Development and Utility of the Observational Research in Oncology Toolbox: Cancer Medications Enquiry Database-Healthcare Common Procedure Coding System (HCPCS)

Abstract

Purpose

Health-care claims are of increasing utility as a rich, real-world data resource for conducting treatment-related cancer research. However, multiple dynamic coding nomenclatures exist, leading to study variability. To promote increased standardization and reproducibility, the National Cancer Institute (NCI) developed the Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) within the Observational Research in Oncology Toolbox.

Methods

The CanMED-HCPCS includes codes for oncology medications that a) have a US Food and Drug Administration-approved indication for cancer treatment or treatment-related symptom management; b) are present in National Comprehensive Cancer Network guidelines; or c) carry an orphan drug designation for treatment or management of cancer. Included medications and their HCPCS codes were primarily identified based on Center for Medicare and Medicaid Services annual HCPCS Indices (2012–2018). To demonstrate the utility of the CanMED-HCPCS, use of systemic treatment for stage II–IV colorectal cancer patients included in the Surveillance, Epidemiology, and End Results-Medicare data (2007–2013) was assessed.

Results

The CanMED-HCPCS (v2018) includes 332 HCPCS codes for cancer-related medications: chemotherapy (156), immunotherapy (74), hormonal therapy (54), and ancillary therapy (48). Observed treatment trends within the NCI Surveillance, Epidemiology, and End Results-Medicare data were as expected; utilization of each treatment type increased with stage, and immunotherapy was largely confined to use among stage IV patients.

Conclusion

The CanMED-HCPCS provides a comprehensive resource that can be used by the research community to facilitate systematic identification of medications within claims or electronic health data using the HCPCS nomenclature and greater reproducibility of cancer surveillance and health services research.

Healthcare claims are often queried to identify receipt of therapy for observational cancer research. Such studies, including those conducted using the National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) Program-Medicare data, which is a linkage of the SEER Program population-based cancer registry data with Medicare enrollment and claims data (1), rely on the accurate identification of individual medications using codes listed on the health-care claims. The Healthcare Common Procedure Coding System (HCPCS) is one of the primary nomenclatures used for coding medication dispensing and administrative billing (2).

Observational studies have identified trends in the utilization, cost, and effectiveness for cancer treatments by assessing which HCPCS codes are included on administrative claims (3–5). However, varying definitions of cancer therapies (eg, what constitutes chemotherapy vs immunotherapy, modes of administration considered, and relevant medications included), the extensive number of HCPCS codes, and the frequent addition and retirement of HCPCS codes have made comparisons and evaluation of treatment across studies challenging. Varying therapy definitions, which have been especially common in cancer care research because of the rapid approval of new therapies, can affect study design (eg, study population and data sources), misclassification, and, thus, results.

Improved standardized capture of cancer treatments in observational oncology research is essential for comparability of results and reproducibility. In supporting this goal, the NCI is developing the Observational Research in Oncology Toolbox, which was conceptualized to guide systematic, standardized, and reproducible research for observational studies that require the use of specialized ontologies or nomenclatures during study design. The first publicly available resource within the toolbox is the Cancer Medications Enquiry Database (CanMED), which includes comprehensive, queryable, ontologic databases of US Food and Drug Administration (FDA)-approved cancer medications and their associated codes.

The focus of this article is the CanMED-HCPCS, which is a database of oncology medications and associated HCPCS codes. The aims of this article are to provide a methodological description of how this database was created and to demonstrate its utility by examining trends in colorectal cancer (CRC) systemic treatment.

Methods

CanMED-HCPCS Development

Data Sources

Multiple data sources were reviewed to ensure a comprehensive list of oncology medications, and associated HCPCS codes were included in CanMED-HCPCS. The two primary data sources utilized were the Center for Medicare and Medicaid Services (CMS) HCPCS Indices, which include all HCPCS drug codes approved for Medicare billing purposes (6), and the SEER*Rx database, which was created by the NCI to facilitate the coding of cancer medications and for use in the development of regimen ontologies for cancer registry data (7). FDA and commercially available drug databases were also cross-referenced and reviewed to ensure completeness, as described in more detail below (8,9).

