Evaluation of the duplication of staging CT scans for localized colon cancer in Medicare population

Abstract

Background

To quantify and characterize duplicated tests done during the staging of localized colon cancer in Medicare population.

Methods

We used the SEER-Medicare linked database to select patients diagnosed with localized colon cancer the years 1996–2009. We considered a patient as adequately staged after having received a colonoscopy, an abdominal CT scan, and a pelvic CT scan. Abdominal and pelvic CT scans performed between complete staging and first cancer-directed treatment, if not ordered due to an acute condition, were considered duplicates. We characterized the institutions providing the tests and evaluated the association with survival using a weighted pooled logistic regression adjusted by baseline and time-varying confounders.

Results

Of 36,291 patients with a complete staging, 2,680 (7.4%) had at least one duplicated test. Patients receiving a duplicate had a higher comorbidity score, were more symptomatic, and had more visits to the emergency department and clinical evaluations. They also were treated with surgery less frequently and had worse survival (HR 1.22, 95% CI 1.16-1.28). The type of institution involved in the staging (non-profit/government centers, proprietary centers, free-standing facilities) was not associated with receiving duplicated tests.

Conclusion

We found a low frequency of duplicated abdominal or pelvic CT scans in the staging of colon cancer in Medicare population.

Introduction

Every year more than 100,000 cases of colon cancer are diagnosed in the United States¹. Appropriate staging of these tumors is necessary for informed therapeutic decisions. Clinical staging is intended to detect metastatic disease that rules out the ability to perform curative intent surgical resection of tumor. Clinical guidelines for the diagnosis and staging of colon cancer recommend the use of colonoscopy, abdominal and pelvic CT scans^2–7, and pathologic examination of the surgical specimen for localized tumors.

Health care costs in the United States are projected to account for 20% of the gross domestic product in 2020⁸. A key measure to cut down costs is the avoidance of services that do not benefit patients^{9, 10}. The “Choosing Wisely” campaign explicitly points at the elimination of duplicated tests as a benefit of promoting conversations between physicians and patients ¹¹. Thus, avoidance of unnecessary tests for the diagnosis and staging of colon cancer might be a potential target for cost-containment measures.

Medicare patients receive coverage for all tests required for diagnosis and staging of colon cancer. While imposing no restrictions on the number of test covered, Medicare encourages patients to avoid unnecessary duplication of tests¹². The proportion of duplicated tests is, however, unknown. If diagnostic workup includes duplicative workup, there is a potential strategy for improving care quality while also controlling health care costs. Here we quantify and characterize the frequency of duplicated tests performed in the fee for service Medicare population during the clinical staging of early stage colon cancer.

Methods

Study population

The study cohort was identified from the SEER-Medicare data, which is a linkage of patient demographic and tumor-specific variables collected by 17 SEER cancer registries across 12 states with Medicare claim files from the Centers for Medicare and Medicaid Services¹³. SEER data are summarized in the Patient Entitlement and Diagnosis Summary File (PEDSF), which is linked with 100% of Medicare claims. For the current study we used Medicare claims from the Inpatient, Outpatient, Home Health Agency, Durable Medical Equipment (DME), Medpar and National Claims History (NCH) files. Provider characteristics are extracted from the Hospital file, which contains information on hospital characteristics for years 1996, 1998 and 2000 to 2009.

Our analysis includes patients 66 years (to allow for at least one year of claims before diagnosis) or older, with histological diagnosis of invasive colon adenocarcinoma between 1996 (when hospital information first became available) and 2009 in a SEER area. We excluded rectal cancer and rectosigmoid tumors (which may require additional staging like MRI or endoscopic ultrasound) and cancers for which the reporting source was nursing home/hospice, autopsy or death certificate. To ensure complete ascertainment of health services, patients had to be enrolled in parts A and B and not in an HMO during the six months before and after diagnosis. We excluded patients diagnosed in Louisiana in 2005 because of the disruption of data collection following hurricane Katrina.

We considered a patient as adequately staged and ready for a therapeutic decision after having received a colonoscopy, an abdominal CT scan, and a pelvic CT scan, as prescribed by the National Comprehensive Cancer Network and European Society for Medical Oncology^2–7. We did not require a chest CT scan which is considered by some guidelines ⁶, but not others^{7, 14} (see Table, Supplemental Digital Content 1, for codes used to identify these tests). Tests are extracted from the claims six months before and after SEER date of diagnosis.

