Price transparency & out-of-pocket payments for medications: Implications of associated delivery fees in the United States.

1
Introduction
Policymakers in the United States are increasingly relying on price transparency as a tool to control health care spending by encouraging patients and providers to consider prices when choosing among treatment options [1]. Consumers often information on the prices of potential treatments. Price transparency tools are intended to address that gap by making both gross (i.e. what insurance pays for services on a beneficiary’s behalf) and out-of-pocket (OOP) payments (i.e. what a patient pays for the drug) available to consumers and providers. The premise behind price transparency is that a better understanding of prices and required payments will allow consumers (i.e. patients) to “shop” for services and incorporate information on prices in their decision-making.
Prescription drugs have been a target of price transparency efforts. In May 2018, the Trump Administration released a “blueprint” to address high drug prices. Greater price transparency for patients enrolled in Medicare Part D, publicly funded health insurance coverage for adults 65 and over and people with disabilities, was one of several mechanisms to lower these prices [2], [3]. In 2019, the Center for Medicare and Medicaid Service (CMS), the government agency that runs the U.S. Medicare program, finalized regulations requiring Part D plans to provide real-time benefit tools that provide customized information on OOP costs for prescription drugs and clinically appropriate treatment alternatives [4]. Efforts exist to extend price transparency requirements to employer-sponsored coverage [5]. On February 5, 2025, President Trump issues an executive order indicating his intent to revive and extend these efforts.
Prostate cancer represents an important context for understanding the implications of these regulations. The costs attributable to the treatment of metastatic prostate cancer care in the United States were estimated at between $5.2 to $8.2 billion per year annually between 2007–2017 and were expected to rise [6]. These high OOP payments can result in significant patient harm including altered medication use, lower spending on food and other essential goods and services, and psychological stress, and can contribute to worse health outcomes and higher mortality [7], [8], [9]. Similar to other malignancies, there are more than six treatment options which improve long-term survival for patients with metastatic castrate-resistant prostate cancer (mCRPC) relative to no treatment. Little evidence exists on comparative effectiveness [10], [11], as few drugs have been tested in head-to-head comparisons. The development pipeline includes additional treatment options and combinations which are also associated with high prices.
A distinguishing feature of mCRPC is that treatment alternatives include both oral therapies, which are generally covered under prescription drug benefits, and injectable therapies, which are generally covered under medical benefits. Injectable therapies, which are administered in a physician’s office usually require additional payments for services such as infusion center fees, clinic visits, and lab monitoring fees, etc. that often occur on the same day as drug administration. While prior work has evaluated the OOP payments required for the drugs themselves [12], there is little evidence of how payments for drug administration impact total OOP payments for these therapies. In the absence of this information, the provisions of price transparency laws, which require reporting of payments for the drug alone, may lead to misleading comparisons of the relative costs of alternative treatment options.
In this study, we use insurance claims for individuals with employer-sponsored health insurance (ESHI) to examine OOP payments for six treatments for mCRPC, differentiating between payments for only the drug (drug only) and payments for the drug as well as payments for drug administration that occur on the same day of drug receipt, which we refer to as “full day” payments.. We also evaluate differences in full day and drug-only OOP payments for each therapy by health plan type.

2
Methods
2.1
Data and study Population
We used the IBM Marketscan Commercial and Medicare Supplemental databases (Marketscan) to identify male patients who initiated treatment with one of six focus drugs (docetaxel, abiraterone, enzalutamide, sipuleucel-T, cabazitaxel, and radium-223) used to treat mCRPC from 07/01/2013–06/30/2019. Docetaxel and cabazitaxel are drugs given by injection, abiraterone and enzalutamide are dosed orally and sipuleucel-T is a vaccine..

