires hospitals to disclose both discounted cash prices and negotiated rates with insurance companies for shoppable services, was intended to stimulate cost-based consumer choice and supply-side competition,^3,9 yet also provides the opportunity to examine the current state of pricing for free flap operations in the US. We hypothesized that given the current economic disconnects within US healthcare, prices for free flap operations would vary substantially and would not correlate with performance metrics.

Methods

Data Sources

We performed a cross-sectional study of pricing data based on a constructed dataset (Supplemental Figure 1). Data from Turquoise Health, a data service platform that aggregates price disclosures for every hospital in the US, formed the foundation of the dataset. This database has been previously used as the substrate of scientific analysis in peer-reviewed studies published within various academic journals.^7,10–14 As of October 2022, Turquoise Health data encompassed 60,755,118 negotiated rates from 65% (4,195) of US hospitals.¹⁵ Current Procedural Terminology (CPT) codes for free flap operations (Table 1) were used to query the Turquoise Health database to extract all disclosed rates for the corresponding procedures. Data from Maryland were excluded due to hospital price controls set by the state’s Global Budget Program.¹⁶ Data were then merged at the hospital level with data from the 2021 Lown Institute Hospitals Index, a ranking of hospital social responsibility that includes facility-level data for over 3,000 hospitals.¹⁷ Herfindahl-Hirschman Index (HHI) 2020 values for a sample of US metro areas were obtained from the Health Care Cost Institute and added to the dataset in order to quantify market concentration.^18,19 All rates were adjusted based on the US Center for Medicare & Medicaid Services’ (CMS) Geographic Adjustment Factor (GAF) to account for nationwide variation in input costs,²⁰ with the exception of Medicare rates that contain intrinsic price adjustment. Relative value units (RVUs) per CPT code were obtained from the CMS Physician Fee Schedule.²¹ This study did not require IRB approval as it did not involve human subjects research by Common Rule.

Table 1.

Current Procedural Terminology (CPT) codes for free flap operations

Code	Description
15756	Free muscle or myocutaneous flap with microvascular anastomosis
15757	Free skin flap with microvascular anastomosis
15758	Free fascial flap with microvascular anastomosis
19364	Breast reconstruction with free flap
20955	Bone graft with microvascular anastomosis; fibula
20956	Bone graft with microvascular anastomosis; iliac crest
20962	Bone graft with microvascular anastomosis; other than fibula, iliac crest, or metatarsal
43496	Free jejunum transfer with microvascular anastomosis
49906	Free omental flap with microvascular anastomosis

Variables

GAF-normalized commercial rate was the main outcome variable of interest. Dependent variables included three Lown performance metrics (value, outcomes, and equity), graded alphabetically where A represents the highest score and D is the lowest score. Methodology for determining Lown performance metrics is publicly available.¹⁷ In brief, value grade reflects cost efficiency (30- and 90-day mortality compared to the cost of care) and avoiding overuse (avoidance of 12 low-value services), weighted in a 3:2 ratio. Outcomes grade reflects clinical outcomes (risk-standardized rates of mortality and readmission), patient safety, and patient satisfaction, weighted in a 5:2:1 ratio. Equity grade reflects community benefit (spending on charity care and Medicaid), inclusivity (hospital population compared to local demographics), and pay equity, weighted in a 2:2:1 ratio. Additional dependent variables included status as a National Cancer Institute (NCI) Cancer Center or American College of Surgeons Level 1 Trauma Center; hospital bed count (extra-small 6–49, small 50–99, medium 100–199, large 200–399, extra-large 400+); safety-net status, defined as the 20% of hospitals with the highest proportion of dual Medicaid and Medicare beneficiaries; profit status; teaching status; and HHI. Lower HHI values indicate less concentrated and thus more competitive markets.¹⁸ HHI values were categorized to indicate low (≤1500), moderate (1501–2500), or high (>2500) market concentration based on US Department of Justice definitions.²² US census division was utilized as a covariate in analyses to control for geographic differences.

