Feasibility of Multi-Institutional Data Collection for Medication Prior Authorization in Pediatric Oncology

Abstract

Purpose

Medication prior authorization (mPA) occurs commonly in US cancer care. While medical oncologists report potential harm resulting from this requirement, the impact of mPA in pediatric oncology is unknown. This study’s primary aim was to test the feasibility of prospectively collecting multi-institutional data on mPA in pediatric oncology, while secondarily assessing for resultant delays in care.

Methods

Pediatric patients with cancer enrolled between September 2021 and December 2022 at three Children’s Oncology Group (COG) institutions participating in the National Cancer Institute Community Oncology Research Program (NCORP). Data collected for each mPA event included the name of the medication, indication, desired initiation date and actual date administered.

Results

Among 68 patients enrolled at three institutions, 38 (56%) were subject to at least one mPA. A total of 69 mPAs occurred in these 38 patients with 36 (52%) for supportive care and 33 (48%) for treatment medications. Ultimately, 61/69 (88%) mPAs were approved as prescribed (33/33 treatment mPAs) with a range of 5–240 minutes of provider/staff time required to resolve. MPAs delayed care in 15/69 (22%) of cases with a range of 1–21 days. Most providers reported minimal or no additional burden to characterizing mPA data.

Conclusion

Despite the complexity and variability of institutional mPA processes, it is feasible to prospectively collect multi-institutional mPA data. No cancer treatment mPA request led to alterations in prescriptions. Delays in care due to mPA affect roughly one quarter of pediatric patients with cancer, the consequences of which remain unknown.

Introduction

The prior authorization (PA) process is a policy utilized by the US health payer system for regulation and cost containment usually targeting prescription drugs, durable medical equipment, and diagnostic radiology. A Kaiser Family Foundation review of 2022 data showed that nearly all (99%) Medicare Advantage enrollees are in plans that require PA for some services, particular those of higher cost.¹ From the latest 2023 report from the Council for Affordable Quality Healthcare^® (CAQH), prior authorization volumes and costs continue to rise and are estimated to have reached over 700 million US dollars annually.² In a 2023 survey of over 1,000 US physicians conducted by the American Medical Association, 94% of responders reported care delays, 78% reported treatment abandonment, and 24% reported a serious adverse event for a patient in their care because of mPA.³ These results stand in contradiction to a 2020 survey conducted by the America’s Health Insurance Plans (AHIP) that reported mPA having a positive impact to quality of care (91%), affordability (91%) and patient safety (84%) with an estimated <10% of enrollees being subject to mPA.⁴ Barriers identified by AHIP included providers not utilizing electronic health records enabled for electronic prior authorization and providers not submitting necessary clinical information with the initial request.

The American Society of Clinical Oncology (ASCO) concluded in a 2022 position statement that PA is the largest barrier to care for insured patients for adult cancer patients.⁵ An included member survey found that 42% of responders reported significant delays in care and 36% ascribed loss of life from mPA. To date, the available analyses have been focused adults with cancer, and minimal information exists in children with cancer. A previous single institutional study focused on pediatric oncology conducted in 2014–2015 found mPAs occur in roughly 60% of patients and took an average of 45 minutes to adjudicate, while only 1.5% mPA led to a change in therapy.⁶ One previous survey in 2016 of the Children’s Oncology Group (COG) affiliates participating in the NCI Community Oncology Research Program (NCORP) identified mPA as a barrier to care by 60% of pediatric oncologists.⁷ There is currently no explanation for the discrepant perspectives of payers and providers, nor is there an understanding of the impact that mPA has on access to care. To date, pediatric and adult data have been largely retrospective, survey-based (not case-based), and contradictory (provider vs. insurer perspectives). We aimed to address these gaps by designing a prospective, multi-institutional study. Specifically, we characterized the occurrence of mPA and prospectively collected data on medications subject to mPA as well as resultant treatment delays. Patient and family demographic characteristics were collected to identify potential disparities in mPA impact across different segments of the population.