2.2
Outcomes
Our outcome of interest was total OOP payments on day of drug receipt (i.e. drug + delivery payments on the day of drug receipt), which we refer to as “full day” OOP payments,for first-line treatments within 6-months of the index date. We also examine OOP payments for drugs only to understand the extent to which drugs as opposed to other types of services contribute to patient OOP spending.
We defined payments for drugs alone based on the service claims with the Healthcare Common Procedure Coding System (HCPCS) codes or outpatient prescription claims with the National Drug Codes. To incorporate payments for drug administration, such as infusion center fees, laboratory tests, and clinic visits, we included all additional claims incurred on the same day of treatment receipt. For oral drugs, the treatment date is the date of the pharmacy fill. OOP payments include the deductible, coinsurance, and co-payment amounts for each paid claim. The deductible, coinsurance and co-payment contribute to an individual’s OOP payment. The deductible is the amount of money an individual/family must pay out of one’s own pocket annually before insurance starts to contribute. Cost sharing may also take the form of a copayment, a fixed fee the patient pays for each prescription or coinsurance, a percentage of the bill that the patient pays.

2.3
Exposures and additional variables
The primary exposure was the type of drug the patient received as first-line therapy.
Our secondary exposure was the type of insurance plan in which a patient was enrolled. We defined health insurance type using Marketscan’s health plan categories: high-deductible (HDHP) or consumer-driven health plan (CDHP), comprehensive, health maintenance organization (HMO) or exclusive provider organization (EPO), and preferred provider organization (PPO) or point-of-service (POS) health plan.
We used six-month pre-index claims to identify co-morbidities and index date claims to identify demographic characteristics.

2.4
Statistical methods
We first report descriptive statistics on the study sample’s characteristics by first-line treatment. We then report descriptive statistics of full day 6-month OOP drug payments and drug-only 6-month OOP payments by first-line treatment, overall and by plan type. All reported payments were after excluding outliers (<1st percentile or > 99th percentile). For each variable, we also calculated the proportion of people making any payment, overall and by plan type. We tested for differences in mean costs across treatments using ANOVA F-tests.
We then employed multivariable models to test for differences in OOP payments after adjusting for patient characteristics. Given the large share of observations with zero OOP payments, we utilized two-part models. First, we used logistic regression to model the probability of having a non-zero OOP payment. Second, we used a multivariable generalized linear model (GLM) with a gamma-distribution and log-link to examine OOP payments amongst patients with a non-zero OOP payment. We calculated regression-adjusted mean costs using a product of expectations from both parts of the model and used bootstrapping to generate 95% confidence intervals (CIs). The regression-adjusted mean costs were compared between treatments using the seemingly unrelated estimation in STATA which allows cross-model tests by combining the estimation results from the first and second parts of the two-part model [13]. We completed this analysis for both the comprehensive OOP payments and OOP payments for drug alone. When performing our regression analyses, we were not able to include radium-223 and cabazitaxel, as the small sample size for these treatment options resulted in unstable estimates.
Next, to examine the relationship between full day OOP payments and insurance type, we performed the above analyses with an interaction between therapy and insurance type. Similar to the model without the interaction term, the unadjusted and regression-adjusted mean costs were compared between treatments within each insurance type.
All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC) and Stata (Release 17. College Station, TX: StataCorp LLC). This study was deemed exempt from Institutional Review Board (IRB) review.

3
Results
We identified 4,298 patients who met inclusion criteria. Table 1 presents patient characteristics by treatment type. Patients who received docetaxel were more generally younger while those receiving the other treatments were more evenly distributed across age groups. Sipuleucel-T was less likely to be administered in later calendar years. Most patients were in urban settings, with the lowest share of patients receiving treatment in the Western United States. Over 50% of patients across treatments had point of service (POS) or preferred provider organization (PPO) health plan types. Patients using cabazitaxel had the lowest comorbidity index (0.4, SD 1.0) while patients using enzalutamide (1.4, SD 1.7) and radium-33 (1.8, SD 2.0) had the highest. Additional information on patient characteristics by drug type, including the prevalence of specific comorbidities, is available in supplemental Table 1.
3.1
Full day 6-month OOP payments by first-line therapy
Unadjusted 6-month mean full day OOP payments varied by drug; they were highest amongst patients who used cabazitaxel ($1380, SD $1534), followed by abiraterone ($842, SD $1561), and lowest among those who used Sipuleucel-T ($658, SD $1328) (Fig. 1 & Table 2). The contribution of payments for the drug to full day OOP payments varied across drugs. Docetaxel accounted for 34% of the full day payment compared to 93% for Enzalutamide (Fig. 1).
Regression-adjusted 6-month full day drug OOP payments were similar to the unadjusted estimates (Table 2). Supplemental Table 2 presents the results of the regressions we used to calculate the adjusted estimates. While the unadjusted estimates are statistically significantly different, the regression adjusted differences across the drugs are not (p = 0.38). We note that we are able to compare the adjusted estimates only across the four drugs with adequate sample sizes.
The probability of a patient having zero full day drug OOP payments varied by drug. In unadjusted analyses, patients who received Sipuleucel-T were more likely to have zero OOP payments (36.9%), when compared to those who received Docetaxel (13.7%), Enzalutamide (11.0%) or Abiraterone (6.6%) (Table 2). In regression-adjusted analyses (Supplemental Table 2), when compared to Docetaxel, Sipuleucel-T was associated with a 78% lower odds (OR 0.22, 95% CI 0.16 to 0.29, p < 0.001) and Abiraterone with a 123% higher odds of having a non-zero OOP full day drug payment (OR 2.23, 95% CI 1.66 to 2.99, p < 0.001).