Statistical Analysis

Variation in commercial rate was measured using within-hospital ratios (WHRs) and across-hospital ratios (AHRs). Within-hospital ratios were defined as the median of the maximum divided by the minimum commercial rate per procedure code and hospital, as per prior methodology.^23–25 Similarly, across hospital ratios were calculated as the median of the 90^th percentile divided by the 10^th percentile median GAF-normalized rate across all hospitals per procedure code. Given the nonparametric distribution of data, median and interquartile ranges were used to quantify average commercial rate and variation per code. Commercial rates among payers were compared using Kruskal-Wallis tests. Violin plots were created to illustrate variation by payer.²⁶

For each CPT code, GAF-normalized commercial rates were divided by the respective CPT code RVU assignment to normalize prices to GAF-adjusted commercial dollars per RVU. These values were then mapped across the continental US Hospital Referral Regions (HRRs) with Tableau (Tableau Software LLC). Weights were assigned based on RVU to account for baseline differences in the value of procedures represented by CPT codes.

Mixed effects linear regression with random intercepts by procedure code was used to determine the association between median GAF-normalized commercial rate and Lown performance metrics, where the association with each metric (value, outcomes, and equity) was modeled using independent regressions. Random intercepts were utilized for procedure codes appreciating that heterogeneity between these codes is not correlated; we allowed inherent variation within each code to be expressed with its own intercept, as opposed to forcing all the codes into a single intercept. An additional multivariable mixed effects linear regression with random intercepts by procedure code modeled the association of median GAF-normalized commercial rate with hospital factors and HHI. The likelihood ratio (LR) test was used to evaluate mixed effects model fitness in comparison to a simple linear regression model. The Akaike information criterion (AIC) was used to gauge model parsimony. Analyses were conducted using Stata/SE Version 17.0 (StataCorp LLC). P-values of less than 0.05 were considered statistically significant.

Results

A total of 675 hospitals disclosed at least one commercial rate for nine CPT codes representing free flap operations, yielding a total of 20,528 unique commercial rates. Across all hospitals, 1.2% were NCI Cancer Centers, 5.3% were Level 1 Trauma Centers, 15.5% were safety-net hospitals, 69.2% were nonprofit, and 44.8% were teaching hospitals. Most hospitals were either of medium (100–199 beds; 24.7%) or large (200–399 beds; 31.1%) size. Among hospitals with HHI data, HHI distribution was 43.3% high concentration (HHI >2500), 23.5% moderate concentration (1501–2500) and 33.2% low concentration (≤1500) markets.

Table 2 displays median GAF-normalized commercial rates, within-hospital ratios, and across-hospital ratios per code. Omental free flaps (CPT 49906) had the lowest median commercial rate of $3,301.44 (IQR $2,079.10 – $6,916.22) while free flaps involving bone graft other than fibula, iliac crest, or metatarsal had the highest median rate of $5,343.35 ($2,949.54 – $9,276.19; Figure 1). Within-hospital ratios varied from 1.00 to 1.71, with iliac crest free flaps (CPT 20956) yielding the highest within-hospital ratios; across-hospital ratios ranged from 5.85 for free skin flaps (CPT 15757) to 7.95 for jejunal flaps (43496). Across all codes, the four most common commercial payers were Blue Cross Blue Shield, United Healthcare, Aetna, and Cigna. Commercial prices differed significantly across these payers for all CPT codes (p<0.001, Kruskal-Wallis; Figure 2). The nationwide distribution of GAF-normalized commercial dollars per RVU for all CPT codes is illustrated in Figure 3.

Table 2.

Descriptive statistics for commercial payer-negotiated rates for free flap procedures

			GAF-normalized commercial rate, $b			Within-hospital ratio (IQR)c	Across-hospital ratiod
Free Flap Procedure (CPT code)a	No. rates	No. hospitals	Median (IQR)	p10	p90
Muscle or myocutaneous (15756)	2,502	520	3,817.85 (2,359.27 – 7,157.23)	1,495.50	12,221.20	1.69 (1.00 – 3.75)	5.87
Skin (15757)	2,493	547	3,959.05 (2,382.85 – 7,724.32)	1,575.69	13,582.05	1.50 (1.00 – 3.75)	5.85
Fascial (15758)	2,310	507	3,665.19 (2,294.24 – 7,108.22)	1,491.91	11,758.97	1.65 (1.00 – 3.91)	6.21
Breast (19364)	2,493	526	4,862.91 (2,894.90 – 8,836.48)	1,858.70	14,700.78	1.57 (1.00 – 4.14)	6.95
Fibula (20955)	2,575	539	5,299.56 (2,883.30 – 9,505.32)	1,890.12	15,071.00	1.53 (1.00 – 3.82)	5.97
Iliac crest (20956)	2,485	525	4,992.91 (2,661.60 – 8,935.35)	1,518.48	14,166.15	1.71 (1.00 – 4.95)	6.40
Other bone (20962)	2,647	551	5,343.35 (2,949.54 – 9,276.19)	1,814.37	14,652.88	1.51 (1.00 – 4.26)	6.34
Jejunum (43496)	1,553	448	3,391.39 (1,978.54 – 6,967.45)	1,250.29	12,221.20	1.00 (1.00 – 3.71)	7.95
Omental (49906)	1,470	441	3,301.44 (2,079.10 – 6,916.22)	1,247.60	12,213.37	1.00 (1.00 – 3.61)	7.59

