Costs of diagnostic and preoperative workup with and without breast MRI in older women with a breast cancer diagnosis

The high incidence of female breast cancer in the United States (U.S.) – estimated to affect 232,670 women in 2014 - [1] results in an aggregate economic burden of care relative to other cancers that is among the highest of all cancers in women [2]. While the majority of breast cancer treatment related costs occur in the first 12 months following diagnosis [2], costs for women as they are diagnosed and deciding upon initial diagnostic/preoperative treatment have not been characterized. Women with breast cancer and their providers face a number of decisions during the diagnostic and preoperative period, including which diagnostic testing, workup and primary treatment options they will pursue. Cost is one of several key factors cited by women with breast cancer as important in weighing their breast cancer care options [3].

Breast MRI has been reimbursed by Medicare and other payers since 1991for women with a breast cancer diagnosis, but only in the past 10 years has its use Fibecome widespread in clinical workup, increasing dramatically from an estimated 1 % in 2001 to between 21 and 50 % of breast cancer cases by 2010, with considerable regional variability [4‐8]. The utility of breast MRI is its high sensitivity – especially compared to mammography – conferring an increased ability to determine extent of disease and detect occult cancer among women with a breast cancer diagnosis [9]. However, breast MRI also increases detection of benign and low malignant potential lesions, which often require further workup in order to make a final assessment [9]. This workup potentially leads to increased costs that may not be offset by other benefits. Preoperative MRI remains controversial in part due to: lack of evidence for improved outcomes, increased potential for false-positives, and the possibility of surgical overtreatment [10‐13]. Given the increasing prevalence of breast MRI during the diagnostic/preoperative period even amidst uncertainty regarding harm and benefit tradeoffs, comparing effectiveness against mammography (+/− ultrasound) is an identified national priority [14].

Comparative effectiveness requires quantifying tradeoffs, a key dimension of which is cost, to determine the relative benefits and harms of alternative clinical practices for use by individuals and public policy makers [15]. In an era of increasingly constrained health care budgets, developing a clear understanding of costs weighed against clinical benefits is critical to rational resource use [15]. Fortunately, evidence of the clinical impacts of preoperative breast MRI is accumulating rapidly. Specifically, results from randomized controlled trials and observational studies are providing insights into preoperative MRI effects on cancer detection, biopsy rates, biopsy yield, surgical treatment, and recurrence [4‐6, 8, 16‐21]. To date, there have been no reports of the impact of diagnostic/preoperative breast MRI on aggregate economic considerations particularly compared to standard workup involving mammography with or without ultrasound. Although the cost of a breast MRI is about 10-times higher than a diagnostic mammogram [22], the overall costs of workups including breast MRI are not known and are relevant both for women and for payers.

The objective of this study was to examine per capita costs (PCC) of imaging and biopsy in the diagnostic/preoperative window among women enrolled in Medicare who did and did not receive MRI during their breast cancer diagnosis and preoperative workups. To address this objective, we used a woman-centered framework rather than societal, by measuring costs as the amount actually paid for services, rather than a nationally-standardized cost estimate. Thus, our aim was to estimate costs at the woman level to provide estimates when considering tradeoffs in the use of breast MRI. The overall goal is to inform the current practice and policy debates over use of diagnostic/preoperative MRI from an economic standpoint to help guide clinically appropriate use of this rapidly diffusing technology [4].

Among the 56,449 women in the study cohort, a breast imaging claim (mammogram, ultrasound or an MRI) for 53,653 women was found within the diagnostic/preoperative window. Of the 53,653, 20 % (N = 10,776) received an MRI. Among the women defined in the MRI group (N = 10,776), 1,713 women had an MRI in the diagnostic period (16 %) and 9,196 women had an MRI in the preoperative period (84 %). 133 women had an MRI in both periods (12 %).

The total unadjusted median diagnostic/preoperative costs were almost double for women with a diagnostic/preoperative MRI vs. without ($2,251 v. $1,152; Table 1). Higher total median costs were found for each increase in median income quartile, for urban versus rural, for each successive year of diagnosis, for lower staged tumors, smaller tumors, tumors with a positive nodal status, and for women whose initial therapy was breast conserving surgery (BCS) with radiation compared to those with mastectomy or BCS alone (Tables 1 and 2). Older age, increasing comorbidities, and having primary surgical treatment closer to diagnosis were associated with lower median costs (Tables 1 and 2). Similar trends were found among the diagnostic/preoperative breast MRI groups.

Among workups not including MRI, negligible deviations from the median imaging costs were observed by SEER registry. However, in the MRI group, wider positive imaging cost deviations were found in the SEER registries of San Francisco, Los Angeles, and New Jersey. Variation in the total median costs for SEER registries are presented geographically as four distinct maps divided by receipt of biopsy or breast imaging and separately by receipt of MRI (Fig. 1). The median biopsy costs were somewhat higher among the MRI group ($1,086 v. $862; MRI vs. no MRI), but there were markedly lower biopsy costs in the New Mexico and the Seattle SEER registries for both diagnostic/preoperative groups (MRI v. no MRI). Compared with other regions in the diagnostic/preoperative MRI group, the highest biopsy median costs were found in Atlanta and San Francisco. For those without diagnostic/preoperative MRI, the highest biopsy costs were found in Detroit and New Jersey.

The unadjusted and adjusted (patient and tumor characteristics) PCCs were similar overall and among the diagnostic/preoperative MRI groups (Table 3). The total adjusted cost difference between the MRI groups was $1,065. Partitioning total costs by imaging and biopsy costs, we found the difference was mainly attributable to costs of imaging ($928).