Methods

A prospective cohort study was opened and conducted in collaboration with three COG sites: Helen DeVos Children’s Hospital (HDVCH) in Grand Rapids, MI, Blank Children’s Hospital in Des Moines, Iowa, and BI-LO Charities Children’s Cancer Center in Greenville, SC. Sites were selected based on institutional interest, available clinical research resources, and geographic distribution. The participating institutions were all COG NCORP sites but received no direct funding support for this work. Medically insured, English-speaking patients less than 26 years of age with newly diagnosed or relapsed cancer were enrolled within 30 days of starting chemotherapy between September 2021 and December 2022. Enrolled patients were monitored for the occurrence of mPA during the subsequent 90-day follow-up period. This time frame was chosen due to the feasibility nature of the study. Site research personnel coordinated with their respective institutional administrative groups responsible for managing mPA so that the site clinical research associate (CRA) would be notified with each mPA occurrence. In addition, site CRAs actively queried their respective system every 30-days to ensure all occurrences were recorded.

Study procedures were tailored by site according to predetermined institutional workflows. At HDVCH, the study team met with their institution’s centralized PA office to inform them of the study and what data were being collected, as well as identify points of contact. Those contacts were notified with the names of enrolled patients and instructed to contact the study CRA if a mPA occurred for that patient. Weekly reminders of enrolled patients and data required were sent to the contacts at the centralized mPA office. Additionally, divisional nurses responsible for addressing mPA were similarly trained. At Blank Children’s, the divisional pharmacist processed most retail and specialty prescriptions which were routed via trained nurses, nurse coordinators and care managers. Referrals for mPA requests were made almost exclusively utilizing the third-party platform, CoverMyMeds^®. This enterprise system generates a mPA “key” which is entered into the electronic health record in an event note (Epic Systems). At BI-LO Charities Children’s Cancer Center, one patient access specialist who regularly received all notifications of mPA requests notified the study team member who then was responsible for data collection and entry.

For every mPA, the data were entered into REDCap^® including the medication name, indication, type of insurance, intended start date and actual start date. Intended start dates were determined by the treating provider. Delays were defined as any separation between intended and actual start dates that were deemed at least possibly due to the mPA request. The dates and attribution were assigned by the treating provider. Sites were asked to record the time spent on mPA adjudication and ultimately, whether the drug was approved, disapproved and/or changed because of the mPA. For every mPA event, the primary medical provider was given a six-question survey (Data Supplement) to assess the medication’s necessity, as defined by Unguru et al⁸, categorization (treatment or supportive care), and attribution of delays related to the mPA request. The Unguru reference was utilized as it remains the only medication-focused resource for defining medical necessity in pediatric oncology. Finally, the provider was asked to rate the burden of completing the requested form. Feasibility was defined by achieving at least 75% provider forms completed for all enrolled patients with no missing fields and by the data collection to be classified by at least 60% of providers as no more than minimally burdensome. Summary statistics were calculated using the name of the medication, the payer, the total clinical staff time spent on each mPA request, and the payer decision outcome of PA request (approval vs. denial). Period prevalence was used to represent the relative frequency of mPA requests and was calculated by dividing the number of requests in the study time frame by the total number of enrolled patients. This investigation was reviewed and approved by the UnityPoint Health Institutional Review Board. Written informed consent was obtained from all patients or legal guardian prior to their participation in this study.

Results

With regards to the primary aim of feasibility, all provider forms (100%) associated with mPAs were complete with all providers reporting minimal (6%) or no (94%) additional burden to the data collection. The characteristics of the 68 patients enrolled at three institutions are shown in Table 1. A little over half (56%) of all patients were enrolled at HDVCH, 34% at Blank, and 10% at BI-LO. The largest portion of patients were diagnosed with acute lymphoblastic leukemia (44%), identified as white/non-Hispanic (87%), and had private health insurance (65%). The median patient age was 11 years with an interquartile range of 4–15.8 years. This study was not statistically powered to analyze mPA variation by treatment site or demographic characteristics. However, trends in mPA requirements were noted by geography (HDVCH 39%, Blank 72%, BI-LO 86%). No trends in mPA requirements were noted by type of insurance (public 50%, private 59%). The number of non-white or Latino patients enrolled in this study were too small to assess for possible associations.

TABLE 1.