Definition of duplicated test

Any abdominal CT scan or pelvic CT scan received between the date when the patient was completely staged (see above) and the date of first treatment was considered a duplicate, with the exception of scans performed because of acute conditions ¹⁵(see Table, Supplemental Digital Content 2, for the list of conditions and codes). Treatment of colon cancer was defined as colon surgery, radiotherapy, or chemotherapy (see Table, Supplemental Digital Content 1 for codes used to identify these treatments). Tests performed beyond 90 days of complete staging were not considered duplicates under the assumption that restaging might be appropriate if a patient has not been treated within 90 days.

Covariates

Demographic characteristics (age, sex, race, marital status, urbanicity), tumor features (TNM stage, grade of tumor differentiation, date of diagnosis), and census tract features (census region, percentage of black population, percentage of residents living below the poverty level, percentage of residents aged 25 or older with less than 12 years of education, percentage of residents speaking English not well/not at all at age 65+, median income) were extracted from the PEDSF file. Comorbidities were summarized using the Deyo-Charlson-Klabunde comorbidity index¹⁶, derived from the inpatient and outpatient Medicare claims for the period between 12 months to 1 month before diagnosis. To assess health services utilization, we computed a “preventive score” ¹⁷, the number of “low complexity visits” in the 24 months one year before diagnosis, and emergency room visits.

The provider performing the tests was linked with the institution information on the Hospital file. The Outsaf and Medpar files, but not the NCH file, contain a variable that allows linkage of providers with institutions without identifiers. NCH claims can correspond to either a test performed by a free-standing facility or to a professional service performed at an institutional provider (and thus also recorded in the Outsaf or Medpar files). We thus classified patients according to the type of institution involved in their staging work-up: all tests performed in institutional non-profit/government centers, at least one test in a proprietary center, and all tests in free-standing facilities or free-standing facilities plus non-profit/government centers (see table in Supplemental Digital Content 1 for the codes used to extract this information).

Mortality Analysis

For each patient, follow-up started at complete staging (see above) and ended at date of death or administrative cutoff date (in PEDSF file, 12/31/2010), whichever occurred earlier. We estimated the mortality hazard ratio (HR) for “receiving at least one duplicated test”versus “not receiving any duplicated test”within 3 months of complete staging. To do so, we fit a weighted pooled logistic model that included an indicator for duplicated tests, a flexible function of time (restricted cubic splines to estimate the baseline hazard) and the baseline covariates described above. We calculated robust standard errors to compute conservative 95% confidence intervals for the effect estimate.

As in previous analyses of exposures that are not fully determined at baseline, we used data replication, censoring, and inverse probability weighting ^18,19 to adjust for the time-varying covariates: visits to the emergency room, clinical evaluations, change in comorbidity index and development of large bowel obstruction. We then stabilized the weights to emulate a uniform duplicated test administration during three months²⁰. Like previous applications of inverse probability weighting ^21–23, we truncated weights at percentile 99. All analyses were conducted with SAS, version 9.3 (SAS Institute, Cary, North Carolina).

Results

Of 75,840 eligible patients, 36,291 had complete staging: 25% stage I, 43% stage II and 33% stage III (Figure 1). We found that 2,680 (7.4%) of patients had at least one duplicated CT scan. Of the 2,680 patients with duplicated tests, 68% received one duplicated abdominal CT scan plus one duplicated pelvic CT scan, and only 8% received more than two duplicated tests (see Table, Supplemental Digital Content 3). After complete staging, a colonoscopy was repeated in 5.5% of the patients.

Figure 1. Flowchart of colon cancer patients.

¹Time frame for diagnostic tests: −180 to +180 days/first treatment since diagnosis. See Methods.

²4,795patients presented a diagnosis of large bowel obstruction and 1,199presented a diagnosis of large bowel perforation, both contraindications for colonoscopy.

Table 1 shows the baseline characteristics of the patients. Patients receiving a duplicated CT scan had a higher comorbidity score; lived in census areas with a higher percentage of high school drop-outs, residents below poverty line, black race/ethnicity, and lower median incomes; and were more likely to have anemia, asthenia, and gastrointestinal symptoms in the six months before diagnosis.

