Israeli, Ron M.D.; Funk, Susan M.B.A.; Reaven, Nancy L. M.A.
Breast reconstruction with autologous tissue is integral to postmastectomy reconstruction. Autologous reconstruction involves the inherent complexity of a donor site, but patients and surgeons often regard autologous reconstruction as an immediate way to regain a breast that will age naturally over time, minimizing impact on patients’ lives. The bulk of the clinical literature on autologous reconstruction consists of case series from the clinical practices of one or, at best, a small number of surgeons who are typically very skilled at these complex procedures. Consequently, as a community, surgeons have limited information about reconstructive surgery costs and outcomes experienced by patients across the nation in the months and years following their procedures. This knowledge is essential for evaluating and ultimately improving patient outcomes and minimizing the cost of achieving consistent results.
As a first step in the evaluative process, we sought to study outcomes, resource use, and costs related to autologous breast reconstruction as broadly as possible, investigating the complications and services that patients experience, the rate at which patients return for these complications and services, and the associated costs. We examined 18-month postsurgical outcomes and costs of three common autologous flap procedures: latissimus dorsi flap with implant or expander, pedicled transverse rectus abdominis myocutaneous (TRAM) flap, and all variations of free flap breast reconstruction. We used national insurance claims data to obtain a divergent sample across geographic regions and care settings, avoiding self-selection bias. Our goal was to provide an assessment of current outcomes nationally to stimulate further investigation and discussion of those aspects of care in which efforts to improve outcomes would be the most fruitful.
PATIENTS AND METHODS
Using 2008 to 2009 insurance claims from U.S. employer-based private and Medicare plans obtained from Truven Health Analytics MarketScan Research Databases, we identified all patients who had a flap breast reconstruction procedure (designated as the “index event”) between January 1 and June 30, 2008. Included reconstructions were defined by Current Procedural Terminology or Healthcare Common Procedure Coding System code as follows: latissimus dorsi flap with implant or expander, 19361; pedicled TRAM flap, 19367, 19368, or 19369; and free flap (any), 19364, S2066 (gluteal artery perforator flap), or S2068 [deep inferior epigastric perforator (DIEP) flap or superficial inferior epigastric artery flap]. Patients were required to have 18 months of continuous insurance coverage following the index event, excepting death.
Patients were excluded if a postindex mastectomy was identified or if their initial reconstruction included a revision to a prior breast reconstruction procedure, a TRAM flap reconstruction involving an implant or tissue expander, or Current Procedural Terminology code 19366 (other reconstruction), because procedures identified by this code could not be defined with sufficient precision. Data were insufficient to differentiate unilateral and bilateral reconstructions.
Each patient’s health insurance claims for an 18-month postindex period (study period) were reviewed to identify services related to the breast reconstruction using diagnosis and procedure codes. Medical complications were identified by diagnosis code and grouped into categories for reporting purposes, as shown in Table 1. A medical complication was included if it occurred during a defined postreconstruction period, and was excluded if analysis of services between the breast procedure and the complication revealed a potential alternative cause. The original study design included postprocedural hernia as a complication; defining codes are listed in Table 1. After review, hernia complications were excluded from analysis because data were not sufficiently detailed to distinguish reconstruction-related hernias from coincidental hernia occurrences.
Episodes of care subsequent to the hospital stay for the initial flap procedure that involved a complication or a breast procedure were identified as postindex events. Events involving multiple complications were classified according to the highest priority complication present (Table 1). Events involving a breast procedure and treatment of a complication were classified according to the complication. In the analysis of patient complication rates, each complication was counted once per patient. In the analysis of event rates, each readmission or return visit for hospital or office care of a complication(s) and/or related breast procedure(s) was counted only once, irrespective of how many conditions or procedures were involved in the event. Postindex event results were normalized as rates per 100 patients to facilitate comparisons between study arms.
Anemia was identified by diagnosis codes occurring before or during the index episode. Other comorbidities were identified by the presence at any point during the 2-year data set of a diagnosis code confirming the condition or, in the case of smoking, obesity, or diabetes, a diagnosis, Current Procedural Terminology code, or Healthcare Common Procedure Coding System code confirming a related service (smoking cessation instruction, bariatric surgery, or insulin pump training, respectively).