Patient Characteristics

	Patients	≥ 1 mPA
	N = 68, No. (%)	YES; N = 38, No. (%)	NO; N = 30, No. (%)
Gender
Female	37 (54)	21 (57)	16 (43)
Male	31 (46)	17 (55)	14 (45)
Median Age (Interquartile Range)	11 years (4–15.8)	11 years (4–15.3)	12 years (4–16.5)
Treatment Site
Michigan	38 (56)	14 (39)	22 (61)
Iowa	23 (34)	18 (72)	7 (28)
South Carolina	7 (10)	6 (86)	1 (14)
Diagnosis
Acute Lymphoblastic Leukemia	30 (44.1)	18 (60)	12 (40)
Hodgkin Lymphoma	9 (13.2)	6 (67)	3 (33)
Non-Hodgkin Lymphoma	5 (7.4)	0 (0)	5 (100)
Osteosarcoma	4 (5.9)	3 (75)	1 (25)
Brain Tumor	4 (5.9)	1 (25)	3 (75)
Soft Tissue Sarcoma	4 (5.9)	3 (75)	1 (25)
Ewing Sarcoma	3 (4.4)	2 (67)	1 (33)
Renal Tumor	3 (4.4)	1 (33)	2 (67)
Germ Cell Tumor	2 (2.9)	0 (0)	2 (100)
Ovarian Carcinoma	1 (1.5)	1 (100)	0 (0)
Acute Myelogenous Leukemia	1 (1.5)	1 (100)	0 (0)
Hepatoblastoma	1 (1.5)	1 (100)	0 (0)
Neuroblastoma	1 (1.5)	1 (100)	0 (0)
Race/Ethnicity
White, non-Hispanic	59 (86.9)	34 (58)	25 (42)
Black	4 (5.8)	1 (25)	3 (75)
White, Hispanic	2 (2.9)	1 (50)	1 (50)
Asian	2 (2.9)	0 (0)	2 (100)
More than one	1 (1.5)	1 (100)	0 (0)
Type of Insurance
Public	24 (35)	12 (50)	12 (50)
Private	44 (65)	26 (59)	18 (41)

mPA = medication prior authorization

Of the total 68 patients enrolled, 38 patients were subject to at least one mPA during the 90-day study period, yielding a period prevalence of 56%. A total of 69 mPAs occurred in these 38 patients with 36 (52%) for supportive care medications and 33 (48%) for treatment medications (Table 2). Five of 36 (14%) supportive care mPAs and 27 of 33 (82%) treatment mPAs occurred for drugs considered essential. mPA delayed care in 15 of 69 (22%) of cases with a median of four days (range 1–21 days); four delays were for treatment medications and 11 for supportive care meds. Delays were rated probably or definitely due to the mPA in 13 of 15 cases and possibly due to the mPA in two of 15 cases. Providers reported feeling very confident in their attribution of delay to the mPA in 13 of 15 delays and somewhat confidant in two of 15 determinations. Ultimately, 61 of 69 mPAs were approved with a median of 15 minutes (range 5–240 minutes) of care team time per mPA. The remaining eight mPAs led to a change in prescription, all for supportive care medications (Figure 1). With regards to potential negative medical outcomes, there was one reported study violation induced when a patient prescribed asparaginase Erwinia chrysanthemi was unable to be received within the time frame suggested by standards of care.

TABLE 2.

Medication Subject to medication prior authorization

Supportive care; N = 36 (52%)		Treatment; N = 33 (48%)
Essential; N = 5 (14%)	Non-essential; N = 31 (86%)	Essential; N = 27 (82%)	Non-essential; N = 6 (18%)
ondansteron x 4	pegfilgrastim x 9	cytarabine x 7	brentuximab vedotin x 3
morphine	filgrastim x 3	mercaptopurine x 4	tisagenlecleucel
	omeprazole x 2	asparaginase x 4	blinatumumab
	famotidine	vincristine x 3	selumetinib
	posaconazole x 4	methotrexate x 3
	oral vancomyocin x 4	imatinib x 2
	IVIG x 2	temozolomide x 2
	micafungin	daunorubicin
	foasaprepitant	cyclophosphamide
	methylnaltrexone x 2
	amlodopine
	bicitra

FIG 1:

Outcomes of medication prior authorization requests.

