Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Estimating Chemotherapy Use Among Patients With a Prior Primary Cancer Diagnosis Using SEER-Medicare Data

Estimating Chemotherapy Use Among Patients With a Prior Primary Cancer Diagnosis Using... Cancer treatment studies commonly exclude patients with prior primary cancers due to difficulties in ascertaining for which site treatment is intended. Surveillance, Epidemiology, and End Results-Medicare patients 65 years and older diagnosed with an index colon or rectal cancer (CRC) or female breast cancer (BC) between 2004 and 2013 were included. Chemotherapy, de- fined as “any chemotherapy” and more restrictively as “chemotherapy with confirmatory diagnoses,” was ascertained based on claims data within 6 months of index cancer diagnosis by prior cancer history. Any chemotherapy use was slightly lower among patients with a prior cancer (CRC: no prior ¼ 17.4%, prior ¼ 16.1%; BC: no prior ¼ 12.9%, prior ¼ 12.0%). With confirma- tory diagnoses required, estimates were lower, especially among patients with a prior cancer (CRC: no prior ¼ 16.8%, prior ¼ 13.6%; BC: no prior ¼ 12.6%, prior ¼ 11.0%). These findings suggest that patients with prior cancers can be included in studies of chemotherapy use; requiring confirmatory diagnoses can increase treatment assignment confidence. The number of cancer survivors is increasing in the United characteristics (eg, histology and stage) at the time of each diag- States (1). Although the main factor for this increase is an aging nosis. However, information on disease progression and recur- population, advances in cancer therapy have also contributed rence is not available in the registry or claims data. to this growth. The increase in cancer survivorship has resulted Although the exclusion of patients with a prior primary can- cer can simplify study design and conclusions, with the number in a growing number of newly reported cancers being diagnosed among individuals with a prior primary cancer diagnosis; in of cancer patients diagnosed with multiple primary cancers 2000, these cancers represented 15% of all cancers diagnosed growing, not including these patients in treatment use studies and 20% in 2015 (2). results in a sizeable segment of the patient population being ex- Most previous studies that have used cancer registry data cluded and, ultimately, understudied. Therefore, in the hopes of linked to claims data to describe treatment use have excluded promoting the inclusion of patients with multiple primaries in patients with a prior primary cancer (3–17). Patients with a prior research, this study aimed to develop methods to estimate che- cancer are often excluded because they and/or their care are motherapy use among individuals with and without a prior pri- perceived to be different than those without a prior cancer. It is mary cancer. To exemplify these methods, we estimated also difficult to determine, using these data, if the observed che- chemotherapy use for colon or rectal cancer (CRC) and female motherapy was administered for the current cancer or for treat- breast cancer (BC). Two definitions of chemotherapy use were ment of a prior cancer (eg, maintenance or recurrence therapy). assessed: 1) any chemotherapy: at least one chemotherapy Registries collect information on the number and sequence of claim within 6 months of an index CRC or female BC cancer di- lifetime primary cancer diagnoses for each person and clinical agnosis, and 2) chemotherapy with confirmatory cancer Received: 19 September 2019; Revised: 9 January 2020; Accepted: 2 February 2020 Published by Oxford University Press 2020. This work is written by US Government employees and is in the public domain in the US. 14 Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |15 diagnoses: all chemotherapy claims within 6 months of index cancer diagnosis. Patients without a prior primary cancer were cancer diagnosis included a diagnosis code for CRC or BC. Both used as a reference for comparison purposes. measures of chemotherapy use were compared by cancer site Patients with an index cancer with unknown month of diag- nosis, unknown stage, or reporting sources of autopsy or death between individuals with and without a prior primary cancer to assess definition robustness. certificate were excluded. Patients who were reported to the registry by nursing home, convalescent home, or hospice were also excluded because these patients are unlikely to have com- Methods plete treatment information available in the claims data. Patients with Medicare date of death before diagnosis date were Data Source also excluded because of inconsistent date data. Patients were required to have continuous Medicare Part A and Part B cover- We used Surveillance, Epidemiology, and End Results (SEER)- age and enrollment in fee-for-service from the month of index Medicare data, which are a linkage of the National Cancer cancer diagnosis through 6 months after index diagnosis or Institute’s population-based SEER cancer registry data with death, whichever occurred first, and were required to be at least Medicare enrollment and claims data (18). The SEER registries 65 years at index diagnosis. Patients were further excluded if are population based and, at the time of analysis, covered the they were diagnosed with a subsequent primary tumor within states of Connecticut, Hawaii, Iowa, New Mexico, Utah, Georgia, 6 months of their index cancer diagnosis. Patients with a prior California, Kentucky, Louisiana, and New Jersey and the metro- primary cancer were excluded if their prior cancer was distant politan areas of Detroit and Seattle-Puget Sound, representing stage IV at diagnosis, because we assumed that any chemother- approximately 28% of the total US population (19). Each registry apy received after the index cancer diagnosis would likely be for collects information on all patients with newly diagnosed (inci- the prior advanced stage cancer. Summary stage 2000 was used dent) cancer who reside within the designated geographic areas, to categorize all prior cancers because this allowed for the most including demographics (eg, age, sex, race), date of diagnosis consistent classification; inclusion was independent of prior (month/year), reporting source, and clinical characteristics (eg, cancer diagnosis year. The selections for the final cohorts are histology and stage) as well as date of and cause of death, if ap- shown in Supplementary Tables 1 and 2 (available online). plicable. For each person diagnosed with cancer, the registry assigns a number to each tumor to indicate diagnosis sequence: 00 is the only known primary tumor for the individual, 01 is the Chemotherapy Claims Ascertainment first known primary tumor for an individual with more than one tumor, 02 is the second known primary tumor for an indi- Chemotherapy use was ascertained based on the Medicare National Claims History (NCH) files. NCH claims are submitted vidual with at least two tumors, and so on. When an individual is diagnosed with their second tumor, the sequence number of by individual providers and include the level of detail needed their first primary tumor changes from 00 to 01 in the SEER for our definition of chemotherapy with confirmatory diagnosis data. codes, as described below, unlike other claim types. Specifically, Medicare is a federally funded health insurance program each NCH claim has a line item Healthcare Common Procedure that covers in-patient (Part A), out-patient (Part B), and prescrip- Coding System (HCPCS) code indicating the service provided (eg, tion drug (Part D) services. The vast majority (84%; 48 million) of chemotherapy agent administered) and a corresponding International Classification of Diseases version 9 diagnosis Medicare beneficiaries in 2016 were ages 65 years and older (20). Medicare claims include services provided in many settings (eg, code. We assessed claims reporting services provided within hospitals, outpatient clinics, and physician offices). Claims for the 6 months after index cancer diagnosis because patients are beneficiaries enrolled in fee-for-service plans include diagnosis most likely to start receiving chemotherapy within this time and procedure codes with specific dates of service. Together, frame (28). For simplicity, oral chemotherapies covered under SEER-Medicare links 96% of SEER cancer cases diagnosed at ages Medicare Part D were not considered. 