Preindex radiation treatment was identified if a Current Procedural Terminology code for a radiation therapy service occurred before the index event, or by history of irradiation (V15.3). Postindex irradiation was identified by Current Procedural Terminology code only. The specific codes used to define radiation therapy services, comorbidities, and breast procedures are provided in Supplemental Digital Content 1. (See Document, Supplemental Digital Content 1, showing the flap analysis, http://links.lww.com/PRS/A943.)
Costs are presented from the health plan perspective as allowed cost reported at actual values in 2008 and 2009 dollars. Average cost is calculated per reconstructed patient to show the contribution of each type of return event to the average total cost per breast reconstruction patient over 18 months.
Results are presented as percentages, mean (SD), and event rates per 100 patients. Comparisons of population characteristics across study arms were conducted using either analysis of variance or chi-square tests, as appropriate. Comparisons of the occurrence rates of each complication across study arms were conducted using chi-square tests. In addition, comparisons of the occurrence of at least one complication across study arms, controlling separately for preindex or postindex radiation treatment, were performed using a logistic regression analysis and odds ratios, including 95 percent confidence intervals. Because of the inherently skewed nature of cost data and unequal cost variances across study arms, comparisons of costs across study arms were carried out using a nonparametric bootstrap randomization analysis (n = 5000 samples) for each type of complication. Nonparametric bootstrap 95 percent confidence intervals are provided for the grand total postindex cost of each complication. Length-of-stay comparison across study arms was performed using analysis of variance. Poisson regression analysis was used to compare the event rates per 100 patients across study arms for each type of complication. In addition, for each type of complication, 95 percent Wald confidence intervals are provided for the total and study arm event rates per 100 patients. All post hoc pairwise comparisons of study arms used the Bonferroni stepdown adjustment procedure of Holm,1 controlling the familywise error rate at 0.05. Values of p < 0.05 were considered to be significant. All statistical analyses were carried out using SAS/STAT software version 9.2 (SAS Institute, Inc., Cary, N.C.).
The study population consisted of 828 female patients, with 274, 302, and 252 patients receiving free flaps, latissimus dorsi flaps with implants or expanders, and TRAM flaps, respectively. Patients receiving free flaps were significantly younger (p < 0.05) than those receiving TRAM flaps (mean age, 48.9 years versus 50.7 years) (Table 2). There were no significant differences in rates of diabetes, smoking, or obesity between the three study arms. Latissimus dorsi flap patients were less likely to have a diagnosis of anemia than were free flap or latissimus dorsi flap patients, and free flap patients were less likely to have undergone postindex radiation treatment compared with TRAM or latissimus dorsi flap patients (Table 2).
Overall, 34.8 percent (288 of 828) of patients experienced at least one postindex complication; differences between study arms were not significant (Table 3). Infection (14.6 percent) and complications of the implant or graft (13.9 percent) were the most commonly reported complications. Differences between study arms were apparent for individual complications. Patients with a latissimus dorsi flap were more likely to experience graft-related complications (19.2 percent) and hematoma/seroma (6.3 percent) than were patients receiving either TRAM or free flaps, but the difference in hematoma complications was statistically significant only in relation to free flap patients (6.3 percent versus 1.8 percent; p < 0.05). Free and TRAM flap patients had higher rates of skin necrosis than did latissimus dorsi flap patients (7.7 percent and 6.4 percent versus 1.7 percent, respectively; p < 0.05). TRAM flap patients were more likely to experience wound complications, including the use of negative-pressure wound therapy devices, than either free flap or latissimus dorsi flap patients (5.6, 2.6, and 1.0 percent, respectively), but the difference was statistically significant only with respect to latissimus dorsi flap patients. No statistically significant differences between the study arms were found for infection, breast pain, procedural complications, or skin or degenerative tissue complications, other than necrosis.