Table 1.

Baseline characteristics of patients with early colon cancer 66 years or older at diagnosis, enrolled in part A and B of Medicare, not in a HMO and with a complete staging(N=36,291).

	Not receiving any duplicate scan(N=33,611)	Receiving at least one duplicate scan(N=2,680)
Sentinel symptom(%)
Anemia	26,007(77)	2,202(82)
Gastrointestinal symptoms1	15,194(45)	1,540(57)
Large bowel obstruction	6,993(21)	794(30)
Abnormal weight loss	5,775(17)	629(23)
Asthenia	14,015(42)	1,391(52)
Comorbidity score(%)
0	17,645(53)	1,117(42)
1	8,514(25)	717(27)
2+	6,121(18)	743(28)
Unknown	1,331(4)	103(4)
Median age when staged(range)	78.4(65.9–106.3)	78.6(65.8–99.2)
Female(%)	19,620(58)	1,564(58)
Race(%)
Caucasian NOS	27,947(83)	2,171(81)
Caucasian, Spanish origin or surname	1,384(4)	117(4)
African American	2,590(8)	250(9)
Asian/Pacific islander	1,526(5)	131(5)
Other/Unknown/unspecified	164(0)	11(0)
Stage(%)
I	8,623(25)	671(25)
II	14,449(43)	1,088(41)
III	10,899(32)	921(34)
Grade of differentiation(%)
Well differentiated	2,711(8)	201(8)
Moderately differentiated	22,841(68)	1,749(65)
Poorly differentiated	6,907(21)	589(22)
Unknown	1,152(3)	141(5)
Median number of low complexity visits(Q1–Q3)2	7(2–13)	7(3–15)
Median preventive score(Q1–Q3)3	2(1–3)	2(1–3)
Urbanicity(%)
Big Metro (≥1 million population)	19,361(58)	1,595(60)
Metro (250,000 to 1 million)	9,135(27)	650(24)
Urban (20,000 to 250,000)	1,821(5)	127(5)
Less Urban (2,500 to 20,000)	2,672(8)	246(9)
Rural (rural or < 2,500 population)	622(2)	62(2)
Marital Status(%)
Single	2,593(8)	219(8)
Married	16,168(48)	1,176(44)
Separated/divorced	2,059(6)	173(6)
Widowed	11,446(34)	1,021(38)
Unknown	1,345(4)	101(4)
SEER registry census region(%)
West	11,309(34)	867(32)
Northeast	9,897(29)	809(30)
Midwest	5,203(15)	395(15)
South	6,741(20)	588(22)
Pacific	461(1)	21(1)
Year of diagnosis(%)
1996–2000	5,580(17)	318(12)
2001–2005	15,812(47)	1,184(44)
2006–2009	12,219(36)	1,178(44)
Census tract features4[median(Q1–Q3)]
% did not complete high school	15.7(9.4–25.5)	17.1(10.4–27.5)
% below poverty line	7.6(4.1–14.3)	8.2(4.4–15.9)
% black race/ethnicity	2.1(0.6–7.7)	2.4(0.7–9.5)
% English not well/at all at 65+	1.6(0–5.5)	1.5(0–5.8)
Median income(USD)	46,163(34,742–61,152)	44,712(33,099–60,007)

¹Gastrointestinal symptoms include abdominal distention, change in bowel habit, constipation, irritable bowel syndrome, diarrhea, obstruction, anemia, abnormal weight loss, asthenia.

²Low complexity visits(as defined by CPT codes, see appendix Table A1) during the years -2 and -3 of diagnosis.

³See methods section for details.

⁴Census tract features are missing for 218 individuals

Patients with duplicated CT scans had more clinical evaluations and were more likely to visit the emergency department in the time span from being completely staged to first treatment (Table 2). Patients with duplicated CT scans also had a longer median time from staging to first treatment (17 days, interquartile range [IQR] from 7 to 35 versus9days, IQR from 3 to 20). First treatment received was surgery in 89% and 96% of the patients with and without duplicates, respectively. The use of chemotherapy or radiotherapy as first treatment was marginal (Table 2).

Table 2.

Cancer treatment-related interventions and outcomes by duplicates.