65 years and older with their Medicare data. Chemotherapy Definition Study Populations We assessed two definitions of chemotherapy. First, patients We selected cancer patients included in the SEER-Medicare data were classified as receiving “any chemotherapy” if they had at who were aged 65 years or older when diagnosed with either least one claim within 6 months of their index cancer diagnosis CRC or female BC between 2004 and 2013. These sites were cho- that included a HCPCS code for a chemotherapy agent, includ- sen because they represent two of the most common second ing oral alternatives to intravenously administered chemo- primary cancers diagnosed among men and women in the therapies (eg, Capecitabine), which are covered under Medicare United States (21) and these patients often receive chemother- Part B. Medications were identified as being a chemotherapy agent per the National Cancer Institute’s Cancer Medications apy (22–26). To use a consistent staging scheme across the in- cluded diagnosis years, we used Derived American Joint Enquiry Database-HCPCS (v 2018) (Supplementary Table 3, avail- Committee on Cancer 6th edition to categorize index CRC and able online) (29). We excluded claims that had International female BC stage (27). Classification of Diseases version 9 diagnosis code for rheuma- We restricted the included diagnosis years to reflect the toid arthritis (714.0) or senile macular degeneration (362.50– most current treatment patterns at the time of analysis and to 362.52) because some chemotherapy agents are also used to ensure site-specific consistent staging systems during the study treat these conditions. Second, using a more conservative defi- period. We identified two mutually exclusive cohorts of patients nition, patients were considered to have had chemotherapy for each cancer site: patients with no prior primary cancer diag- only if all of their chemotherapy claims included a diagnosis nosis at “index cancer” diagnosis (sequence 00 or 01) and code consistent with the index cancer diagnosis in the registry patients with a prior primary cancer (sequence 02) at index data (Supplementary Table 4, available online). Diagnosis codes Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 16 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 Table 1. Demographics of patients diagnosed at age 65 years or older with CRC or female BC by prior primary cancer diagnosis, SEER-Medicare (2004–2013) CRC Female BC Prior primary cancer diagnosis Prior primary cancer diagnosis † † No* Yes No* Yes No. % No. % No. % No. % All 95 881 100.0 13611 100.0 120 873 100.0 16 480 100.0 Age, y 65–74 39 580 41.3 5 163 37.9 63 664 52.7 8210 49.8 75–84 38 354 40.0 5981 43.9 43 415 35.9 6236 37.8 85þ 17 947 18.7 2467 18.1 13 794 11.4 2034 12.3 Sex Males 44 507 46.4 7748 56.9 — — — — Females 51 374 53.6 5863 43.1 120 873 100.0 16 480 100.0 Stage 0 7207 7.5 1441 10.6 20 923 17.3 3925 23.8 I 22 512 23.5 3961 29.1 52 151 43.1 7710 46.8 II 25 894 27.0 3481 25.6 31 156 25.8 3449 20.9 III 22 486 23.5 2924 21.5 10 074 8.3 977 5.9 IV 17 782 18.5 1804 13.3 6569 5.4 419 2.5 Race White 81 052 84.5 11 734 86.2 104 638 86.6 14 558 88.3 Black 9044 9.4 1309 9.6 10 087 8.3 1201 7.3 Other 5423 5.7 >557 >4.1 5636 4.7 699 4.2 Missing 362 0.4 <11 <0.1 512 0.4 22 0.1 *Patients diagnosed with a first primary cancer (sequence 00, 01) and second primary cancer not diagnosed 0–6 months after first primary cancer. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage. on line diagnosis only were assessed. Patients could have more definition based on site-specific guideline-recommended che- than one chemotherapy claim; therefore, we summarized the motherapy agents (eg, if all chemotherapy claims included a consistency with the cancer registry data across all chemother- listed agent, then the patient was considered to have had che- apy claims for each person. Agreement was categorized as al- motherapy for the index cancer) (22–26). ways, sometimes, and never. These categories, thereby, represented a measure of confidence that the chemotherapy Results should be attributed to the index cancer. We considered a pa- tient to have had confirmatory cancer diagnoses if their chemo- We identified 95 881 and 13 611 CRC cancer patients without therapy claims always included a diagnosis code that was and with a prior primary cancer, and 120 873 and 16 480 female consistent with their index cancer diagnosis per the registry. BC cancer patients without and with a prior primary cancer, re- Due to small numbers and patient confidentiality concerns, spectively (Table 1). Patients with a prior primary cancer tended results were not presented for patients included in the some- to be older than those without, especially among patients diag- times and never agreement groups. nosed with CRC. CRC patients with a prior cancer were also more likely to be male than female compared with CRC patients without a prior cancer. Early (0–I) stage cancers were less com- Statistical Analysis mon among patients without a prior cancer (CRC: 31.0% vs Characteristics of included patients with index CRC and female 39.7%; BC: 60.4% vs 70.6%). The distribution of race was similar across all cohorts. BC were compared by prior cancer history. Then the number and percent of patients who received any chemotherapy and Overall, receipt of any chemotherapy within 6 months of an chemotherapy with confirmatory cancer diagnoses were esti- index cancer diagnosis was slightly lower among patients with mated and compared within and by prior primary cancer status a prior cancer for both CRC (no prior cancer ¼ 17.4%, prior can- for both CRC and female BC. All estimates were calculated strat- cer ¼ 16.1%; Table 2) and female BC (no prior cancer ¼ 12.9%, ified by stage and age. prior cancer ¼ 12.0%; Table 3). However, when stratified by stage Among patients with a prior cancer, sensitivity analyses and age at index cancer, receipt of any chemotherapy tended to were conducted to determine how chemotherapy estimates be slightly higher among patients with a prior cancer, especially varied by time between cancer diagnoses (<5 years vs 5 years) if the index cancer was early (0–I) stage. Chemotherapy esti- and concordance of cancer sites (first and second cancer sites mates were moderately lower when restricted to patients who were the same vs different). Additionally, we assessed the level had claims with confirmatory diagnoses of CRC (no prior cancer of agreement for assigning chemotherapy to the index cancer ¼ 16.8%, prior cancer ¼ 13.6%; Table 2) and female BC (no prior site between our confirmatory diagnosis definition and a cancer ¼ 12.6%, prior cancer ¼ 11.0%; Table 3). Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |17 Table 2. Number and percent of patients receiving any chemotherapy* and the subset who had confirmatory cancer diagnoses within 6 months of CRC diagnosis in 2004–2013 by prior primary can- cer diagnosis, stage, and age, SEER-Medicare Prior primary cancer diagnosis † ‡ No Yes Received chemotherapy Received chemotherapy Patients receiving any Patients receiving any Received any and had confirmatory Received any and had confirmatory Total § ¶ chemotherapy who had Total § ¶ chemotherapy who had chemotherapy CRC diagnoses chemotherapy CRC diagnoses Age at patients confirmatory diagnoses patients confirmatory diagnoses Stage# diagnosis, y No. No. % of all patients No. % of all patients % No. No. % of all patients No. % of all patients % 0 65–74 3568 50 1.4 46 1.3 92.0 659 52 7.9 38 5.8 73.1 75–84 2689 <42 <1.6 <39 <1.5 >90.0 609 26 4.3 18 3.0 69.2 85þ 950 <11 <1.2 <11 <1.2 — 173 <11 <6.4 <11 <6.4 100.0 I 65–74 9501 435 4.6 416 4.4 95.6 1513 127 8.4 99 6.5 78.0 75–84 9187 258 2.8 241 2.6 93.4 1796 74 4.1 54 3.0 73.0 85þ 3824 50 1.3 45 1.2 90.0 652 12 1.8 <11 <1.7 <91.7 II 65–74 9622 1901 19.8 1868 19.4 98.3 1135 227 20.0 186 16.4 81.9 75–84 10 718 949 8.9 916 8.5 96.5 1571 153 9.7 125 8.0 81.7 85þ 5554 79 1.4 78 1.4 98.7 775 <20 <2.6 12 1.5 >60.0 III 65–74 9452 4709 49.8 4627 49.0 98.3 1151 538 46.7 479 41.6 89.0 75–84 8925 2765 31.0 2712 30.4 98.1 1256 389 31.0 351 27.9 90.2 85þ 4109 296 7.2 290 7.1 98.0 517 36 7.0 28 5.4 77.8 IV 65–74 7437 2971 39.9 2795 37.6 94.1 705 273 38.7 234 33.2 85.7 75–84 6835 1870 27.4 1740 25.5 93.0 749 216 28.8 179 23.9 82.9 85þ 3510 291 8.3 273 7.8 93.8 350 49 14.0 39 11.1 79.6 Total 95 881 16 667 17.4 16 086 16.8 96.5 13 611 2193 16.1 1853 13.6 84.5 *Chemotherapy agents identified from Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018 https://seer.cancer.gov/oncologytoolbox/canmed/hcpcs/. Numbers less than 11 cases have been masked to protect patient confidentiality. CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a first primary cancer (sequence 00, 01) who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. §Any Medicare National Claims History claim for chemotherapy within 6 months of index cancer diagnosis. ¶All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with CRC. #Derived Adjusted American Joint Committee on Cancer Stage, 6th edition for the index CRC. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 18 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 Table 3. Number and percent of patients receiving any chemotherapy* and the subset who had confirmatory cancer diagnoses within 6 months of female BC diagnosis in 2004–2013 by prior primary cancer diagnosis, stage, and age, SEER-Medicare Prior Primary Cancer Diagnosis † ‡ No Yes Received chemotherapy Received chemotherapy Patients who received Patients receiving any Received any and had confirmatory Received any and had confirmatory Total § ¶ any chemotherapy who had Total § ¶ chemotherapy who had chemotherapy BC diagnoses chemotherapy BC diagnoses Age at patients confirmatory diagnoses patients confirmatory diagnoses Stage# diagnosis, y No. No. % of all patients No. % of all patients % No. No. % of all patients No. % of all patients % 0 65–74 12 563 28 0.2 13 0.1 46.4 2237 145 6.5 129 5.8 89.0 75–84 6996 12 0.2 <11 <0.2 — 1371 33 2.4 29 2.1 87.9 85þ 1364 0 0.0 0 0.0 — 317 <11 <3.5 <11 <3.5 I 65–74 27 904 2667 9.6 2623 9.4 98.4 3748 493 13.2 456 12.2 92.5 75–84 19 159 592 3.1 561 2.9 94.8 3023 149 4.9 126 4.2 84.6 85þ 5088 38 0.7 29 0.6 76.3 939 15 1.6 11 1.2 73.3 II 65–74 15 358 5425 35.3 5366 34.9 98.9 1585 516 32.6 487 30.7 94.4 75–84 11 250 1554 13.8 1522 13.5 97.9 1322 196 14.8 174 13.2 88.8 85þ 4548 96 2.1 91 2.0 94.8 542 21 3.9 16 3.0 76.2 III 65–74 4866 2634 54.1 2592 53.3 98.4 463 215 46.4 199 43.0 92.6 75–84 3568 1112 31.2 1094 30.7 98.4 350 107 30.6 100 28.6 93.5 85þ 1640 118 7.2 116 7.1 98.3 164 17 10.4 15 9.1 88.2 IV 65–74 2973 800 26.9 755 25.4 94.4 177 45 25.4 37 20.9 82.2 75–84 2442 406 16.6 369 15.1 90.9 170 25 14.7 22 12.9 88.0 85þ 1154 58 5.0 50 4.3 86.2 72 <11 <15.3 <11 <15.3 Total 120 873 15 540 12.9 15 187 12.6 97.7 16 480 1984 12.0 1806 11.0 91.0 *Chemotherapy agents identified from Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018 https://seer.cancer.gov/oncologytoolbox/canmed/hcpcs/. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a first primary cancer (sequence 00, 01) who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. §Any Medicare National Claims History claim for chemotherapy within 6 months of index cancer diagnosis. ¶All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with BC. #Derived Adjusted American Joint Committee on Cancer Stage, 6th edition for the index BC. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |19 Table 4. Number and percent of second primary CRC and female BC* patients who received any chemotherapy by whether or not their chemo- therapy was associated with a confirmatory site-specific diagnosis code, years since first primary cancer diagnosis, and concordance of first and second primary tumor sites, SEER-Medicare. Number of years from first First primary and primary cancer diagnosis to second primary cancer Received second primary cancer diagnosis site are same site recommended chemo Received chemotherapy <5 years 5 years Same site Other sites and had confirmatory Total Yes No Agreement ‡ § § § § Cancer diagnoses No. No. % No. % No. % No. % No. No. % CRC Yes 1853 1326 71.6 527 28.4 659 35.6 1194 64.4 1838 15 88.4 No 340 260 76.5 80 23.5 17 5.0 323 95.0 239 101 Total 2193 1586 72.3 607 27.7 676 30.8 1517 69.2 2077 116 Female BC Yes 1806 1194 66.1 612 33.9 1299 71.9 507 28.1 1804 <11 >91.5% No 178 150 84.3 28 15.7 18 10.1 160 89.9 164 >11 Total 1984 1344 67.7 640 32.3 1317 66.4 667 33.6 1968 16 *Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diag- nosis 0–6 months after index cancer diagnosis date. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Any Medicare National Claims History claim for chemotherapy as identified chemotherapy agents identified from the Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018. All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with CRC or fe- male BC, respectively. §Row percent. Sensitivity analyses among the 2193 CRC and 1984 BC We required confirmatory diagnoses on claims as a first step patients who had a prior cancer and were identified as having in creating a more automated method to attribute chemotherapy at least one chemotherapy claim within 6 months of their index to a specific cancer among patients with multiple cancers found in large population-based data. This definition resulted in a mea- cancer diagnosis were conducted. These analyses indicated that sure that was easy to execute (eg, only considered claim diagno- a longer time interval (<5 years vs 5 years) between cancer di- sis codes), yet we believe it can provide high confidence in agnoses was associated with being classified as receiving che- discerning if chemotherapy was administered for the index can- motherapy for the index cancer, per our confirmatory diagnosis cer. This definition was conservative because it included only definition, especially for female BC (CRC: 28.4% vs 23.5%; BC: patients for whom there was always agreement between chemo- 33.9% vs 15.7%; Table 4). Not surprisingly, having a prior concor- therapy claim diagnosis codes and the registry cancer site. This dant site cancer was also associated with being classified as re- definition can be refined depending on the research question. ceiving chemotherapy for the index cancer, per our Although we found high agreement when receipt of chemother- confirmatory diagnosis definition. Finally, we found high agree- apy was classified by agent, specific chemotherapy agents could ment between assigning chemotherapy to the index cancer us- be reviewed. If an agent is consistent with guidelines for the in- ing our confirmatory diagnosis definition and a definition based dex cancer then this could be considered evidence that the che- on site-specific guideline-recommended agents (CRC ¼ 88.4%; motherapy was administered for the index cancer, even if the female BC ¼ 91.5%). claim diagnosis codes were not always consistent with the regis- try data. The National Comprehensive Cancer Network clinical Discussion guidelines are an accessible source of chemotherapy drug and regimen information and could be used to build lists of recom- In this study, we estimated receipt of chemotherapy within mended cancer-specific drugs (24–26). However, this approach 6 months of index cancer diagnosis among patients with a prior would require additional consideration when an agent is recom- primary cancer diagnosis compared with patients without a mended for treatment for both the first and second cancers. prior cancer for two common cancer sites: CRC and female BC. Our approach for assigning chemotherapy to a cancer site dif- Overall, CRC and female BC patients with a prior primary cancer fers from two prior studies that included patients with multiple had only slightly lower chemotherapy use than patients with- primary cancers to study treatment patterns and outcomes. In out a prior cancer. Stratified by age and stage, CRC and female one study, patients diagnosed with multiple primary cancers BC patients with a prior primary cancer tended to have slightly were assigned to one type of cancer based on a hierarchical list higher chemotherapy use than patients without a prior cancer. of cancers (eg, if a patient’s claims data included diagnoses codes Additionally, more than 95% of CRC and female BC patients indicating more than one cancer type, all documented chemo- without a prior cancer who received chemotherapy had confir- therapy was assigned to the higher ranked cancer) (30). In an- matory cancer diagnoses listed on their chemotherapy claims. other analysis, chemotherapy was assigned to a specific cancer if For patients with a prior primary cancer who received chemo- at least 60% of the therapy claims included diagnosis codes indic- therapy, a lower percent had claims with confirmatory diagno- ative of only one cancer site (31). Neither study included cancer ses (85% CRC, 91% BC). The lower consistency between registry data linked to therapy claims as was done in our study. diagnoses listed in the registry and claim data among patients In addition to our chemotherapy definition being useful for with a prior cancer may be due to the chemotherapy being ad- assessing chemotherapy use for health services research, it ministered as treatment for a prior cancer. may be relevant to cancer registry functions. Registries could Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 20 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 apply the site-specific chemotherapy definition to automati- diagnosis, and it appears that more time between diagnosis in- cally determine receipt of chemotherapy and assign it to a given creased the likelihood of assigning the chemotherapy to the in- cancer. When there is always agreement between chemother- dex cancer. apy claims diagnosis codes and the registry cancer site, chemo- In conclusion, patients diagnosed with multiple primary therapy could automatically be reported as “yes” for the cancer. tumors have commonly been excluded from claims-based stud- Manual review of records could then be implemented for the ies of chemotherapy use due to the complexities in distinguish- remaining cases that may have greater complexity and thus did ing whether treatment was being used for the index or prior not always have diagnosis codes that matched the registry can- cancer. An Institute of Medicine Report on Delivery of High- cer site. Thereby, use of our chemotherapy definition could in- Quality Cancer Care in 2013 recommended as one of its 10 prior- crease cancer registry efficiency. ity areas to “Expand breadth of cancer research data collected We found that a very small number of in situ (stage 0) first so that they reflect the population with the disease, such as the CRCs and female BC received chemotherapy per our definitions. elderly and patients with comorbid conditions” (32). This study Given that chemotherapy is rarely administered for the treat- demonstrated a simple approach to determine for which cancer ment of in situ cancers, potential explanations are overtreat- chemotherapy is administered and thus can aid researchers ment, incorrect diagnosis listed on the claims, or possibly and cancer registrars to more accurately and efficiently deter- upstaging of the cancer based on later tests, which was not mine treatment intent. These findings suggest that requiring reflected in the registry data. confirmatory diagnoses can increase treatment assignment Our study demonstrated the use of claims data to estimate confidence. receipt of chemotherapy for patients with prior cancers. The confirmatory diagnosis codes on claims can be used in multiple Notes ways to expand and improve chemotherapy attribution at the cancer registries and in treatment pattern studies. However, we Author Affiliations: Data Analytics Branch, Surveillance did have some limitations that need to be considered. Some of Research Program, Division of Cancer Control and Population the diagnosis codes we used to determine whether the claim Sciences, National Cancer Institute, Rockville, MD (CL, AM); was for the