Projecting total costs and health consequences of increasing mt-sDNA utilization for colorectal cancer screening from the payer and integrated delivery network perspectives

Abstract

Aims: To evaluate total costs and health consequences of a colorectal cancer (CRC) screening program with colonoscopy, fecal immunochemical tests (FIT), and expanded use of multitarget stool DNA (mt-sDNA) from the perspectives of Integrated Delivery Networks (IDNs) and payers in the United States.

Materials and methods: We developed a budget impact and cost-consequence model that simulates CRC screening for eligible 50- to 75-year-old adults. A status quo scenario and an increased mt-sDNA scenario were modeled. The status quo includes the current screening mix of colonoscopy (83%), FIT (11%), and mt-sDNA (6%) modalities. The increased mt-sDNA scenario increases mt-sDNA utilization to 28% over 10 years. Costs for both the IDN and the payer perspectives incorporated diagnostic and surveillance colonoscopies, adverse events (AEs), and CRC treatment. The IDN perspective included screening program costs, composed of direct nonmedical (e.g. patient navigation) and indirect (e.g. administration) costs. It was assumed that IDNs do not incur the costs for stool-based screening tests or bowel preparation for colonoscopies.

Results: In a population of one million covered lives, the 10-year incremental cost savings incurred by increasing mt-sDNA utilization was $19.6 M for the IDN and $4.4 M for the payer. The incremental savings per-person-per-month were $0.16 and $0.04 for the IDN and payer, respectively. For both perspectives, increased diagnostic colonoscopy costs were offset by reductions in screening colonoscopies, surveillance colonoscopies, and AEs. Extending screening eligibility to 45- to 75-year-olds slightly decreased the overall cost savings.

Limitations: The natural history of CRC was not simulated; however, many of the utilized parameters were extracted from highly vetted natural history models or published literature. Direct nonmedical and indirect costs for CRC screening programs are applied on a per-person-per modality basis, whereas in reality some of these costs may be fixed.

Conclusions: Increased mt-sDNA utilization leads to fewer colonoscopies, less AEs, and lower overall costs for both IDNs and payers, reducing overall screening program costs and increasing the number of cancers detected while maintaining screening adherence rates over 10 years.

Keywords:

• Colorectal cancer screening
• mt-sDNA
• FIT
• colonoscopy

JEL CLASSIFICATION CODES:

• B23
• B41
• C01
• C51

View correction statement:

Correction

Introduction

Colorectal cancer (CRC) is the second leading cause of cancer-related deaths in the United States¹. There is significant evidence that screening for CRC reduces mortality^2–14 and that most cancer deaths are preventable by screening and early detection⁶. The U.S. Preventive Services Task Force (USPSTF) recommendations suggest that there is substantial screening benefit for adults at average risk for CRC, with either a high-sensitivity stool-based test or a direct visualization examination of the colon and rectum². The USPSTF recommends that individuals 50–75 years of age, at average risk for CRC be screened, and the American Cancer Society (ACS) recommends that screening begins at the age of 45 years (qualified recommendation)⁸ however, the risks and consequences of different screening methods vary².

Colorectal cancer screening is an on-going public health challenge and opportunity. There are multiple screening options with different screening intervals and variable levels of evidence supporting cost-effectiveness, performance, risk of harm, patient preference, and benefit for the prevention and early detection of CRC. The USPSTF and ACS prioritize screening uptake and adherence over the endorsement of a particular screening modality^2,8. There are no head-to-head high quality data that compare the performance, effectiveness, or patient preference for colonoscopy, fecal immunochemical tests (FIT), fecal occult blood tests (FOBT), and multitarget stool DNA (mt-sDNA, [Cologuard^®]). Current CRC screening options included in guidelines can be categorically described as structural examinations (colonoscopy, CT colonography, flexible sigmoidoscopy), and stool-based tests (FIT, FOBT, mt-sDNA).

Through shared decision-making discussions with providers, patients may choose from screening options with varying degrees of preparation, time requirements, and invasiveness. For example, colonoscopy requires sedation, full bowel preparation, and can lead to adverse events (AEs) such as gastrointestinal bleeding or bowel perforation^15–18. However, colonoscopy screening identifies CRC, and prevents CRC through detection and removal of precancerous polyps. In contrast, noninvasive screening modalities such as mt-sDNA and FIT require neither bowel preparation nor sedation. Stool-based screening tests have varying performance characteristics (sensitivity and specificity) for the detection of CRC and precancerous lesions¹⁹. Positive stool-based screening tests should be followed by diagnostic colonoscopy and as appropriate, therapeutic polyp removal, and/or cancer diagnosis and treatment.

mt-sDNA is a molecular diagnostic test composed of 11 biomarkers, which include 10 DNA markers and a marker for hemoglobin. The DNA markers represent acquired changes in DNA (methylations, mutations) that are associated with colorectal neoplasia (precancer, cancer). The 11 biomarkers are analyzed and the results are combined in a logistic regression algorithm to yield a ‘positive’ or a ‘negative’ result. A head-to-head cross-sectional clinical study of FIT and mt-sDNA found that the mt-sDNA test has a sensitivity of 92% and specificity of 87%, whereas a leading commercially available FIT test has a sensitivity of 74% and specificity of 95% for CRC¹⁹. The specificities of annual FIT and every 3°years mt-sDNA are theoretically similar²⁰ although these have not been evaluated in a study.