Discussion

In this study, we demonstrate the feasibility of prospective, multi-institutional data collection on mPA in pediatric oncology. Over half of enrolled patients were subject to at least one mPA in a 90-day window, most of which were considered essential treatments for cancer and ultimately approved. Importantly, delays in care related to mPA was documented in roughly 1 in 4 patients which have unknown consequences with regards to outcome. We documented various workarounds that may have prevented delays and therefore our findings may underestimate the burden. These included filing mPAs well in advance (when possible) of need, providing bridge funding (out of pocket or philanthropy) to cover the cost of the medication, and delivering medication as an inpatient. While these efforts may have prevented delays, they add a currently uncalculated financial and psychological burden on patient families and treating institutions. As each institution had distinct processes for mPA identification and management, we attempted to informally assess best practices. Both decentralized and centralized methods yielded equally effective data capture. For future studies, either model would be acceptable and difficult to standardize given variation in institutional resources. One trend noted was that dedicated resources to mPA management led to more effective enrollment and data capture.

The impact of delayed treatment and supportive care medications (immune stimulants, antibiotics, laxatives, antacids, pain relievers, antiemetics, etc) is difficult to measure and would require further study. The definition of essential medicines in pediatric oncology is dynamic and somewhat elusive and we therefore utilized the only collated resource.⁸ However, this list was not designed to be complete or kept up to date beyond the 2019 publication date. For example, selumetinib for plexiform neurofibroma and brentuximab vedotin for Hodgkin Lymphoma were not included as “essential” but have since been FDA-approved for those respective indications after publication of our referenced document.⁸ Supportive care medications such as growth factors and anti-infectives (oral vancomycin for clostridium difficile) are considered standards of care for their respective indications, and are also not included in our utilized reference. Despite these limitations, we believe that providing insights into the prospective, multi-institutional mPA data collection is an important step in the process of enhancing efficiency and improving access in pediatric oncology. Until more effective systematic solutions are created, pediatricians and subspecialists have an online mechanism created through the American Academy of Pediatrics’ Payer Advocacy Committee to submit individual patient concerns (https://form.jotform.com/Subspecialty/aapcodinghotline). Though intended for specific cases, there are examples of single cases leading to broader payer policy change.

According to the latest data from the Centers for Medicare & Medicaid Services, US healthcare expenditures equaled 17.3% of the US gross domestic product in 2022 and continues to grow at rates that have economists concerned about future sustainability. ⁹ Eliminating wasteful spending represents the least controvertible method to reduce the burden on patients, the healthcare system, and the US economy. In 2019, an estimated $265 billion (~7% of health care expenditures) was spent on administrative complexity.¹⁰ While wasteful spending in healthcare has important economic consequences, the impact on patient care remains central. Previous assessments of the effect of mPA on cancer care have been limited to general surveys with contradictory perspectives arising from providers and payors. The primary aim of this study was to test the feasibility of prospective, multi-institutional data collection for mPA in pediatric oncology, their characterization, and association with delays in care. We found this to be feasible and given the documented cost and potential negative cancer care impact of mPA, modifications to the mPA process are justifiable.

If the general principle of mPA is to ensure the right patient gets the right drug at the right time, it is essential that providers and payors use the correct resources and the best available evidence to make medical necessity determinations. Currently, this is accomplished by cross-referencing existing, evidence-based compendia and resources that rely on FDA approval and/or large randomized controlled trials. When considering diseases which affect large numbers of patients, this approach is entirely achievable and justifiable. However for patients with rare diseases like pediatric cancer, the pathway to developing standards of care differs by necessity. Often clinical trials conducted through large multi-institutional research consortia studies establish new standards of care, however, these entities are not structured nor recognized as clinical care benchmarks. Adding FDA-labelled indications is not routinely pursued when these clinical trials establish efficacy in pediatric cancer trials. Therefore, there remains discord between what pediatric oncologists and payors utilize to determine medical necessity. This is highlighted by the fact that mPAs in our study were requested for treatments that have been considered standard of care for decades including cytarabine, mercaptopurine, asparaginase, vincristine, methotrexate for acute lymphoblastic leukemia. Recently the National Comprehensive Cancer Network (NCCN) has incorporated pediatric-specific guidelines for several pediatric cancers, guidelines that serve as a reference point for coverage discussions. Embedding NCCN-endorsed standards, which are updated frequently, into the Code of Federal Regulations and state Medicaid policies has the potential to reduce the burden of mPAs in pediatric oncology.