	Not receiving any duplicate scan (N=33,611)	Receiving at least one duplicate scan (N=2,680)
Median(Q1–Q3) time from complete staging to first duplicate, days	N/A	4(1–14)
Median(Q1–Q3) time from complete staging to first treatment, days	9(3–20)	17(7–35)
Patients with a clinical evaluation between complete staging and first treatment1(%)	26,564(79)	1,930(72)
Median(Q1–Q3) number of evaluations	2(2–4)	4(2–6)
Patients visiting the emergency department between complete staging and first treatment(%)	3,587(11)	788(29)
Median(Q1–Q3) number of visits	1(1–1)	1(1–1)
First treatment(%)
Surgery	32,190(96)	2,383(89)
Chemotherapy	78(0)	32(1)
Radiotherapy	99(0)	21(1)
No cancer-specific therapy	1,244(4)	244(9)
Months of follow-up	1,885,882	119,649
Number of deaths(%)	18,296(54)	1,697(63)
Number of colon cancer deaths(%)	6,596(20)	673(25)
Adjusted rate ratio for all cause mortality	Reference	1.22(1.16–1.28)
Adjusted rate ratio for colon cancer mortality	Reference	1.23(1.14–1.32)

¹Clinical evaluation consists on any of the following: new outpatient, established outpatient, hospital observation services, new inpatient, established inpatient, observation/inpatient care services, outpatient consultation, inpatient consultation, follow-up inpatient consultation, confirmatory consultation, nursing facility services and team conference.

Fifty percent of patients received complete staging in non-profit/government centers, 8% received at least one staging test in a proprietary center, and 42% received staging tests in free-standing facilities with or without tests in institutional non-profit/government centers. The percentage of patients receiving duplicates was 6%, 9% and 8% in these three groups, respectively.

The all-cause mortality HR for having received a duplicated CT scan was 1.22 (95% CI 1.16-1.28). The corresponding HR for colon cancer-specific mortality was 1.23 (95% CI 1.14-1.32) (see Table, Supplemental Digital Content 4, for more details on the survival analysis).

Discussion

We found that 7%of abdominal or pelvic CT scans were duplicated in the staging of localized colon cancer in Medicare patients. Compared with patients without duplicated CT scans, those with duplicates had a higher comorbidity index, were more symptomatic, visited the emergency room more often, and received surgery as first treatment less often. These findings suggest that patients receiving duplicate tests were more frail and complex, which may warrant the additional testing.

The higher mortality among patients receiving duplicate CT scans also suggests that a the duplicates may often be clinically indicated for reasons not captured in the Medicare data, such as performance status and abnormal test results. This explanation is further supported by the attenuation of the mortality HR after adjusting for baseline and time-varying confounders, together with the smaller attenuation observed for cancer-specific mortality(see Table, Supplemental Digital Content 4, for more details on the survival analysis).

The short time span from complete staging to the first treatment (median 9 days) indicates a timely administration of treatment to patients with localized colon cancer. Though patients receiving duplicates are treated a few days later on average, it is unlikely that this delay can explain the association between duplicated CT scans and mortality.

The cost of cancer care is estimated to grow from $125 billion in 2010 to $173 billion in 2020 in the US ²⁴. Aging of the US population is argued as one of the drivers of this cost increase ²⁵ and, in the case of colorectal cancer, the 12 months following diagnosis account for most of the expenses ²⁴. Our analysis targeted elderly population in the initial phase of colorectal cancer diagnosis, and provided reassurance of an adequate use of Medicare resources in this population.

Our analysis has the data limitations inherent to claim-based analyses and is restricted to patients over 66 years residing in SEER states. There is a possibility of occasional coding of rectal cancer as colon cancer, or vice versa. However, the small proportion of radiotherapy as first therapy suggest this potential miscoding would have been infrequent. Some diagnostic tests may have been missed if some patients were using health care providers outside Medicare. However, when we restricted the analysis to the 27,158 individuals with an evaluation for a colon cancer-related symptom in the six months before diagnosis (i.e., those more likely to have been diagnosed and staged within Medicare), results did not change materially.

In summary, we found a 7% frequency of duplicated CT scans for disease staging, which may be partly explained by the higher complexity of these patients, and timely delivery of treatment among elderly Medicare patients with localized colon cancer.