Oncology Medication Inclusion Strategy

Oncology medications were eligible for inclusion in the CanMED-HCPCS if they a) had an FDA-approved indication for cancer treatment or treatment-related symptom management, b) were present in the National Comprehensive Cancer Network (NCCN) guidelines, or c) carried an orphan drug designation for the treatment or management of cancer.

The CMS HCPCS Indices were reviewed from 2012 to 2018 to identify Medicare-reimbursable cancer-related medications. In addition, any oncology medications listed in the SEER*Rx database meeting the above criteria were included. The medications identified from either list were reviewed to remove duplicates, misclassified medications (eg, those not actually used for cancer-related treatment), and experimental drugs (eg, clinical trial use exclusively). The combined list of medications was then cross-referenced with two commercially available drug databases, Micromedex and Lexicomp (8,9), to further ensure completeness. Only FDA-approved medications were eligible for inclusion; thus, approval status was verified via Drugs@FDA (10). Medications that are no longer FDA approved are retained to allow for retrospective assessments. Each of the above steps was validated independently by two pharmacists.

Treatment Categories

To ensure consistency with cancer surveillance efforts within the SEER cancer registries, oncologic medications were assigned to a mutually exclusive treatment category: chemotherapy, immunotherapy, hormonal therapy, and ancillary therapy. Chemotherapy is defined as systemic treatment that targets various phases of the cell cycle to destroy cancer cells or interrupt cellular signaling through targeted inhibition of key pathways (11). Immunotherapy is defined as systemic treatment that primarily targets an immune response and mediates tumor growth through immune modification (eg, monoclonal antibodies, checkpoint inhibitors, and vaccines). Hormonal therapy, which is used to treat hormonally linked cancers (eg, breast and prostate cancers), is defined as functional endocrine modifiers that affect hormone production and reduce tumor effects. Ancillary medications are defined as medications required for or directly associated with the administration of chemotherapy, immunotherapy, or hormonal therapy. Many drugs that are used for the management of disease onset, cancer progression, and treatment exacerbations could be included as ancillary medications; however, here, ancillary medications are restricted to chemotherapy administration-related medications or medications specially given for treatment-related symptom management.

HCPCS Nomenclature

HCPCS codes can be divided into two components: Level I codes managed by the American Medical Association as Common Procedure Terminology (CPT) and Level II codes managed by the CMS (2). CPT/Level I HCPCS, or CPT codes, were first developed in 1966. Level II HCPCS are five-digit alphanumeric codes that were introduced by CMS in 1978 for reimbursement of items and services not covered by the CPT/Level I codes, including the administration of cancer-related medications and subsequently adopted by other health care insurers. With each Level II HCPCS code, the letter indicates the type of code (eg, J: permanent codes for medications; C: temporary codes using the outpatient perspective payment system) (12). Level III HCPCS codes were discontinued in 2003 in response to legal requirements that CMS have a uniform coding system. In the CanMED-HCPCS, only Level II HCPCS codes are included. Discontinued HCPCS codes are retained with a discontinuation date to allow for historic analyses.

Special Methodological Considerations

Medications used to treat cancer can be used for other disease indications. For example, erythropoietin has a different HCPCS code when used to treat anemia secondary to cancer treatment effects than when used to treat anemia secondary to end-stage renal disease. The CanMED-HCPCS is cancer focused and thus only includes HCPCS codes for oncologic medications. Assessing utilization of erythropoietin for cancer treatment using all erythropoietin HCPCS codes could result in overestimation and misclassification, whereas limiting to cancer-related HCPCS codes should result in more accurate cancer-related utilization, assuming the accurate code was listed on the health care claim. There are other ancillary medications, such as prednisone, that can be used broadly for noncancer indications and do not have cancer-specific HCPCS codes. As a result, certain ancillary therapy HCPCS codes included in the CanMED-HCPCS are not indication specific.