A 2018 joint effort between the American Medical Association, the American Hospital Association, AHIP, the American Public Health Association, BlueCross BlueShield Association, and the Medical Group Management Association generated a consensus statement on improving the PA process.¹¹ Five mutually agreed upon principles areas were identified as having the highest impact; selective implementation based upon providers’ performance, regular program review, policy transparency with efficient communication, continuity of patient care, and systems automation. Much of what was agreed upon appears in a Congressional bill, unanimously passed in 2022 by the US House of Representatives (Improving Seniors’ Timely Access to Care Act; H.R.3173). Under this legislation, Medicare Advantage plans would be required to establish an electronic PA program that provides real-time decisions in response to requests for items and services that are routinely approved in addition to other stipulations. Additionally, nine states have recently passed legislation that reforms the PA process in their jurisdiction, with more proposals introduced in 2024. Lastly, in 2024 the Centers for Medicare & Medicaid Services (CMS) finalized the Interoperability and Prior Authorization Final Rule, which set requirements for timeframes, denial reasons, metrics and use of electronic measures.¹² While this represents an enormous advance in system-wide improvements, the applicability for pediatric cancer care could be negligible given the extremely low proportion of pediatric patients who receive Medicare benefits. To ensure pediatric patients with cancer and their healthcare providers similarly benefit, private policy and Medicaid changes are also required which will require the assistance of disease specific and/or age specific advocates, so that children do not get left behind on these important changes.

Several observations and limitations of this pilot study are worth noting and may inform further study and/or policy decisions. Though our period prevalence of 56% closely matched previous pilot work in this area, the current study period only included a 90-day window of treatment that excluded up to the first month of treatment. With current treatments for some pediatric cancer lasting over two years, it is possible that our results underestimate the prevalence throughout the entire treatment period. The total subject enrollment across the three sites over a 16-month period was relatively low. Our study was conducted during the COVID-19 pandemic, a time which saw a dramatic drop in cancer trial screening and enrollment.¹³ In addition, one of our three sites experienced significant clinical staff turnover, another known barrier to clinical trial enrollment.¹⁴ While not statistically powered to analyze mPA variation by treatment site or demographic characteristics, there appeared to be variation in mPA prevalence according to site and no statistical association between mPA occurrence and race/ethnicity or public vs. private insurance. Further studies are required to assess potential differences between site mPA prevalence, but potentially could be linked to variation in payer policy. Non-English-speaking patients were excluded from this study and therefore we cannot assess whether language barriers independently impact mPA frequency or delays.

In summary, our study found that prospective, multiple institutional data collection on mPA in pediatric oncology is feasible and the magnitude of occurrence and impact on care is substantial. Our results suggest that all mPA for treatment medications are ultimately approved with considerable resourcing and delays in care for almost a quarter of patients subject to mPA. Current proposed legislative solutions have the potential to improve care delivery for adults with cancer but may have no impact on children unless advocacy efforts are undertaken to ensure Medicaid and private payer policy follow Medicare policy changes.

Supplementary Material

Data Supplement: Provider survey assigned to each medication prior authorization.

Context Summary.

Key Objective

Is it possible to collect prospective, multi-institutional data on medication prior authorization (mPA) in pediatric oncology.

Knowledge Generated

Prospective, multi-institutional mPA data collection is feasible. We have confirmed that over half of pediatric oncology patients are subject to at least one mPA. Every cancer treatment prescription was approved with a median time of 15 minutes spent on resolution. Delays in care attributed to mPAs occurred in 22% of cases.

Relevance

mPA in pediatric oncology when studied prospectively rarely leads to changes in treatment, consumes resources and in some cases, delays care. Revisions in mPA policies are needed to reduce wasteful spending and improve patient care.

Abbreviation Key

mPA

Medication Prior Authorization

COG

Children’s Oncology Group

US

United States

CRA

Clinical Research Associate

CAQH

Council for Affordable Quality Healthcare^®

AHIP

America’s Health Insurance Plans

NCI

National Cancer Institute

NCORP

NCI Community Oncology Research Program

CCDR

Clinical Cancer Delivery Research

HDVCH

Helen DeVos Children’s Hospital