CPT, Current Procedural Terminology. GAF, Geographic Adjustment Factor. IQR, interquartile range.

^aCPT descriptions are abbreviated. For full descriptions, see Table 1.

^bCalculated based on the median payer-negotiated rates at each hospital.

^cCalculated as the median of the median of the maximum payer-negotiated rate divided by the minimum payer-negotiated rate per hospital.

^dCalculated as the median of the 90^th percentile median GAF-normalized rate divided by the 10^h percentile median GAF-normalized rate across all hospitals.

Figure 1. Variation in Median GAF-Normalized Commercial Rate by CPT Code.

Whiskers represent 10^th (lower) to 90^th (upper) percentiles. Horizontal line indicates median, circles indicate mean.

Figure 2. Variation in price among the four most common commercial insurers for free fibula (top), breast (middle), and jejunal (bottom) flaps.

BCBC, Blue Cross Blue Shield. United, United Healthcare. Additional CPT codes are not shown

Figure 3. GAF-Normalized Commercial Rate per RVU by Hospital Referral Region (HRR).

Image created using Tableau (Tableau Software LLC). Designs and/or data used to generate this image are attributed to Mapbox and OpenStreetMap (OSM); the latter uses data available under the Open Database License (openstreetmap.org/copyright).

Output for mixed effects linear regressions is shown in Table 3 for the respective Lown performance metrics. For value (i.e., cost efficiency based on 30- and 90-day mortality, and avoidance of low-value services), median GAF-normalized commercial rates were not significantly correlated with any grade. For outcomes (i.e., risk-standardized rates of mortality and readmission, patient safety, and patient satisfaction), median GAF-normalized commercial rate was significantly higher for grade A compared to grade D hospitals (p=0.032). For equity (i.e., community benefit, inclusivity, and pay equity), commercial rates for hospitals with B and C grades were significantly higher than those of hospitals with a D grade (p<0.009); hospitals with an A grade demonstrated no significant difference in commercial rate compared to hospitals with a D grade.

Table 3.

Mixed Effects Linear Regressions for GAF-Normalized Commercial Rate (US$) by Code for Lown Performance Metrics

	Coefficient	95% Confidence Interval	p-value
Value Grade
A	466.76	−2,475.09 – 3,408.61	0.756
B	454.03	−2,489.15 – 3,397.20	0.762
C	−314.22	−3,275.51 – 2,647.07	0.835
D	Ref	-	-
Outcomes Grade
A	3,219.90	277.44 – 6,162.35	0.032
B	2,425.19	−510.78 – 5,361.17	0.105
C	2,409.60	−557.78 – 5,376.97	0.111
D	Ref	-	-
Equity Grade
A	688.58	−318.47 – 1,695.63	0.180
B	1,374.88	413.47 – 2,336.29	0.005
C	1,294.74	338.73 – 2,250.74	0.008
D	Ref	-	-

GAF, geographic adjustment factor. Ref, reference. Note: horizontal lines separate independent regression. Value and Outcomes regressions, N=3,172. Equity regression, N=3,415.