Additionally, sometimes there is lag-time between when a medication is approved for use and when a HCPCS code is assigned. To make the CanMED-HCPCS as comprehensive as possible, all identified cancer-related systemic medications are included regardless of whether the medication has or will have an associated HCPCS code. As such, if an identified medication did not have a HCPCS code when the CanMED-HCPCS was last updated, the HCPCS code will be listed as not available (“NA”).

Finally, HCPCS codes are primarily used for parenteral medications; however, some oral medications are assigned HCPCS codes such as alternatives to intravenously administered medications (eg, capecitabine) or medications with very specific administration guidelines that allow for initial dosing of certain oral medications (eg, imatinib). Therefore, some oral medications are included in the CanMED-HCPCS; these medications are easily identifiable via the oral treatment indicator variable.

CanMED-HCPCS Database Interface

The CanMED-HCPCS interface includes features that can be used to search by specific drug name (brand or generic) to identify all related HCPCS codes or, conversely, search by HCPCS code to identify a drug name (Figure 1). In addition, the user can limit the search, for example, to medications that were approved during a certain time period or to a specific treatment category (eg, chemotherapy, immunotherapy, hormone therapy, or ancillary therapy). The specific variables available in the CanMED-HCPCS include HCPCS code, Generic Name, Brand Name, Strength, SEER*Rx Category, Major Drug Class, Minor Drug Class, Oral Treatment Indicator, FDA Approval Year, FDA Discontinuation Year, CMS Effective Date, CMS Discontinuation Date, and Status of Product in Use. Query results are exportable to Excel for use with common programming languages such as SAS and R.

Figure 1.

Cancer Medication Enquiry Database-Healthcare Common Procedure Coding System (CanMED-HCPCS) Database Interface.

Use-Case Example

To demonstrate the utility of the CanMED-HCPCS for cancer research, systemic treatment patterns for CRC patients in the SEER-Medicare database were assessed. Patients were included if they were diagnosed with a first primary CRC between 2007 and 2013 at age 66 or older and survived for at least 6 months after diagnosis. Patients also had to have continuous fee-for-service Medicare Part A and B from diagnosis through 6 months postdiagnosis. Systemic therapy is guideline recommended for patients diagnosed with stages II-IV CRC (Derived AJCC Stage, 6th ed.); therefore, patients with earlier or unknown stage disease were excluded.

The assessment of medication use was based on Medicare Part B claims through 2014; Medicare Part A and Part D claims do not include HCPCS codes and, thus, were not considered. Medications assessed were identified using the NCCN guidelines, which allowed for characterization of stage-specific treatment. The corresponding HCPCS codes for the guideline-recommended medications were identified using the CanMED-HCPCS and aggregated by treatment category. Chemotherapies that are used to treat CRC but are only administered orally (eg, regorafenib, trifluridine, and tipiracil) are billed using National Drug Codes (NDCs) and were not assessed; these medications are not reported to CMS via HCPCS codes and were, thus, beyond the objective of this study. An exception is capecitabine; as an orally administered prodrug of the IV-administered 5-Flurouracil, it is covered under Medicare Part B and thus has been assigned HCPCS codes. Although capecitabine is known to be underascertained in Medicare Part B claims, it was included (13). Hormonal therapy was also not assessed because it is not a CRC guideline-recommended treatment.

Descriptive statistics assessing the receipt of each treatment category within 6 months of diagnosis, overall and by stage and calendar year, were calculated to describe treatment trends.

Results

The CanMED-HCPCS (11_2018v1.2.4) includes 496 entries and 332 HCPCS codes for oncology medications: chemotherapy (156), immunotherapy (74), hormonal therapy (54), and ancillary therapy (48) (14). Output from the CanMED-HCPCS allows for the tracking of specific medications to different classification groups. Additional detail on how the medications were mapped to the categories, and major and minor drug classifications, is provided in Supplementary Table 1 (available online). For example, bevacizumab was classified as immunotherapy; major class: monoclonal antibody; minor class: VEGFR.