Of the 38 % women (N = 20,570) who had at least one breast event (MRI, Mammogram, Ultrasound or Biopsy) during the preoperative period, 72 % (14,834/20,570) of the women had an imaging event and 58 % (11,947/20,570) had a biopsy. Within the diagnostic/perioperative window, we found that in the diagnostic period, the cost difference between the MRI and no MRI groups was $185 ($159 for imaging and $31 for biopsy); in the preoperative period, the difference was $369 ($779 for imaging and $15 for biopsy; Table 3).

Our study is the first to report diagnostic/preoperative costs for breast cancer in the Medicare population, and to compare those costs between women with and without diagnostic/preoperative MRI in their clinical workup. We found that the total adjusted PCC for the diagnostic/preoperative period were $1,541 and were nearly double for women with diagnostic/preoperative MRI ($2,382) compared to without ($1,317).

Diagnostic/preoperative costs did not differ dramatically over the study period, but decreased with more comorbidities, higher stage, and larger tumors. Geographic variation in costs was greatest for biopsies, regardless of receipt of diagnostic/preoperative MRI, and was very low for imaging without diagnostic/preoperative MRI. Overall, we found that diagnostic/preoperative MRI was associated with higher PCCs by an estimated $1,065 after adjusting for patient and tumor characteristics. However, the difference in imaging costs ($928) between those with and without diagnostic/preoperative MRI was considerably higher than the difference in biopsy costs ($138). Cost differences between the MRI v. no MRI groups were more notable in the preoperative period compared to the diagnostic period, which seems largely attributable to the cost of MRI itself. The national payment amount for a bilateral MRI found in the physician fee schedule (ref: http://​www.​cms.​gov/​apps/​physician-fee-schedule/​search/​) ranged from $994 to $905 in the years 2007 to 2009; hence the imaging costs difference of $928 found across the years in this study appears to be due to the cost of an MRI.

Biopsy cost differences were minimal in both the diagnostic and preoperative periods. Thus, use of MRI in diagnostic and preoperative workup for breast cancer does not appear to lead to excess costs beyond the cost of the MRI exam. These differences between the diagnostic/preoperative MRI groups as well as the cost trends observed in the geographic, patient and tumor characteristics provide an economic perspective of the costs attributed to diagnostic/preoperative workup.

Our study is difficult to compare to prior literature, given the lack of cost estimates for breast imaging and biopsies in prior studies. However, estimating diagnostic/preoperative costs in breast cancer overall, and in relation to use of diagnostic/preoperative MRI is important for several reasons as cost measures: 1) provide critical inputs for comparative effectiveness research to guide clinical decision making; 2) facilitate consumer-centered public cost reporting; and 3) inform decisions about costs once a breast cancer diagnosis occurs which may influence women’s choices of initial treatment [3]. This study fills an important gap in our current understanding of the cost implications of diagnostic/preoperative breast MRI. That is, while not conclusive, most studies to date have not reported improved outcomes from diagnostic/preoperative breast MRI, particularly for reoperation and recurrence [5, 27‐29]. Without clinical benefit, the increased costs associated with diagnostic/preoperative MRI might be interpreted as harmful to women and health care systems. The results presented here provide cost estimates, which can be combined with measures needed to model harms from women’s perspectives and which will be evaluated in relation to any benefits, in comparative effectiveness models.

Public reporting of costs is an important component of women’s ability to be knowledgeable about the likely cost of breast cancer care and to make informed choices. The Agency for Healthcare Research and Quality (AHRQ) reported a general lack of consumer-centered cost reporting, which is a necessary component for women to choose high-value health care (when linked to quality) [14]. As Keselman et al. note, patients need data to inform their decisions to be effective health care consumers [30].

Relatedly, even though diagnostic/preoperative care is only a portion of the costs of cancer care [2, 31, 32], concern over costs once a diagnosis occurs may influence women’s choices of initial treatment [3]. Qualitative evidence has revealed financial concerns and monetary stress for women newly diagnosed with breast cancer [3], which suggests that cost may play a role in planning initial treatment and breast cancer management.

Although this cross-sectional study was large, population-based, and used publicly available data, several limitations should be noted. First, we did not account for type of biopsy, such as core versus excisional, which is likely to be associated with variation in cost. Separate studies are currently underway to fully characterize detailed aspects of diagnostic/preoperative biopsy use in women with and without diagnostic/preoperative MRI. Furthermore, because claims data may be inexact with respect to date of service and date of claim, we performed a sensitivity analysis of our definition of diagnostic/preoperative window to determine whether biopsies performed to establish the diagnosis may be classified as diagnostic/preoperative. Our analyses were robust to the assessment of the ‘start’ of the diagnostic/preoperative window based on the putative diagnostic biopsy. Also, as in all Medicare claims-based studies, the potential for unmeasured confounding exists, and our findings are only generalizable to the Medicare population. Patterns of use and costs may be different for women younger than the >65 yrs. old included in this study. Another limitation of this claims-based analysis is the lack of information on women’s preferences and clinical decision making related to breast cancer care. Finally, although not a limitation, we do note that our cost measure approach (per capita based on amount paid) is one of several methods to evaluate cost. Another common approach is to use a standardized national estimate for each type of service and apply those costs to the services identified in claims. The per capita amounts paid costs measures used in this study are likely to be more useful at the individual woman level for informing clinical decisions. The standardized national estimate approach may be better used if one seeks to compare health care systems, do small area analysis, or inform the societal perspective. Both of these two methods are acceptable, but are different “lenses” through which to interpret costs.

Since we found differences in total costs between the MRI groups, the results suggest the excess costs associated with diagnostic/preoperative MRI could be factored into women’s decisions as a potential harm (with the marginal increase estimated at about $1,065). These estimates may prove useful for those engaged in comparative effectiveness studies of breast cancer diagnosis and for women and their healthcare providers as they consider alternative care pathways.