Table 4 displays the output of the mixed effects linear regression for median GAF-normalized commercial rate as a function of facility-level variables, stratifying by procedure code. Higher commercial rates were associated with small compared to extra-small hospital size (coefficient $6,278.46, 95%CI $4,349.26 - $8,207.66, p<0.001), nonprofit status (coefficient $819.56, 95%CI $266.18 – $1,372.94, p=0.004), and HHI >2500 compared to ≤1500 (coefficient $1,934.88, 95%CI $1,237.02 – $2,632.74, p<0.001). Lower commercial rates were associated with safety-net hospitals (coefficient -$1,681.33, 95%CI -$2,415.83 – -$820.83, p<0.001) and teaching hospitals (coefficient -$1,022.90, 95%CI -$1,565.18 – -$480.62, p<0.001). In other words, median GAF-normalized commercial rate was -$1,618.33 lower for safety-net compared to non-safety-net hospitals and median GAF-normalized commercial rate was -$1,022.90 lower for teaching compared to non-teaching hospitals, controlling for CPT code.

Table 4.

Multivariable Mixed Effects Linear Regression for Median GAF-Normalized Commercial Rate (US$) by Code, Facility-Level Variables

	Coefficient	95% Confidence Interval	p-value
NCI Cancer Center
No	Ref	-	-
Yes	1,263.91	−543.15 – 3,070.97	0.170
Level 1 Trauma Center
No	Ref	-	-
Yes	−169.46	−1,180.95 – 842.03	0.743
Size (bed count)
Extra-small (6–49)	Ref	-	-
Small (50–99)	6,278.46	4,349.26 – 8,207.66	<0.001
Medium (100–199)	−424.36	−1,831.91 – 983.20	0.555
Large (200–399)	442.08	−959.50 – 1,843.65	0.536
Extra-large (400+)	−119.44	−1,593.11 – 1,354.22	0.874
Safety-net hospital
No	Ref	-	-
Yes	−1,618.33	−2,415.83 – −820.83	<0.001
Non-profit hospital
No	Ref	-	-
Yes	819.56	266.18 – 1,372.94	0.004
Teaching hospital
No	Ref	-	-
Yes	−1,022.90	−1,565.18 – −480.62	<0.001
HHI
≤1500	Ref	-	-
1501–2500	292.38	−455.05 – 1,039.82	0.443
>2500	1,934.88	1,237.02 – 2,632.74	<0.001
US census division
New England	Ref	-	-
Middle Atlantic	−1,641.79	−5,003.37 – 1,719.79	0.338
East North Central	−835.02	−4,190.91 – 2,520.87	0.626
West North Central	−523.47	−3,859.12 – 2,812.19	0.758
South Atlantic	2,290.91	−972.21 – 5,554.04	0.169
East South Central	−3,815.27	−7,133.57 – −496.97	0.024
West South Central	−4,425.99	−7,700.52 – −1,151.46	0.008
Mountain	−2,610.80	−5,901.05 – 679.46	0.120
Pacific	1,245.96	−2,084.80 – 4,576.72	0.463

GAF, geographic adjustment factor. HHI, Hirfindahl-Hirschman Index. NCI, National Cancer Institute. Ref, reference. N=2,071.

Discussion

In this cross-sectional study of commercial pricing data for free flap operations from 675 hospitals nationwide, we demonstrate substantial variation in price per procedure both within and across hospitals. For instance, for an iliac crest free flap (CPT 20956), the median negotiated rate with a given commercial insurer was on average 1.71 times that of another insurer for the same procedure at the same hospital. For a jejunal free flap (CPT 43496), the median rate negotiated by hospitals differed by a factor of 7.95 on average. Rates differed significantly by payer, though dispersion of rates for a given payer and code across hospitals and plans was also evident. Higher commercial rates were correlated with nonprofit status and more concentrated markets. Lower commercial rates were correlated with safety-net status and teaching hospitals. Higher commercial rates were also inconsistently correlated with better outcomes (grade A compared to D only) and equity (grade B and C compared to D only) and not associated with value.