 

Use-Case Example

Medications included in the NCCN Guidelines for CRC treatment were identified using the CanMED-HCPCS as shown in Table 1. From 2007 to 2013, there were 34 114 patients in the SEER-Medicare data diagnosed with stage II-IV CRC: 44.2% with stage II, 37.5% with stage III, and 18.3% with stage IV (Table 2). During the study period, guideline-directed chemotherapy was administered in 12.2% (annual range = 14.2–9.7%) of stage II patients, 46.7% (annual range = 48.6–44.4%) of stage III patients, and 60.5% (annual range = 58.8–63.4%) of stage IV patients. Similar trends by stage and year were observed for ancillary therapy. Due to low utilization among stage II–III patients (and lack of guideline recommendations), trends in immunotherapy were only reported among stage IV patients. Overall, 41.0% of stage IV patients received immunotherapy (annual range = 38.0–42.9%).

Table 1.

NCCN Guideline-Recommended CRC treatment and associated CanMED-HCPCS^*,† variables for assessing treatment trends in SEER-Medicare, 2007–2013

Generic	Brand	Administration route	FDA approval	Category	Stage II	Stage III	Stage IV	FDA indication	NCCN guidelines	HCPCS
Bevacizumab	Avastin	IV	2004	Immunotherapy			X	Y	Y	Q2024	C9257		C9214	J9035	S0116
Cetuximab	Erbitux	IV	2004	Immunotherapy			X	Y	Y	C9215	J9055
Fluorouracil	Adrucil	IV	1962	Chemotherapy	X	X	X	Y (only stage IV)	Y	J9190
Irinotecan	Camptosar	IV	1996	Chemotherapy			X	Y (only stage IV)	Y	J9206	C9474
Leucovorin Calcium	Wellcovorin	IV only indicated	1952	Chemotherapy	X	X	X	Y	Y	J0640
Levoleucovorin	Fusilev	IV	2008	Chemotherapy		X	X	Y (only stage IV)	Y	J0641
Oxaliplatin	Eloxatin	IV	2002	Chemotherapy	X	X	X	Y	Y	C9205	J9263
Panitumumab	Vectibix	IV	2006	Immunotherapy			X	Y	Y	C9235	J9303
Pembrolizumab	Keytruda	IV	2014	Immunotherapy			X	Y	Y	C9027	J9271
Ramucirumab	Cyramza	IV	2014	Immunotherapy			X	Y	Y	C9025	J9308
Ziv-aflibercept	Zaltrap	IV	2012	Chemotherapy			X	Y	Y	C9296	J9400
Capecitabine	Xeloda	Oral	1998	Chemotherapy	X	X	X	Y	Y	J8520	J8521

^*CanMED-HCPCS (Version 2018) is available at: https://seer.cancer.gov/oncologytoolbox/canmed/hcpcs/. CRC = colorectal cancer; CanMED-HCPCS = Cancer Medications Enquiry Database-Healthcare Common Procedure Coding System; FDA = US Food and Drug Administration; NCCN = National Comprehensive Cancer Network; SEER = National Cancer Institute Surveillance, Epidemiology, and End Results.

†The following drugs were excluded because they are not captured by HCPCS claims: regorafenib (oral), trifluridine, and tipiracil (oral). For the purposes of this analysis, capecitabine was included; however, given it is an oral agent, it may be underascertained. Assessment of oral agents should include claims billed using National Drug Codes.

Table 2.

Trends in the receipt of NCCN guideline-recommended treatment among colorectal cancer patients diagnosed between 2007–2013 in SEER-Medicare using medication codes identified via the CanMED-HCPCS v.2018.