3.2
Differences by plan type in full day 6-month OOP payments by first-line therapy
For most drugs, full day OOP payments were highest for patients enrolled in CDHPs and lowest for those in comprehensive plans (Fig. 2). The exception is enzalutamide, for which payments were highest for patients in EPOs or HMOs. The extent to which full day OOP payments vary by first-line therapy differed across plan types (Fig. 2, Supplemental Table 3). For patients in Comprehensive, CDHP/HDHP, or POS/PPO plans, unadjusted and regression-adjusted mean full day drug payments were not statistically different across treatments. There was greater variation in full day drug payments across treatments amongst patients in EPO/HMO plans (p < 0.001 for both unadjusted and adjusted estimates), with full day payments highest for patients using enzalutamide and lowest for those using docetaxel (Fig. 2, Supplemental Table 3).
While the percentage of the full day payment that is attributable to the drug alone varies across plan type, the share is lowest for docetaxel for each plan type (Fig. 2). Less variation exists across drugs within each plan type when considering full day rather than drug-only OOP payments. For drug-only payments, differences by first-line therapy within each plan type are statistically significant in both adjusted and unadjusted analyses (Supplemental Table 3 and Supplemental Table 4).

4
Discussion
Our results point to the importance of considering payments both for drugs and drug administration fees when comparing OOP payments for injectable and non-injectable drugs. We find relatively little difference in OOP payments across first-line treatments when accounting for OOP payments associated with drug administration, particularly for the four most frequently prescribed drugs (docetaxel, abiraterone, enzalutamide and sipuleucel-T) during the time period we examine. While there is variation in the proportion of payments for drug alone relative to full day payments across treatments, overall, regression-adjusted mean full day OOP payments are not statistically different across treatments for mCRPC. Consistent with our prior work [13], we also document that, when limiting OOP payments to those for the first-line drug alone, significant differences exist across these four drugs. OOP payments of injection therapies alone are significantly lower than those of oral therapies for the treatment of mCRPC. A primary driver of the similarities in full day OOP payments across drugs is that, while Docetaxel has far lower OOP payments for the drug alone, administration of the drug requires other services, such as clinic visits, infusion center fees, and laboratory tests, which can be associated with significant payments.
Our results also suggest that average payments may mask important differences across patients using the same drug. While OOP payments for docetaxel were similar to those for other drugs when considering fees for drug administration, interestingly, docetaxel was still associated with a higher likelihood of having a zero OOP than abiraterone, and a lower likelihood of having a zero OOP than Sipuleucel-T. So, while, on average, docetaxel may have similar full day OOP payments, for some patients, the full day OOP payment may still be lower for docetaxel compared to other alternative treatments.
Our evaluation further demonstrates that when examining full day OOP payments across treatments, the share of full day payment attributable to the drug alone varies across health plan. This is important because, in theory, if a clinician has drug only payment estimates, he/she may think that he/she can estimate full day payments by adding a certain percent onto the drug payment estimate, alone. However, this strategy will lead to inaccurate information and could contribute to decision-making that is not in line with a patient’s values/goals.
Our evaluation has several limitations. First, we may overestimate OOP full day payments across drugs as we are unable to account for coupons and use of drug-assistance programs. We may also underestimate payments associated with drug administration as we are unable to account for other services that may be required for drug administration and/or drug receipt but that do not occur on the day of drug receipt. For both oral and infusion drugs, many of the non-drug components and ancillary costs (i.e. lab fees) may be outside of the one-day window, but this may be especially true for oral therapies because we use the pharmacy fill date as the treatment date. However, basic lab fees required for regular monitoring are fairly nominal and frequently occur less frequently than labs and clinic visits for routine infusion treatments. Similarly, the one-day window may not capture other medications required for infusion drug delivery in certain patient populations, such as growth factors. We