treatment of cancer were generic codes (eg, encoun- Healthcare Assessment Research Branch, Healthcare Delivery ter for antineoplastic chemotherapy). Thus, there was no way to Research Program, Division of Cancer Control and Population tell whether a claim with these codes matched the index cancer. Sciences, National Cancer Institute, Bethesda, MD (LE, DW, JW); However, the majority of the chemotherapy claims assessed in- Information Management Services, Inc., Calverton, MD (TM) cluded confirmatory cancer codes (data not shown). There is no conflict of interest disclosures from any authors. Additionally, if the first primary cancer was the same cancer site as the second primary cancer, our chemotherapy definition will not be able to determine if the diagnosis codes on the che- References motherapy claims were truly for the index cancer. Furthermore, 1. Bluethmann SM, Mariotto AB, Rowland JH. Anticipating the “Silver with the shift to targeted therapies, which are used to treat mul- Tsunami”: prevalence trajectories and comorbidity burden among older can- cer survivors in the United States. Cancer Epidemiol Biomarkers Prev. 2016;25(7): tiple cancers, it will become increasingly challenging to identify 1029–1036. which cancer is the target of a specific treatment. Depending on 2. National Cancer Institute Surveillance, Epidemiology, and End Results the research question, these factors may or may not be impor- Program SEER 18 Regs Research Data, Nov 2017 Sub (1973-2015). National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance tant. We only investigated second cancers; however, the per- Systems Branch. http://www.seer.cancer.gov. Published 2017. Accessed cent of patients diagnosed with more than two cancers is small December 15, 2018. (4%). Assigning and interpreting treatment patterns for patients 3. Warren JL, Yabroff KR, Meekins A, Topor M, Lamont EB, Brown ML. Evaluation of trends in the cost of initial cancer treatment. J Natl Cancer Inst. with more than one prior cancer increases the complexity; how- 2008;100(12):888–897. ever, requiring all diagnosis codes to match the registry data 4. Bradley CJ, Yabroff KR, Warren JL, Zeruto C, Chawla N, Lamont EB. Trends in could still indicate high confidence that the chemotherapy was the treatment of metastatic colon and rectal cancer in elderly patients. Med administered for the index cancer. Furthermore, we did not in- Care. 2016;54(5):490–497. 5. Morton LM, Dores GM, Schonfeld SJ, et al. Association of chemotherapy for clude patients if they had a prior cancer that was distant stage solid tumors with development of therapy-related myelodysplastic syn- IV at diagnosis, because we assumed that any chemotherapy re- drome or acute myeloid leukemia in the modern era. JAMA Oncol. 2018;5(3): ceived after the index cancer diagnosis would likely be for the 318–325. doi: 10.1001/jamaoncol.2018.5625 6. Tong L, Ahn C, Symanski E, Lai D, Du XL. Effects of newly developed chemo- prior, advanced-stage cancer, and we would not have been able therapy regiments, comorbidities, chemotherapy-related toxicities on the to distinguish treatment between the two cancers. changing patterns of the leading causes of death in elderly patients with Confirmatory cancer diagnoses on the claims can be difficult CRC. Ann Oncol. 2014;25(6):1234–1242. 7. Gilden DM, Kubisiak JM, Pohl GM, et al. Treatment patterns and cost- to assess for oral chemotherapies, particularly for agents cov- effectiveness of first line treatment of advanced non-squamous non-small ered under Medicare Part D because the associated claims do cell lung cancer in Medicare patients. J Med Econ. 2017;20(2):151–161. not include diagnosis codes. If assessing oral agents is para- 8. Medeiros BC, Satram-Hoang S, Hurst D, Hoang KQ, Momin F, Reyes C. Big data mount for a given study, it is recommended that the specific analysis of treatment patterns and outcomes among elderly acute myeloid leukemia patients in the United States. Ann Hematol. 2015;94(7):1127–1138. agents be assessed to determine whether the regimen is consis- 9. Satram-Hoang S, Reyes C, Hoang KQ, Momin F, Skettino S. Treatment practice tent with treatment guidelines for a particular cancer. Lastly, in the elderly patient with chronic lymphocytic leukemia – analysis of the for patients with a prior cancer, time between prior and index combined SEER and Medicare database. Ann Hematol. 2014;93(8):1335–1344. 10. Shahinian VB, Yong-Fang K, Freeman JL, Orihuela E, Goodwin JS. Increasing cancer diagnoses was not considered in our definitions; the use of gonadotropin-releasing hormone agonists for the treatment of local- ideal temporal cut-point for assigning treatment to a given can- ized prostate carcinoma. Cancer. 2005;103(8):1615–1624. cer could vary by multiple variables (eg, cancer site, stage, and 11. Obeidat NA, Pradel FG, Zuckerman IH, DeLisle S, Mullins CD. Outcomes of irinotecan-based chemotherapy regimens in elderly Medicare patients with recurrence rate). However, our sensitivity analyses by number metastatic CRC. Am J Geriatr Pharmacother. 2009;7(6):343–354. of years between prior and index cancer did show that for both 12. Rajan SS, Stearns SC, Lyman GH, Carpenter WR. Effect of primary prophylac- CRC and female BC, the majority of the second primary cancers tic G-CSF use on chemotherapy administration for elderly BC patients. Breast were diagnosed less than 5 years after the first primary cancer Cancer Res Treat. 2011;130(1):255–266. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |21 13. Griffiths RI, Barron RL, Gleeson ML, et al. Granulocyte-colony stimulating fac- 23. American Cancer Society. Treatment of colon cancer, by stage. 2019. https:// tor use and medical costs after initial adjuvant chemotherapy in older www.cancer.org/cancer/colon-rectal-cancer/treating/by-stage-colon.html. patients with early-stage BC. Pharmacoeconomics. 2012;30(2):103–118. Accessed December 15, 2019. 14. Garcia C, Yao A, Camacho F, Balkrishnan R, Cantrell LA. A SEER-Medicare 24. National Comprehensive Cancer Network. NCCN guidelines for treatment of analysis of the impact of metformin on overall survival in ovarian cancer. cancer by site: BC. https://www.nccn.org/professionals/physician_gls/. Gynecol Oncol. 2017;146(2):346–350. Accessed April 18, 2019. 15. Yu D, Gao P, Song Y, et al. The differences on efficacy of oxaliplatin in locally 25. National Comprehensive Cancer Network. NCCN guidelines for treatment of advanced colon cancer between mucinous and nonmucinous adenocarci- cancer by site: colon cancer. https://www.nccn.org/professionals/physician_ noma. Cancer Med. 2018;7(3):600–615. gls/. Published January 15, 2019. Accessed April 18, 2019. 16. Sharma M, Holmes HM, Mehta HB, et al. The concomitant use of tyrosine ki- 26. National Comprehensive Cancer Network. NCCN guidelines for treatment of nase inhibitors and proton pump inhibitors: prevalence predictors, and im- cancer by site: rectal cancer. https://www.nccn.org/professionals/physician_ pact on survival and discontinuation of therapy in older adults with cancer. gls/. Published January 15, 2019. Accessed April 18, 2019. Cancer. 2019;125(7):1155–1162. 27. Derived AJCC Cancer Staging. National Cancer Institute. Surveillance, 17. Rocque GB, Kandhare PG, Williams CP, et al. Visualization of sequential treat- Epidemiology, and End Results Program, 2019. seer.cancer.gov/seerstat/ ments in metastatic BC. J Clin Oncol Clin Cancer Inform. 2019;3:1–8. variables/ajcc-stage/derived-ajcc.html. Accessed July 8, 2019. 18. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the 28. Howlader N, Ward KC, Warren JL, Campbell DS, Coyle L, Mariotto AB. SEER-Medicare data: content, research applications, and generalizability Assessment of oncology practice billing claims for capturing chemotherapy: to the United States elderly population. Med Care. 2002;40(8 suppl): a pilot study in Georgia SEER Cancer Registry. J Natl Cancer Inst Monogr. 2020; IV3–IV18. 2020(55):82–88. 19. Registry Groupings in SEER Data and Statistics. http://seer.cancer.gov/regis 29. Rivera DR, Lam CJK, Enewold L, et al. Development and utility of the observa- tries/terms.html. Accessed June 15, 2019. tional research in oncology toolbox: Cancer Medications Enquiry Database 20. 2016 Medicare Enrollment Section. https://www.cms.gov/Research- (CanMED) Healthcare Common Procedure Coding System (HCPCS). J Natl Statistics-Data-and-Systems/Statistics-Trends-and-Reports/ Cancer Inst Monogr. 2020;2020:(55):39–45. CMSProgramStatistics/2016/2016_Enrollment.html. Published March 15, 30. Hayes J, Hoverman JR, Brow ME, et al. Cost differential by site of service for can- 2017. Accessed June 15, 2019. cer patients receiving chemotherapy. Am J Manag Care. 2015;21(3):e189–e196. 21. American Cancer Society. Cancer Facts & Figures 2019. Atlanta, GA: American 31. Avalere Health, LLC. Total cost of cancer care by site of service: physician of- Cancer Society; 2019. fice vs outpatient hospital. 2012. 22. American Cancer Society. Treatment of BC by stage. 2019. https://www.can 32. Ganz PA. Institute of Medicine report on delivery of high-quality cancer care. cer.org/cancer/breast-cancer/treatment/treatment-of-breast-cancer-by- stage.html. Accessed December 15, 2019. J Oncol Pract. 2014;10(3):193–195. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI Monographs Oxford University Press

Estimating Chemotherapy Use Among Patients With a Prior Primary Cancer Diagnosis Using SEER-Medicare Data

Download PDF
8 pages

Loading next page...