Although differences were found in the proportion of patients whose newly constructed flaps were exposed to postindex radiation therapy, multiple comparisons found no relationship between radiation exposure and the incidence of complications. Adjusted odds ratios and 95 percent confidence intervals for the likelihood of a patient experiencing at least one complication are listed in Table 4. Despite the lack of significant differences in most of the per-patient complication rates, there were significant differences in the number of times patients returned for services associated with complications (Table 5). TRAM flap patients experienced the highest overall rate of return for treatment of complications, including the highest rates of return for treatment of infections, procedural complications, and wound complications or negative-pressure wound therapy. TRAM flap patients returned, on average, three times for treatment of infections and four times for wound complications during the study period. Return rates for procedures with no documented complication were high across all study arms but were highest for latissimus dorsi flap patients [116.6 postindex events in 100 patients versus 105.5 events in 100 patients for free flap patients (not significant) and 87.7 events in 100 patients for TRAM flap patients (significant versus latissimus dorsi and free flap patients; both p < 0.05)].
Return Visits: Breast Procedures
As noted in Table 5, latissimus dorsi flap patients experienced the greatest number of additional procedures overall (139.4 events in 100 patients), followed by free flap patients (117.9 events in 100 patients) and TRAM flap patients (98.0 events in 100 patients). Table 6 shows the rate per 100 patients of the various individual procedures performed during return visits, with or without complications, in the 18-month study period, segmented by study arm.
Table 7 summarizes costs related to index and postindex events. The mean total cost of the breast reconstruction, including the initial reconstruction and related downstream care over the 18-month study period, was $56,205 per reconstructed patient for free flap patients, $33,380 for TRAM flap patients, and $30,783 for latissimus dorsi flap patients (p < 0.05 for difference between free versus latissimus dorsi flap and free versus TRAM flap episodes). The index event was the major contributor to total costs, representing 78 percent of total cost for free flap patients, 73 percent of total cost for TRAM flap patients, and 67 percent of total cost for latissimus dorsi patients. Cost differences for the index event were highly significant between the three study arms (p < 0.0001). Mean length of stay for patients undergoing inpatient free flap procedures was significantly longer compared with both latissimus dorsi flap and TRAM flap procedures (free flap, 4.5 ± 2.45 days; TRAM flap, 3.9 ± 2.02 days; and latissimus dorsi flap, 2.5 ± 1.17 days; p < 0.05). Longer length of stay was a major contributor to the higher costs of free flap reconstructions. In addition, 11 percent of the latissimus dorsi flap reconstructions occurred in the outpatient setting, lowering the average cost.
Return visits contributed $10,297 to the average cost of flap breast reconstruction and its sequelae over the 18-month study period. These visits included treatments for documented complications and additional procedures with no reported complications. Total postindex costs per reconstructed patient (including complication treatments and noncomplicated additional procedures) averaged $11,987, $10,156, and $8605 for free, latissimus dorsi, and TRAM flap reconstructions, respectively (p = not significant). Infection was the costliest individual complication, contributing $2529 to the average 18-month cost of a TRAM flap reconstruction, $1390 to a free flap reconstruction, and $718 to a latissimus dorsi flap reconstruction (p = not significant). The only complication category that showed a significant difference between study arms was complications of implant/graft/mesh; latissimus dorsi flap patients had the highest costs of treatment, $2145 per patient versus $1178 and $774 for free and TRAM flap patients, respectively (p < 0.05, latissimus dorsi versus TRAM flaps).
In this analysis of 828 female patients undergoing reconstruction with free flaps, latissimus dorsi flaps with implants or expanders, or TRAM flaps after mastectomy, we found overall rates of complications for these complex procedures to be comparable to other published studies of flap procedures.2–6 To our knowledge, this is the first study to examine the number of return visits for treatment of individual complications. In this study, TRAM flap patients sought significantly more care for treatment of complications than either free or latissimus dorsi flap patients. However, the higher return rates associated with TRAM flap procedures did not translate into significantly higher costs of treatment. We found total costs higher, on average, and complication rates nonsignificantly lower for free flap versus latissimus dorsi flap patients, similar to Damen and colleagues’ finding of significantly higher short- and medium-term costs and significantly lower complication rates for DIEP compared with latissimus dorsi flaps with or without implants.7 Damen and colleagues excluded their initial 24 DIEP flap patients as a learning curve, whereas our study made no such exclusion.