also do not include the indirect costs of treatment, such as time off from work and other indirect costs which may be substantial. Second, while we are trying to capture data for patients who received first-line treatment for mCRPC, some patients may have received these treatments when in the non-metastatic or non-castrate resistant state. However, prior work has used comparable criteria when evaluating treatments for patients with mCRPC, which sets a precedent for our methods [14]. Third, we may underestimate the OOP payments required to receive these medications as we only include OOP payments for individuals who received treatment and do not include individuals who may not have received treatments given the high OOP payments. Fourth, payments could be underestimated as we estimate payments for each drug across six months of therapy and not all patients completed a full six-month treatment course. However, duration of therapy is similar across treatments, so estimated payments in this evaluation would likely underestimate monthly payments similarly across treatments, as our estimates are calculated across 6 months of treatment.
These limitations notwithstanding, our findings have important implications for patients, clinicians, and policymakers. For patients, it is critical to understand that even if he/she knows the payments required for the drug alone, these payments likely only represent a portion of the payments that will be required to receive the medication. It is critical to discuss other payments that may be incurred such as infusion center fees, how often one has to return for clinic visits and lab monitoring, and to consider other indirect costs associated with treatment.
For clinicians, even if they have accurate OOP payment information for drugs, this data may not be sufficient to help guide decision-making. Payments for other services that are required for drug administration may be substantial. Payments for these associated services may also vary by insurance type, making estimation and payment discussions even more challenging. It is therefore critical for clinicians and/or their clinic staff to discuss other direct (i.e. payments for the drug, infusion center, labs) and indirect (i.e. follow-up and time off from work, time traveling to appts) payments that may be required for drug delivery.
For policymakers, while price transparency is important to help guide decision-making, price transparency for drugs alone is not enough. In fact, patients may be misguided by using price information for drugs alone. Payment information for other services required for drug administration should be incorporated into price transparency estimates. Furthermore, quality measures should include the presence of cost discussions that focus on both direct and indirect treatment costs; follow-up and monitoring requirements are critical for these discussions.

5
Conclusion
Because payments for drug alone make up only a portion of payments required for drug delivery/receipt, price transparency tools which focus on prices for drugs alone is not enough to help guide decision-making. Furthermore, this proportion may vary by both drug and insurance type so using simple rules of thumb to estimate full day OOP payments for drug delivery will not work given different across and within plan type in the relationship between drug and drug delivery payments. To use price information to help guide decision-making, it will be critical to understand the comprehensive payments (i.e. direct costs) required for drug administration. Future research should include payments associated with treatment not on the same day and evaluate variation in indirect costs of different treatments.

CRediT authorship contribution statement
Deborah R. Kaye: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Hui-Jie Lee: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation. Daniel J. George: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Charles D. Scales: Writing – review & editing, Writing – original draft, Resources, Methodology, Data curation. M.Kate Bundorf: Writing – review & editing, Writing – original draft, Visualization, Supervision, Methodology, Formal analysis, Data curation, Conceptualization.

Ethics approval and consent to participate
This study was deemed exempt from IRB review.

Funding
This work was supported in part by the 2021 Urology Care Foundation (UCF) Research Scholar Award Program, the Society of Urologic Oncology (SUO), the National Cancer Institute of the National Institutes of Health (NIH) (1-K08CA267062-01A1), and National Center For Advancing Translational Sciences of the NIH under Award Number UL1TR002553. The content is solely the responsibility of the author and does not necessarily represent the official views of the UCF, SUO, and/or NIH.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.