 
/lp/oxford-university-press/estimating-chemotherapy-use-among-patients-with-a-prior-primary-cancer-OMGIubzvkq

References (28)

Publisher
Oxford University Press
Copyright
Copyright © 2022 Oxford University Press
ISSN
1052-6773
eISSN
1745-6614
DOI
10.1093/jncimonographs/lgaa005
Publisher site
See Article on Publisher Site

Abstract

Cancer treatment studies commonly exclude patients with prior primary cancers due to difficulties in ascertaining for which site treatment is intended. Surveillance, Epidemiology, and End Results-Medicare patients 65 years and older diagnosed with an index colon or rectal cancer (CRC) or female breast cancer (BC) between 2004 and 2013 were included. Chemotherapy, de- fined as “any chemotherapy” and more restrictively as “chemotherapy with confirmatory diagnoses,” was ascertained based on claims data within 6 months of index cancer diagnosis by prior cancer history. Any chemotherapy use was slightly lower among patients with a prior cancer (CRC: no prior ¼ 17.4%, prior ¼ 16.1%; BC: no prior ¼ 12.9%, prior ¼ 12.0%). With confirma- tory diagnoses required, estimates were lower, especially among patients with a prior cancer (CRC: no prior ¼ 16.8%, prior ¼ 13.6%; BC: no prior ¼ 12.6%, prior ¼ 11.0%). These findings suggest that patients with prior cancers can be included in studies of chemotherapy use; requiring confirmatory diagnoses can increase treatment assignment confidence. The number of cancer survivors is increasing in the United characteristics (eg, histology and stage) at the time of each diag- States (1). Although the main factor for this increase is an aging nosis. However, information on disease progression and recur- population, advances in cancer therapy have also contributed rence is not available in the registry or claims data. to this growth. The increase in cancer survivorship has resulted Although the exclusion of patients with a prior primary can- cer can simplify study design and conclusions, with the number in a growing number of newly reported cancers being diagnosed among individuals with a prior primary cancer diagnosis; in of cancer patients diagnosed with multiple primary cancers 2000, these cancers represented 15% of all cancers diagnosed growing, not including these patients in treatment use studies and 20% in 2015 (2). results in a sizeable segment of the patient population being ex- Most previous studies that have used cancer registry data cluded and, ultimately, understudied. Therefore, in the hopes of linked to claims data to describe treatment use have excluded promoting the inclusion of patients with multiple primaries in patients with a prior primary cancer (3–17). Patients with a prior research, this study aimed to develop methods to estimate che- cancer are often excluded because they and/or their care are motherapy use among individuals with and without a prior pri- perceived to be different than those without a prior cancer. It is mary cancer. To exemplify these methods, we estimated also difficult to determine, using these data, if the observed che- chemotherapy use for colon or rectal cancer (CRC) and female motherapy was administered for the current cancer or for treat- breast cancer (BC). Two definitions of chemotherapy use were ment of a prior cancer (eg, maintenance or recurrence therapy). assessed: 1) any chemotherapy: at least one chemotherapy Registries collect information on the number and sequence of claim within 6 months of an index CRC or female BC cancer di- lifetime primary cancer diagnoses for each person and clinical agnosis, and 2) chemotherapy with confirmatory cancer Received: 19 September 2019; Revised: 9 January 2020; Accepted: 2 February 2020 Published by Oxford University Press 2020. This work is written by US Government employees and is in the public domain in the US. 14 Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |15 diagnoses: all chemotherapy claims within 6 months of index cancer diagnosis. Patients without a prior primary cancer were cancer diagnosis included a diagnosis code for CRC or BC. Both used as a reference for comparison purposes. measures of chemotherapy use were compared by cancer site Patients with an index cancer with unknown month of diag- nosis, unknown stage, or reporting sources of autopsy or death between individuals with and without a prior primary cancer to assess definition robustness. certificate were excluded. Patients who were reported to the registry by nursing home, convalescent home, or hospice were also excluded because these patients are unlikely to have com- Methods plete treatment information available in the claims data. Patients with Medicare date of death before diagnosis date were Data Source also excluded because of inconsistent date data. Patients were required to have continuous Medicare Part A and Part B cover- We used Surveillance, Epidemiology, and End Results (SEER)- age and enrollment in fee-for-service from the month of index Medicare data, which are a linkage of the National Cancer cancer diagnosis through 6 months after index diagnosis or Institute’s population-based SEER cancer registry data with death, whichever occurred first, and were required to be at least Medicare enrollment and claims data (18). The SEER registries 65 years at index diagnosis. Patients were further excluded if are population based and, at the time of analysis, covered the they were diagnosed with a subsequent primary tumor within states of Connecticut, Hawaii, Iowa, New Mexico, Utah, Georgia, 6 months of their index cancer diagnosis. Patients with a prior California, Kentucky, Louisiana, and New Jersey and the metro- primary cancer were excluded if their prior cancer was distant politan areas of Detroit and Seattle-Puget Sound, representing stage IV at diagnosis, because we assumed that any chemother- approximately 28% of the total US population (19). Each registry apy received after the index cancer diagnosis would likely be for collects information on all patients with newly diagnosed (inci- the prior advanced stage cancer. Summary stage 2000 was used dent) cancer who reside within the designated geographic areas, to categorize all prior cancers because this allowed for the most including demographics (eg, age, sex, race), date of diagnosis consistent classification; inclusion was independent of prior (month/year), reporting source, and clinical characteristics (eg, cancer diagnosis year. The selections for the final cohorts are histology and stage) as well as date of and cause of death, if ap- shown in Supplementary Tables 1 and 2 (available online). plicable. For each person diagnosed with cancer, the registry assigns a number to each tumor to indicate diagnosis sequence: 00 is the only known primary tumor for the individual, 01 is the Chemotherapy Claims Ascertainment first known primary tumor for an individual with more than one tumor, 02 is the second known primary tumor for an indi- Chemotherapy use was ascertained based on the Medicare National Claims History (NCH) files. NCH claims are submitted vidual with at least two tumors, and so on. When an individual is diagnosed with their second tumor, the sequence number of by individual providers and include the level of detail needed their first primary tumor changes from 00 to 01 in the SEER for our definition of chemotherapy with confirmatory diagnosis data. codes, as described below, unlike other claim types. Specifically, Medicare is a federally funded health insurance program each NCH claim has a line item Healthcare Common Procedure that covers in-patient (Part A), out-patient (Part B), and prescrip- Coding System (HCPCS) code indicating the service provided (eg, tion drug (Part D) services. The vast majority (84%; 48 million) of chemotherapy agent administered) and a corresponding International Classification of Diseases version 9 diagnosis Medicare beneficiaries in 2016 were ages 65 years and older (20). Medicare claims include services provided in many settings (eg, code. We assessed claims reporting services provided within hospitals, outpatient clinics, and physician offices). Claims for the 6 months after index cancer diagnosis because patients are beneficiaries enrolled in fee-for-service plans include diagnosis most likely to start receiving chemotherapy within this time and procedure codes with specific dates of service. Together, frame (28). For simplicity, oral chemotherapies covered under SEER-Medicare links 96% of SEER cancer cases diagnosed at ages Medicare Part D were not considered. 