	Stage II					Stage III						Stage IV
Year of diagnosis	Total	Received recommended* chemotherapy		Received any ancillary therapy†		Total	Received recommended* chemotherapy		Received any ancillary therapy†		Total	Received recommended* chemotherapy		Received any ancillary therapy†		Received any immunotherapy‡
	No.	No.	%	No.	%	No.	No.	%	No.	%	No.	No.	%	No.	%	No.	%
2007	2477	351	14.2	352	14.2	2170	1054	48.6	1014	46.7	938	552	58.8	538	57.4	392	41.8
2008	2400	319	13.3	300	12.5	2018	938	46.5	915	45.3	944	564	59.7	545	57.7	389	41.2
2009	2180	283	13.0	272	12.5	1883	904	48.0	862	45.8	926	550	59.4	538	58.1	379	40.9
2010	2147	273	12.7	240	11.2	1743	819	47.0	770	44.2	947	584	61.7	566	59.8	406	42.9
2011	2054	233	11.3	191	9.3	1758	791	45.0	708	40.3	856	516	60.3	487	56.9	325	38.0
2012	2034	204	10.0	178	8.8	1682	789	46.9	743	44.2	817	491	60.1	475	58.1	321	39.3
2013	1775	173	9.7	153	8.6	1542	685	44.4	653	42.3	823	522	63.4	513	62.3	348	42.3
Total	15 067	1836	12.2	1686	11.1	12 796	5980	46.7	5665	44.3	6251	3779	60.5	3662	58.6	2560	41.0

^*Recommended chemotherapy by stage: stage II included capecitabine, fluorouracil, leucovorin calcium, or oxaliplatin; stage III included capecitabine, fluorouracil, leucovorin calcium, levoleucovorin, or oxaliplatin; and stage IV included capecitabine, fluorouracil, irinotecan, leucovorin calcium, levoleucovorin, oxaliplatin, or ziv-aflibercept. CanMED-HCPCS = Cancer Medication Enquiry Database-Healthcare Common Procedure Coding System; NCCN = National Comprehensive Cancer Network.

†Any ancillary therapy: any therapy listed as Ancillary Therapy in CanMED v2018. Non-guideline use for other stages was minimal and is not reported.

^‡Recommended immunotherapy in stage IV included bevacizumab, cetuximab, pembrolizumab, panitumumab, and ramucirumab.

Discussion

The primary objective of the CanMED-HCPCS is to provide a current, comprehensive resource that provides a standardized method of identifying HCPCS codes for systemic oncology medications or, conversely, the identification of medications associated with HCPCS codes extracted from claims. Cancer registries can use the CanMED-HCPCS to appropriately identify medications for registry operations, such as case finding or supplementing registry treatment data using health-care claims. Additionally, the broader cancer research community can use the CanMED-HCPCS to more quickly identify necessary HCPCS codes (eg, for a specific treatment category, drug class, or agent) and increase comparability or reproducibility with other cancer-related studies.

The SEER-Medicare use-case example served as a demonstration of the utility of the CanMED-HCPCS to standardize the identification and categorization of cancer-related medications for a research study. Through this example, we explored the use of various therapeutic categories and illustrated trends by stage and year.

Chemotherapy use increased with advanced stage as expected, as earlier stage patients might be eligible for modalities such as surgery prior to use of chemotherapy. The noticeable difference in use of ancillary therapy between stage II, III, and IV treatments is likely related to the increased use of chemotherapy and immunotherapy in advanced disease along with the potential complications related to more complex regimens or adverse effects, which require additional treatment or supportive care. The NCCN guidelines recommend immunotherapy use only in stage IV CRC, and there was notable (40.9%) use among these patients (Table 2). Since newer immunotherapies continue to have expanded labeling and new indication approvals, it is increasingly important to have a current reference for identifying the correct HCPCS codes associated with these medications for observational studies.