Of interest is the frequency with which patients returned for reconstruction-related services that were unrelated to complications. Our study suggests that in actual practice, revisions and other procedures performed for reasons other than complications were quite common following all three types of flap reconstruction, occurring approximately once per patient on average. Additional procedures for reasons unrelated to complications are frequently mentioned in other studies but are generally considered unremarkable events to improve aesthetic outcomes. In our analysis, although TRAM flap patients were most likely to return for treatment of complications, they were least likely to return for additional procedures unrelated to complications. Latissimus dorsi flap patients were most likely to return for this category of services (117 events in 100 patients) compared with free (106 events in 100 patients) or TRAM flap patients (88 events in 100 patients). However, even within the least frequently returning group (TRAM flap), returns for procedures unrelated to complications occurred at a rate of nearly once per patient. Breast revision (Current Procedural Terminology code 19380) was the most frequently performed procedure during the postindex period for all flap procedures; at least half of all women experienced at least one revision. Surprisingly, the average cost of procedures unrelated to complications ($6752 per patient) overall was roughly double the cost of treating complications ($3545 per patient) for the three reconstructive approaches.
The relative frequency and cost of returns for revisions/procedures not related to complications raises two issues. First, there is a need to set appropriate expectations for these reconstructive procedures. The American Society of Plastic Surgeons notes in its consumer information regarding breast reconstruction that, “In some situations, it may not be possible to achieve optimal results with a single surgical procedure and another surgery may be necessary.”8 Our results suggest that nearly all patients return for additional services. The consequence for failing to acknowledge that breast reconstruction procedures typically require more than one procedure to recreate a natural-appearing and natural-feeling breast is that we risk failing our patients’ expectations. We also invite health insurers to erroneously assume that flap breast reconstructions are “one-and-done” procedures, to reimburse accordingly, and to consider all postprocedure services either complications or purely aesthetic in nature and therefore subject to noncoverage policies or other sanctions.
Second, issues concerning appropriate patient selection abound. Our findings suggest that the reasons for downstream procedures may be fundamentally different for each of the three flap procedures under examination. Latissimus dorsi flap patients had higher return rates for complications related to the expander/implant and hematomas and the highest rate of downstream procedures unrelated to complications, whereas TRAM flap patients had higher return rates for infections, procedural/wound problems, and degenerative skin or tissue problems and the lowest rate of downstream procedures unrelated to complications. Rates for free flap patients were generally in the middle. Kroll9 found that better patient selection reduced the rate of breast necrosis and flap loss by nearly half in patients receiving DIEP flaps. Garvey et al.10 found better outcomes with DIEP flaps versus pedicled TRAM flaps irrespective of patient risk factors. If, as our findings suggest, the challenges inherent in each type of flap procedure are fundamentally different, patient selection has to be parsed carefully. Finally, rates of complications remain high across all procedures. New techniques and technologies that show promise in mitigating complications should be encouraged and rigorously evaluated.
Studies using insurance claim databases have inherent limitations because analyses are limited to services reported by diagnosis and procedure codes, which lack the precision of medical chart data and can be manipulated to maximize reimbursement. In particular, existing diagnosis codes do not provide sufficient clarity to isolate flap necrosis (an important outcome metric in flap reconstructions), to easily differentiate mastectomy and donor-site complications, or to distinguish between donor-site and comorbid hernias (a limitation that directly impacted our ability to analyze and compare rates of reconstruction-related hernias). Finally, identification of patients having prereconstruction radiation therapy may be incomplete because of limited availability of prereconstruction data. The retrospective nature of analysis makes it difficult to directly control for other factors that may influence outcomes. Insurance claims data offer a means of preliminarily assessing alternative surgical approaches in large patient populations and allow for comparison of alternative surgical approaches performed contemporaneously in multiple centers and geographic regions. This is particularly important when most of the available evidence about surgical procedures is based on single-site data from experienced surgeons.
Our analysis of 828 patients across the United States who had undergone autologous breast reconstructions in the first half of 2008 illuminates the complexity of these procedures. Return visits for procedures not involving documented complications occurred at rates suggesting that patients, the surgical community, and the health care system would benefit by recognizing the inherently staged nature of these reconstructions. Additional research should focus on better characterizing these additional procedures and the role of patient selection in improving outcomes.
Editorial support for this article was provided by Peloton Advantage, LLC (Parsippany, N.J.), and funded by LifeCell Corp. (Branchburg, N.J.).
The opinions expressed in this article are those of the authors.