65 years and older with their Medicare data. Chemotherapy Definition Study Populations We assessed two definitions of chemotherapy. First, patients We selected cancer patients included in the SEER-Medicare data were classified as receiving “any chemotherapy” if they had at who were aged 65 years or older when diagnosed with either least one claim within 6 months of their index cancer diagnosis CRC or female BC between 2004 and 2013. These sites were cho- that included a HCPCS code for a chemotherapy agent, includ- sen because they represent two of the most common second ing oral alternatives to intravenously administered chemo- primary cancers diagnosed among men and women in the therapies (eg, Capecitabine), which are covered under Medicare United States (21) and these patients often receive chemother- Part B. Medications were identified as being a chemotherapy agent per the National Cancer Institute’s Cancer Medications apy (22–26). To use a consistent staging scheme across the in- cluded diagnosis years, we used Derived American Joint Enquiry Database-HCPCS (v 2018) (Supplementary Table 3, avail- Committee on Cancer 6th edition to categorize index CRC and able online) (29). We excluded claims that had International female BC stage (27). Classification of Diseases version 9 diagnosis code for rheuma- We restricted the included diagnosis years to reflect the toid arthritis (714.0) or senile macular degeneration (362.50– most current treatment patterns at the time of analysis and to 362.52) because some chemotherapy agents are also used to ensure site-specific consistent staging systems during the study treat these conditions. Second, using a more conservative defi- period. We identified two mutually exclusive cohorts of patients nition, patients were considered to have had chemotherapy for each cancer site: patients with no prior primary cancer diag- only if all of their chemotherapy claims included a diagnosis nosis at “index cancer” diagnosis (sequence 00 or 01) and code consistent with the index cancer diagnosis in the registry patients with a prior primary cancer (sequence 02) at index data (Supplementary Table 4, available online). Diagnosis codes Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 16 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 Table 1. Demographics of patients diagnosed at age 65 years or older with CRC or female BC by prior primary cancer diagnosis, SEER-Medicare (2004–2013) CRC Female BC Prior primary cancer diagnosis Prior primary cancer diagnosis † † No* Yes No* Yes No. % No. % No. % No. % All 95 881 100.0 13611 100.0 120 873 100.0 16 480 100.0 Age, y 65–74 39 580 41.3 5 163 37.9 63 664 52.7 8210 49.8 75–84 38 354 40.0 5981 43.9 43 415 35.9 6236 37.8 85þ 17 947 18.7 2467 18.1 13 794 11.4 2034 12.3 Sex Males 44 507 46.4 7748 56.9 — — — — Females 51 374 53.6 5863 43.1 120 873 100.0 16 480 100.0 Stage 0 7207 7.5 1441 10.6 20 923 17.3 3925 23.8 I 22 512 23.5 3961 29.1 52 151 43.1 7710 46.8 II 25 894 27.0 3481 25.6 31 156 25.8 3449 20.9 III 22 486 23.5 2924 21.5 10 074 8.3 977 5.9 IV 17 782 18.5 1804 13.3 6569 5.4 419 2.5 Race White 81 052 84.5 11 734 86.2 104 638 86.6 14 558 88.3 Black 9044 9.4 1309 9.6 10 087 8.3 1201 7.3 Other 5423 5.7 >557 >4.1 5636 4.7 699 4.2 Missing 362 0.4 <11 <0.1 512 0.4 22 0.1 *Patients diagnosed with a first primary cancer (sequence 00, 01) and second primary cancer not diagnosed 0–6 months after first primary cancer. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage. on line diagnosis only were assessed. Patients could have more definition based on site-specific guideline-recommended che- than one chemotherapy claim; therefore, we summarized the motherapy agents (eg, if all chemotherapy claims included a consistency with the cancer registry data across all chemother- listed agent, then the patient was considered to have had che- apy claims for each person. Agreement was categorized as al- motherapy for the index cancer) (22–26). ways, sometimes, and never. These categories, thereby, represented a measure of confidence that the chemotherapy Results should be attributed to the index cancer. We considered a pa- tient to have had confirmatory cancer diagnoses if their chemo- We identified 95 881 and 13 611 CRC cancer patients without therapy claims always included a diagnosis code that was and with a prior primary cancer, and 120 873 and 16 480 female consistent with their index cancer diagnosis per the registry. BC cancer patients without and with a prior primary cancer, re- Due to small numbers and patient confidentiality concerns, spectively (Table 1). Patients with a prior primary cancer tended results were not presented for patients included in the some- to be older than those without, especially among patients diag- times and never agreement groups. nosed with CRC. CRC patients with a prior cancer were also more likely to be male than female compared with CRC patients without a prior cancer. Early (0–I) stage cancers were less com- Statistical Analysis mon among patients without a prior cancer (CRC: 31.0% vs Characteristics of included patients with index CRC and female 39.7%; BC: 60.4% vs 70.6%). The distribution of race was similar across all cohorts. BC were compared by prior cancer history. Then the number and percent of patients who received any chemotherapy and Overall, receipt of any chemotherapy within 6 months of an chemotherapy with confirmatory cancer diagnoses were esti- index cancer diagnosis was slightly lower among patients with mated and compared within and by prior primary cancer status a prior cancer for both CRC (no prior cancer ¼ 17.4%, prior can- for both CRC and female BC. All estimates were calculated strat- cer ¼ 16.1%; Table 2) and female BC (no prior cancer ¼ 12.9%, ified by stage and age. prior cancer ¼ 12.0%; Table 3). However, when stratified by stage Among patients with a prior cancer, sensitivity analyses and age at index cancer, receipt of any chemotherapy tended to were conducted to determine how chemotherapy estimates be slightly higher among patients with a prior cancer, especially varied by time between cancer diagnoses (<5 years vs 5 years) if the index cancer was early (0–I) stage. Chemotherapy esti- and concordance of cancer sites (first and second cancer sites mates were moderately lower when restricted to patients who were the same vs different). Additionally, we assessed the level had claims with confirmatory diagnoses of CRC (no prior cancer of agreement for assigning chemotherapy to the index cancer ¼ 16.8%, prior cancer ¼ 13.6%; Table 2) and female BC (no prior site between our confirmatory diagnosis definition and a cancer ¼ 12.6%, prior cancer ¼ 11.0%; Table 3). Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |17 Table 2. Number and percent of patients receiving any chemotherapy* and the subset who had confirmatory cancer diagnoses within 6 months of CRC diagnosis in 2004–2013 by prior primary can- cer diagnosis, stage, and age, SEER-Medicare Prior primary cancer diagnosis † ‡ No Yes Received chemotherapy Received chemotherapy Patients receiving any Patients receiving any Received any and had confirmatory Received any and had confirmatory Total § ¶ chemotherapy who had Total § ¶ chemotherapy who had chemotherapy CRC diagnoses chemotherapy CRC diagnoses Age at patients confirmatory diagnoses patients confirmatory diagnoses Stage# diagnosis, y No. No. % of all patients No. % of all patients % No. No. % of all patients No. % of all patients % 0 65–74 3568 50 1.4 46 1.3 92.0 659 52 7.9 38 5.8 73.1 75–84 2689 <42 <1.6 <39 <1.5 >90.0 609 26 4.3 18 3.0 69.2 85þ 950 <11 <1.2 <11 <1.2 — 173 <11 <6.4 <11 <6.4 100.0 I 65–74 9501 435 4.6 416 4.4 95.6 1513 127 8.4 99 6.5 78.0 75–84 9187 258 2.8 241 2.6 93.4 1796 74 4.1 54 3.0 73.0 85þ 3824 50 1.3 45 1.2 90.0 652 12 1.8 <11 <1.7 <91.7 II 65–74 9622 1901 19.8 1868 19.4 98.3 1135 227 20.0 186 16.4 81.9 75–84 10 718 949 8.9 916 8.5 96.5 1571 153 9.7 125 8.0 81.7 85þ 5554 79 1.4 78 1.4 98.7 775 <20 <2.6 12 1.5 >60.0 III 65–74 9452 4709 49.8 4627 49.0 98.3 1151 538 46.7 479 41.6 89.0 75–84 8925 2765 31.0 2712 30.4 98.1 1256 389 31.0 351 27.9 90.2 85þ 4109 296 7.2 290 7.1 98.0 517 36 7.0 28 5.4 77.8 IV 65–74 7437 2971 39.9 2795 37.6 94.1 705 273 38.7 234 33.2 85.7 75–84 6835 1870 27.4 1740 25.5 93.0 749 216 28.8 179 23.9 82.9 85þ 3510 291 8.3 273 7.8 93.8 350 49 14.0 39 11.1 79.6 Total 95 881 16 667 17.4 16 086 16.8 96.5 13 611 2193 16.1 1853 13.6 84.5 *Chemotherapy agents identified from Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018 https://seer.cancer.gov/oncologytoolbox/canmed/hcpcs/. Numbers less than 11 cases have been masked to protect patient confidentiality. CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a first primary cancer (sequence 00, 01) who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. §Any Medicare National Claims History claim for chemotherapy within 6 months of index cancer diagnosis. ¶All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with CRC. #Derived Adjusted American Joint Committee on Cancer Stage, 6th edition for the index CRC. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 18 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 Table 3. Number and percent of patients receiving any chemotherapy* and the subset who had confirmatory cancer diagnoses within 6 months of female BC diagnosis in 2004–2013 by prior primary cancer diagnosis, stage, and age, SEER-Medicare Prior Primary Cancer Diagnosis † ‡ No Yes Received chemotherapy Received chemotherapy Patients who received Patients receiving any Received any and had confirmatory Received any and had confirmatory Total § ¶ any chemotherapy who had Total § ¶ chemotherapy who had chemotherapy BC diagnoses chemotherapy BC diagnoses Age at patients confirmatory diagnoses patients confirmatory diagnoses Stage# diagnosis, y No. No. % of all patients No. % of all patients % No. No. % of all patients No. % of all patients % 0 65–74 12 563 28 0.2 13 0.1 46.4 2237 145 6.5 129 5.8 89.0 75–84 6996 12 0.2 <11 <0.2 — 1371 33 2.4 29 2.1 87.9 85þ 1364 0 0.0 0 0.0 — 317 <11 <3.5 <11 <3.5 I 65–74 27 904 2667 9.6 2623 9.4 98.4 3748 493 13.2 456 12.2 92.5 75–84 19 159 592 3.1 561 2.9 94.8 3023 149 4.9 126 4.2 84.6 85þ 5088 38 0.7 29 0.6 76.3 939 15 1.6 11 1.2 73.3 II 65–74 15 358 5425 35.3 5366 34.9 98.9 1585 516 32.6 487 30.7 94.4 75–84 11 250 1554 13.8 1522 13.5 97.9 1322 196 14.8 174 13.2 88.8 85þ 4548 96 2.1 91 2.0 94.8 542 21 3.9 16 3.0 76.2 III 65–74 4866 2634 54.1 2592 53.3 98.4 463 215 46.4 199 43.0 92.6 75–84 3568 1112 31.2 1094 30.7 98.4 350 107 30.6 100 28.6 93.5 85þ 1640 118 7.2 116 7.1 98.3 164 17 10.4 15 9.1 88.2 IV 65–74 2973 800 26.9 755 25.4 94.4 177 45 25.4 37 20.9 82.2 75–84 2442 406 16.6 369 15.1 90.9 170 25 14.7 22 12.9 88.0 85þ 1154 58 5.0 50 4.3 86.2 72 <11 <15.3 <11 <15.3 Total 120 873 15 540 12.9 15 187 12.6 97.7 16 480 1984 12.0 1806 11.0 91.0 *Chemotherapy agents identified from Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018 https://seer.cancer.gov/oncologytoolbox/canmed/hcpcs/. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Patients diagnosed with a first primary cancer (sequence 00, 01) who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diagnosis 0–6 months after index cancer diagnosis date. §Any Medicare National Claims History claim for chemotherapy within 6 months of index cancer diagnosis. ¶All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with BC. #Derived Adjusted American Joint Committee on Cancer Stage, 6th edition for the index BC. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |19 Table 4. Number and percent of second primary CRC and female BC* patients who received any chemotherapy by whether or not their chemo- therapy was associated with a confirmatory site-specific diagnosis code, years since first primary cancer diagnosis, and concordance of first and second primary tumor sites, SEER-Medicare. Number of years from first First primary and primary cancer diagnosis to second primary cancer Received second primary cancer diagnosis site are same site recommended chemo Received chemotherapy <5 years 5 years Same site Other sites and had confirmatory Total Yes No Agreement ‡ § § § § Cancer diagnoses No. No. % No. % No. % No. % No. No. % CRC Yes 1853 1326 71.6 527 28.4 659 35.6 1194 64.4 1838 15 88.4 No 340 260 76.5 80 23.5 17 5.0 323 95.0 239 101 Total 2193 1586 72.3 607 27.7 676 30.8 1517 69.2 2077 116 Female BC Yes 1806 1194 66.1 612 33.9 1299 71.9 507 28.1 1804 <11 >91.5% No 178 150 84.3 28 15.7 18 10.1 160 89.9 164 >11 Total 1984 1344 67.7 640 32.3 1317 66.4 667 33.6 1968 16 *Patients diagnosed with a second primary cancer (sequence 02) whose first primary cancer was not distant (IV) stage and who did not have a subsequent cancer diag- nosis 0–6 months after index cancer diagnosis date. Numbers less than 11 cases have been masked to protect patient confidentiality. BC ¼ breast cancer; CRC ¼ colon or rectal cancer; SEER ¼ Surveillance, Epidemiology, and End Results. Any Medicare National Claims History claim for chemotherapy as identified chemotherapy agents identified from the Cancer Medications Enquiry Database (CanMED)-Healthcare Common Procedure Coding System (HCPCS) version 2018. All Medicare National Claims History chemotherapy claims within 6 months of index cancer diagnosis had a claim diagnosis code that was consistent with CRC or fe- male BC, respectively. §Row percent. Sensitivity analyses among the 2193 CRC and 1984 BC We required confirmatory diagnoses on claims as a first step patients who had a prior cancer and were identified as having in creating a more automated method to attribute chemotherapy at least one chemotherapy claim within 6 months of their index to a specific cancer among patients with multiple cancers found in large population-based data. This definition resulted in a mea- cancer diagnosis were conducted. These analyses indicated that sure that was easy to execute (eg, only considered claim diagno- a longer time interval (<5 years vs 5 years) between cancer di- sis codes), yet we believe it can provide high confidence in agnoses was associated with being classified as receiving che- discerning if chemotherapy was administered for the index can- motherapy for the index cancer, per our confirmatory diagnosis cer. This definition was conservative because it included only definition, especially for female BC (CRC: 28.4% vs 23.5%; BC: patients for whom there was always agreement between chemo- 33.9% vs 15.7%; Table 4). Not surprisingly, having a prior concor- therapy claim diagnosis codes and the registry cancer site. This dant site cancer was also associated with being classified as re- definition can be refined depending on the research question. ceiving chemotherapy for the index cancer, per our Although we found high agreement when receipt of chemother- confirmatory diagnosis definition. Finally, we found high agree- apy was classified by agent, specific chemotherapy agents could ment between assigning chemotherapy to the index cancer us- be reviewed. If an agent is consistent with guidelines for the in- ing our confirmatory diagnosis definition and a definition based dex cancer then this could be considered evidence that the che- on site-specific guideline-recommended agents (CRC ¼ 88.4%; motherapy was administered for the index cancer, even if the female BC ¼ 91.5%). claim diagnosis codes were not always consistent with the regis- try data. The National Comprehensive Cancer Network clinical Discussion guidelines are an accessible source of chemotherapy drug and regimen information and could be used to build lists of recom- In this study, we estimated receipt of chemotherapy within mended cancer-specific drugs (24–26). However, this approach 6 months of index cancer diagnosis among patients with a prior would require additional consideration when an agent is recom- primary cancer diagnosis compared with patients without a mended for treatment for both the first and second cancers. prior cancer for two common cancer sites: CRC and female BC. Our approach for assigning chemotherapy to a cancer site dif- Overall, CRC and female BC patients with a prior primary cancer fers from two prior studies that included patients with multiple had only slightly lower chemotherapy use than patients with- primary cancers to study treatment patterns and outcomes. In out a prior cancer. Stratified by age and stage, CRC and female one study, patients diagnosed with multiple primary cancers BC patients with a prior primary cancer tended to have slightly were assigned to one type of cancer based on a hierarchical list higher chemotherapy use than patients without a prior cancer. of cancers (eg, if a patient’s claims data included diagnoses codes Additionally, more than 95% of CRC and female BC patients indicating more than one cancer type, all documented chemo- without a prior cancer who received chemotherapy had confir- therapy was assigned to the higher ranked cancer) (30). In an- matory cancer diagnoses listed on their chemotherapy claims. other analysis, chemotherapy was assigned to a specific cancer if For patients with a prior primary cancer who received chemo- at least 60% of the therapy claims included diagnosis codes indic- therapy, a lower percent had claims with confirmatory diagno- ative of only one cancer site (31). Neither study included cancer ses (85% CRC, 91% BC). The lower consistency between registry data linked to therapy claims as was done in our study. diagnoses listed in the registry and claim data among patients In addition to our chemotherapy definition being useful for with a prior cancer may be due to the chemotherapy being ad- assessing chemotherapy use for health services research, it ministered as treatment for a prior cancer. may be relevant to cancer registry functions. Registries could Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 20 | J Natl Cancer Inst Monogr, 2020, Vol. 2020, No. 55 apply the site-specific chemotherapy definition to automati- diagnosis, and it appears that more time between diagnosis in- cally determine receipt of chemotherapy and assign it to a given creased the likelihood of assigning the chemotherapy to the in- cancer. When there is always agreement between chemother- dex cancer. apy claims diagnosis codes and the registry cancer site, chemo- In conclusion, patients diagnosed with multiple primary therapy could automatically be reported as “yes” for the cancer. tumors have commonly been excluded from claims-based stud- Manual review of records could then be implemented for the ies of chemotherapy use due to the complexities in distinguish- remaining cases that may have greater complexity and thus did ing whether treatment was being used for the index or prior not always have diagnosis codes that matched the registry can- cancer. An Institute of Medicine Report on Delivery of High- cer site. Thereby, use of our chemotherapy definition could in- Quality Cancer Care in 2013 recommended as one of its 10 prior- crease cancer registry efficiency. ity areas to “Expand breadth of cancer research data collected We found that a very small number of in situ (stage 0) first so that they reflect the population with the disease, such as the CRCs and female BC received chemotherapy per our definitions. elderly and patients with comorbid conditions” (32). This study Given that chemotherapy is rarely administered for the treat- demonstrated a simple approach to determine for which cancer ment of in situ cancers, potential explanations are overtreat- chemotherapy is administered and thus can aid researchers ment, incorrect diagnosis listed on the claims, or possibly and cancer registrars to more accurately and efficiently deter- upstaging of the cancer based on later tests, which was not mine treatment intent. These findings suggest that requiring reflected in the registry data. confirmatory diagnoses can increase treatment assignment Our study demonstrated the use of claims data to estimate confidence. receipt of chemotherapy for patients with prior cancers. The confirmatory diagnosis codes on claims can be used in multiple Notes ways to expand and improve chemotherapy attribution at the cancer registries and in treatment pattern studies. However, we Author Affiliations: Data Analytics Branch, Surveillance did have some limitations that need to be considered. Some of Research Program, Division of Cancer Control and Population the diagnosis codes we used