This example confirmed that codes from CanMED-HCPCS can support analyses that assess temporal trends in treatment and concordance with guideline recommendations. Beyond this specific example, codes from CanMED-HCPCS can be used to standardize coding between studies, thus improving the consistency across treatment-related cancer care research. CanMED is a valuable resource to use when creating a cohort to define the appropriate claims-based cancer treatment and to reduce the potential for missed treated cases or misclassification biases (treated vs untreated). Applications for use may include, but are not limited to utilization and dissemination of medications, treatment trends, adherence to guideline-recommended treatment or quality of care, health outcomes, exposure definition for adverse event studies, and general pharmacoepidemiology or health services research.

There are limitations to the development and use of the CanMED-HCPCS. The included HCPCS codes were identified using publicly available coding information. Therefore, it is possible that some historical codes were not included and some fields (eg, CMS starts date) may be left blank if unavailable in the public data. However, we diligently attempted to identify all relevant HCPCS codes, and this is reflected in the most comprehensive, public resource for oncologic therapies to our knowledge. The constantly changing landscape of cancer treatment also makes the prospective development of the CanMED-HCPCS both challenging and necessary. There may be a lag time between when a new therapy is FDA approved and when it is included in the CanMED-HCPCS, given possible delays at CMS in assigning a unique HCPCS code to new medications. Newly approved medications can be billed using a general chemotherapy HCPCS code (eg, J8999 or J9999) if a medication is administered prior to being assigned a specific HCPCS code, though identification of an individual medication exposure in claims data is not possible until they are assigned specific HCPCS codes. However, to stay as relevant as possible, the CanMED-HCPCS will be maintained through a dual review process by NCI clinicians, and any changes will be updated on a yearly basis.

There are other limitations that researchers need to consider when using CanMED-HCPCS. Certain medications, such as methotrexate, can be used for cancer and for multiple other conditions (eg, treatment of rheumatoid and juvenile arthritis). Users should be cautious when including such medications in their cancer-related treatment analyses. It is recommended that medications without indication-specific HCPCS codes should accompany a cancer diagnosis (International Classification of Diseases code), be a specific part of an administered cancer regimen (eg, R-CHOP), or be used in a cancer-specific algorithm in order to be considered cancer-related treatment (15–20). For increased confidence, this recommendation may be extended to any medication assessed.

Researchers need to be cognizant that identification of cancer-related therapies solely using HCPCS codes will lead to underascertainment as more cancer-related medications are exclusively being administered orally and are billed used a different coding nomenclature: NDCs (eg, Medicare Part D). As such, these oral medications are not typically assigned HCPCS codes, the exceptions being initial dosing of oral agents and oral alternatives to IV agents. However, for completeness, all oncology medications will be included in CanMED-HCPCS; if the agent does not have an associated HCPCS code, it will be listed as “NA” in the database. All oral agents will also be included in the companion CanMED-NDC (Rivera et al. NDC). Researchers conducting administrative claims-based studies of cancer treatment are advised to use both the CanMED-HCPCS and CanMED-NDC, when applicable, to identify a comprehensive list of relevant codes to understand exposure and longitudinal medication utilization.

Reproducibility in cancer treatment research is important, given the vast amount of available data resources and the need for real-world evidence, to allow patients and providers to make informed decisions. Standardizing treatment-related research is a pivotal step in the direction of better evidence generation. There are currently no specific guidelines for identifying medication codes; however, the NCI seeks to foster systematic, standardized, and reproducible cancer research for observational studies, and use of CanMED-HCPCS may facilitate standardization of future cancer treatment research. The CanMED-HCPCS provides a consolidated, comprehensive resource that can be used to increase reproducibility of cancer surveillance and health services research using claims or electronic health data that include the HCPCS nomenclature.

Notes

Affiliations of authors: Surveillance Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute, Rockville, MD (DRR, CJKL, VIP, QT, LD, AMN, ABM, LP); Information Management Services, Inc., Calverton, MD (SB, TSM, BO); Healthcare Delivery Research Program, Division of Cancer Control and Population, Sciences, National Cancer Institute, Rockville, MD (LE, DPW).

The authors have no conflicts of interest to declare.