The large degree of variation in commercial prices observed in our study is not unique to free flap operations. Our group has previously demonstrated similar magnitudes of variation for both autologous (within-hospital ratios 1.6 – 2.0, across-hospital ratios 4.5 – 11.2) and alloplastic (within-hospital ratios 1.9 – 2.5, across-hospital ratios 11.5 – 18.3) breast reconstruction;⁷ primary and secondary cleft lip and palate repair, and cleft rhinoplasty (within-hospital ratios 2.0 – 2.9, across-hospital ratios 5.4 – 13.7);¹³ and lymphedema excisional (within-hospital ratios 1.63 – 2.07, across-hospital ratios 4.60 – 6.29) and physiologic (within-hospital ratios 1.01 – 3.03, across-hospital ratios 5.23 – 10.36) procedures.¹⁴ Of note, the median GAF-normalized commercial rates and ratios presented in this paper for CPT 19364 differ from our prior publication,⁷ as this paper includes more updated pricing data. Other authors using analogous methodology for within- and across-hospital ratio calculations for otolaryngology procedures and telemedicine services have found comparable rates of variation.^23–25,27 Urologic procedures;²⁶ imaging studies, cardiac surgery, joint replacement, vaginal childbirth;²⁸ and shoppable surgical services in general¹² have additionally been the subject of commercial price variation research, though use of alternative methodology precludes direct comparison to our findings. Nonetheless, collectively these studies form a convincing body of evidence suggesting that there is enormous variation within the pricing structure of US healthcare, including that of microsurgical reconstruction.

Is commercial price variation for free flap operations problematic? Based on our results, the answer to this question is yes. Consistent with our previous findings,⁷ higher commercial rates were associated with more concentrated (i.e. less competitive) markets. While some argue that consolidations lead to market efficiencies that lower costs and/or improve quality through economies-of-scale,^29,30 data also support the opposing viewpoint that consolidations lead to higher prices reflective of monopolistic markets, without a benefit to value or equity.⁴ For breast reconstruction, greater market competition has been associated with increased odds of breast free flap reconstruction in general and among racial minorities.³¹ The relationship between market concentration, price, and equity is similarly tested in our study, where safety-net and teaching hospitals were associated with lower commercial rates. Though hospitals with an equity grade of B and C negotiated higher rates than those with a D grade, this association did not extend to the most equitable hospitals (grade A) in comparison to grade D facilities. In addition, though the detected association of higher commercial rates with better outcomes for grade A versus D hospitals was favorable, this correlation did not translate into an association of commercial rates with value. Since value in healthcare is defined as outcomes over costs,³² this suggests that outcomes did not improve sufficiently in relation to higher prices to overcome the counter effects of higher costs. It is also worth noting that higher prices were associated with nonprofit status; this may be an aberration, as we have found the opposite association in our prior work,⁷ or may reflect the increasingly blurred distinction between the behavior of for-profit and nonprofit hospitals in the US.³³ Overall, the associations, or lack thereof, between facility-level and performance metric data in this study do not provide adequately robust evidence to justify the large amount of variation in commercial rates for microsurgical reconstruction.

Limitations of our study are largely related to high rates of noncompliance with the Hospital Price Transparency Final Rule. Turquoise Health lacks data from 35% of US hospitals due to noncompliance.¹⁵ While early surveys of compliance indicated minimal bias in disclosure rates based on revenue status,³⁴ more recent data from acute care and cancer center-accredited hospitals suggest that compliant hospitals are more likely to have less revenue per patient per day, receive more Medicaid payments, and/or have lower average charge-to-cost ratios.^35,36 In January 2022, CMS increased the penalty for nondisclosure to a minimum of $300 per day and a maximum of $5,500 per day, depending on bed count;³⁷ nonetheless, some high-revenue hospitals likely find it advantageous to withhold disclosures of uniquely favorable rates despite the penalty. Furthermore, these rates encompass only facility prices and do not involve fees related to independent practitioners.³⁸ In addition, the Lown Institute definitions of value, outcomes, and equity are not specific to plastic surgery and omit several metrics key to measuring performance of microsurgical reconstruction, such as flap survival rate and health-related quality of life. Future studies should seek to correlate commercial rates with more responsive metrics.

Conclusion

Commercial rates for free flap procedures varied substantially within and across hospitals. Higher rates demonstrated significant associations with market concentration, while lower rates were correlated with safety-net and teaching hospitals. Despite some favorable associations with outcomes and equity, these performance metrics were not consistently associated with commercial rate, and value demonstrated no significant association. The long-term effects of price transparency in plastic surgery have yet to be determined. Though plastic surgery is unique among surgical specialties in its relatively high degree of preference sensitivity,³⁹ it remains to be seen whether price transparency initiatives alone will have the desired effect on lowering healthcare spending given the complicated nature of microsurgical reconstruction and effects of both brand and the physician/patient relationship on demand-side decision-making.⁴⁰ Additional measures beyond price transparency may be needed to reduce waste within the current pricing structure of free flap operations.