to determine whether the claim Sciences, National Cancer Institute, Rockville, MD (CL, AM); was for the treatment of cancer were generic codes (eg, encoun- Healthcare Assessment Research Branch, Healthcare Delivery ter for antineoplastic chemotherapy). Thus, there was no way to Research Program, Division of Cancer Control and Population tell whether a claim with these codes matched the index cancer. Sciences, National Cancer Institute, Bethesda, MD (LE, DW, JW); However, the majority of the chemotherapy claims assessed in- Information Management Services, Inc., Calverton, MD (TM) cluded confirmatory cancer codes (data not shown). There is no conflict of interest disclosures from any authors. Additionally, if the first primary cancer was the same cancer site as the second primary cancer, our chemotherapy definition will not be able to determine if the diagnosis codes on the che- References motherapy claims were truly for the index cancer. Furthermore, 1. Bluethmann SM, Mariotto AB, Rowland JH. Anticipating the “Silver with the shift to targeted therapies, which are used to treat mul- Tsunami”: prevalence trajectories and comorbidity burden among older can- cer survivors in the United States. Cancer Epidemiol Biomarkers Prev. 2016;25(7): tiple cancers, it will become increasingly challenging to identify 1029–1036. which cancer is the target of a specific treatment. Depending on 2. National Cancer Institute Surveillance, Epidemiology, and End Results the research question, these factors may or may not be impor- Program SEER 18 Regs Research Data, Nov 2017 Sub (1973-2015). National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance tant. We only investigated second cancers; however, the per- Systems Branch. http://www.seer.cancer.gov. Published 2017. Accessed cent of patients diagnosed with more than two cancers is small December 15, 2018. (4%). Assigning and interpreting treatment patterns for patients 3. Warren JL, Yabroff KR, Meekins A, Topor M, Lamont EB, Brown ML. Evaluation of trends in the cost of initial cancer treatment. J Natl Cancer Inst. with more than one prior cancer increases the complexity; how- 2008;100(12):888–897. ever, requiring all diagnosis codes to match the registry data 4. Bradley CJ, Yabroff KR, Warren JL, Zeruto C, Chawla N, Lamont EB. Trends in could still indicate high confidence that the chemotherapy was the treatment of metastatic colon and rectal cancer in elderly patients. Med administered for the index cancer. Furthermore, we did not in- Care. 2016;54(5):490–497. 5. Morton LM, Dores GM, Schonfeld SJ, et al. Association of chemotherapy for clude patients if they had a prior cancer that was distant stage solid tumors with development of therapy-related myelodysplastic syn- IV at diagnosis, because we assumed that any chemotherapy re- drome or acute myeloid leukemia in the modern era. JAMA Oncol. 2018;5(3): ceived after the index cancer diagnosis would likely be for the 318–325. doi: 10.1001/jamaoncol.2018.5625 6. Tong L, Ahn C, Symanski E, Lai D, Du XL. Effects of newly developed chemo- prior, advanced-stage cancer, and we would not have been able therapy regiments, comorbidities, chemotherapy-related toxicities on the to distinguish treatment between the two cancers. changing patterns of the leading causes of death in elderly patients with Confirmatory cancer diagnoses on the claims can be difficult CRC. Ann Oncol. 2014;25(6):1234–1242. 7. Gilden DM, Kubisiak JM, Pohl GM, et al. Treatment patterns and cost- to assess for oral chemotherapies, particularly for agents cov- effectiveness of first line treatment of advanced non-squamous non-small ered under Medicare Part D because the associated claims do cell lung cancer in Medicare patients. J Med Econ. 2017;20(2):151–161. not include diagnosis codes. If assessing oral agents is para- 8. Medeiros BC, Satram-Hoang S, Hurst D, Hoang KQ, Momin F, Reyes C. Big data mount for a given study, it is recommended that the specific analysis of treatment patterns and outcomes among elderly acute myeloid leukemia patients in the United States. Ann Hematol. 2015;94(7):1127–1138. agents be assessed to determine whether the regimen is consis- 9. Satram-Hoang S, Reyes C, Hoang KQ, Momin F, Skettino S. Treatment practice tent with treatment guidelines for a particular cancer. Lastly, in the elderly patient with chronic lymphocytic leukemia – analysis of the for patients with a prior cancer, time between prior and index combined SEER and Medicare database. Ann Hematol. 2014;93(8):1335–1344. 10. Shahinian VB, Yong-Fang K, Freeman JL, Orihuela E, Goodwin JS. Increasing cancer diagnoses was not considered in our definitions; the use of gonadotropin-releasing hormone agonists for the treatment of local- ideal temporal cut-point for assigning treatment to a given can- ized prostate carcinoma. Cancer. 2005;103(8):1615–1624. cer could vary by multiple variables (eg, cancer site, stage, and 11. Obeidat NA, Pradel FG, Zuckerman IH, DeLisle S, Mullins CD. Outcomes of irinotecan-based chemotherapy regimens in elderly Medicare patients with recurrence rate). However, our sensitivity analyses by number metastatic CRC. Am J Geriatr Pharmacother. 2009;7(6):343–354. of years between prior and index cancer did show that for both 12. Rajan SS, Stearns SC, Lyman GH, Carpenter WR. Effect of primary prophylac- CRC and female BC, the majority of the second primary cancers tic G-CSF use on chemotherapy administration for elderly BC patients. Breast were diagnosed less than 5 years after the first primary cancer Cancer Res Treat. 2011;130(1):255–266. Downloaded from https://academic.oup.com/jncimono/article/2020/55/14/5837285 by DeepDyve user on 13 July 2022 C. J. K. Lam et al. |21 13. Griffiths RI, Barron RL, Gleeson ML, et al. Granulocyte-colony stimulating fac- 23. American Cancer Society. Treatment of colon cancer, by stage. 2019. https:// tor use and medical costs after initial adjuvant chemotherapy in older www.cancer.org/cancer/colon-rectal-cancer/treating/by-stage-colon.html. patients with early-stage BC. Pharmacoeconomics. 2012;30(2):103–118. Accessed December 15, 2019. 14. Garcia C, Yao A, Camacho F, Balkrishnan R, Cantrell LA. A SEER-Medicare 24. National Comprehensive Cancer Network. NCCN guidelines for treatment of analysis of the impact of metformin on overall survival in ovarian cancer. cancer by site: BC. https://www.nccn.org/professionals/physician_gls/. Gynecol Oncol. 2017;146(2):346–350. Accessed April 18, 2019. 15. Yu D, Gao P, Song Y, et al. The differences on efficacy of oxaliplatin in locally 25. National Comprehensive Cancer Network. NCCN guidelines for treatment of advanced colon cancer between mucinous and nonmucinous adenocarci- cancer by site: colon cancer. https://www.nccn.org/professionals/physician_ noma. Cancer Med. 2018;7(3):600–615. gls/. Published January 15, 2019. Accessed April 18, 2019. 16. Sharma M, Holmes HM, Mehta HB, et al. The concomitant use of tyrosine ki- 26. National Comprehensive Cancer Network. NCCN guidelines for treatment of nase inhibitors and proton pump inhibitors: prevalence predictors, and im- cancer by site: rectal cancer. https://www.nccn.org/professionals/physician_ pact on survival and discontinuation of therapy in older adults with cancer. gls/. Published January 15, 2019. Accessed April 18, 2019. Cancer. 2019;125(7):1155–1162. 27. Derived AJCC Cancer Staging. National Cancer Institute. Surveillance, 17. Rocque GB, Kandhare PG, Williams CP, et al. Visualization of sequential treat- Epidemiology, and End Results Program, 2019. seer.cancer.gov/seerstat/ ments in metastatic BC. J Clin Oncol Clin Cancer Inform. 2019;3:1–8. variables/ajcc-stage/derived-ajcc.html. Accessed July 8, 2019. 18. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the 28. Howlader N, Ward KC, Warren JL, Campbell DS, Coyle L, Mariotto AB. SEER-Medicare data: content, research applications, and generalizability Assessment of oncology practice billing claims for capturing chemotherapy: to the United States elderly population. Med Care. 2002;40(8 suppl): a pilot study in Georgia SEER Cancer Registry. J Natl Cancer Inst Monogr. 2020; IV3–IV18. 2020(55):82–88. 19. Registry Groupings in SEER Data and Statistics. http://seer.cancer.gov/regis 29. Rivera DR, Lam CJK, Enewold L, et al. Development and utility of the observa- tries/terms.html. Accessed June 15, 2019. tional research in oncology toolbox: Cancer Medications Enquiry Database 20. 2016 Medicare Enrollment Section. https://www.cms.gov/Research- (CanMED) Healthcare Common Procedure Coding System (HCPCS). J Natl Statistics-Data-and-Systems/Statistics-Trends-and-Reports/ Cancer Inst Monogr. 2020;2020:(55):39–45. CMSProgramStatistics/2016/2016_Enrollment.html. Published March 15, 30. Hayes J, Hoverman JR, Brow ME, et al. Cost differential by site of service for can- 2017. Accessed June 15, 2019. cer patients receiving chemotherapy. Am J Manag Care. 2015;21(3):e189–e196. 21. American Cancer Society. Cancer Facts & Figures 2019. Atlanta, GA: American 31. Avalere Health, LLC. Total cost of cancer care by site of service: physician of- Cancer Society; 2019. fice vs outpatient hospital. 2012. 22. American Cancer Society. Treatment of BC by stage. 2019. https://www.can 32. Ganz PA. Institute of Medicine report on delivery of high-quality cancer care. cer.org/cancer/breast-cancer/treatment/treatment-of-breast-cancer-by- stage.html. Accessed December 15, 2019. J Oncol Pract. 2014;10(3):193–195.

Journal

JNCI Monographs – Oxford University Press

Published: May 1, 2020

Keywords: cancer; chemotherapy regimen; medicare; cancer diagnosis; seer program

There are no references for this article.