Modeling studies have estimated the clinical and economic outcomes of colorectal screening modalities^3–7. The majority of these studies assess the cost-effectiveness of screening from the societal perspective over a screening lifetime. Several are natural history models, including the Cancer Intervention and Surveillance Modeling Network (CISNET) models that simulate adenoma advancement into cancer and typically assume ideal (100%) adherence to screening regimens²¹. Although these models are informative, there is a complementary need to quantify the total costs and health consequences of screening alternatives from the perspectives of Integrated Delivery Networks (IDNs) and payers. Integrated Delivery Networks consist of integrated provider networks, such as hospitals, urgent care centers, and laboratories. A shorter time horizon versus a screening lifetime should be simulated to reflect the frequent migration of individuals among IDNs and health plans. Furthermore, real-world applications of economic analyses should reflect imperfect adherence to screening, as well as potential shifting of patient choice of screening modalities over time.

Attempts to overcome the limitations to practical application of economic evaluations have occurred. Frazier et al.³ and Vijan et al.⁷ performed cost-effectiveness analyses with imperfect cross-sectional and repeat screening compliance. Subramanian et al.⁶ incorporated patient preference for screening tests by assigning proportions of the modeled population to various modalities. Although these economic models address some of the limitations described above, they still focus on a societal perspective and do not address comparative costs of screening from the perspectives of IDNs and payers.

The objectives of this analysis were to examine costs from the perspectives of (1) IDNs and (2) payers, as well as to assess health consequences, such as adherence to CRC screening guidelines, number of detected CRC cases, and number of AEs.

Methods

Model overview

A 10-year Markov cohort model with annual cycles was developed to assess the costs and consequences of CRC screening with different modalities at varying utilizations and compliance levels. One million adults enter the model in the IDN and payer simulations. One million members is a common US payer size and is often used in health economic evaluations^22–24 because it is easily scalable to an individual payer’s or IDN’s population. Based on the calculations from US Census data²⁵ 38% of covered lives were eligible for screening based on age (50–75 years), and an assumed 80% of those were considered to be at average risk for CRC, a calculation based on familial predisposition²⁶. The study included an average risk population only because recommended screening timing and frequency differs between average and high risk populations. The USPSTF and ACS recommendations apply to average risk patients only^2,27. The U.S. Multi-Society Task Force of Colorectal Cancer (MSTF), however, provides recommendations for high risk individuals²⁰. Depending on various individual risk factors, the MSTF guidelines recommend screening with colonoscopy only, beginning at the age of 40 years or younger and at various intervals spanning from 3 to 5 years²⁰. With an average risk population, the simulated screening-eligible population for each perspective was 302,000 adults.

The screening algorithm was based on the 2016 USPSTF CRC screening recommendations². Eligible individuals in the model are screened with mt-sDNA, FIT, or colonoscopy. Flexible sigmoidoscopy and CT colonography are not commonly utilized in the US²⁷ and were therefore not included. If the test is negative, mt-sDNA screeners enter a tunnel (nonscreening) state and are eligible for screening again in 3°years. Negative FIT screeners are eligible for rescreening annually, and colonoscopy screeners with negative results enter a tunnel state for the remainder of the time horizon, as guidelines recommend rescreening at 10 years².

Individuals with positive stool-based screening results are referred to diagnostic colonoscopy. If a polyp or colorectal neoplasia is detected during a diagnostic or screening colonoscopy, a biopsy is performed and the pathology is determined to be either a non-neoplastic finding, a nonadvanced adenoma, an advanced adenoma, or cancer (Figure 1). Post-polypectomy surveillance is based on polyp categorization and the 2012 American Gastroenterological Association (AGA) guidelines²⁸. If an advanced adenoma is detected, the individual is referred for a surveillance colonoscopy in 3 or 5 years depending on adenoma size, and histology²⁸. Individuals with a negative diagnostic or surveillance colonoscopy enter a tunnel state for the remainder of the model horizon. The model allows mt-sDNA and FIT screeners to change screening modalities during each eligible cycle over the 10-year period. Furthermore, a proportion (16%) of all colonoscopies were required to be repeated due to poor bowel preparation²⁹ and AE rates due to colonoscopy were assumed to be based on whether or not a polypectomy is performed^18,30–33 (Table 1).

Figure 1. Conceptual model framework.

Table 1. Clinical inputs.

(a) Screening compliance.
Screening modality	First-round screening compliance (%)	Repeat modality screening tests (%)^a	Diagnostic colonoscopy compliance for positives (%)
mt-sDNA	67.5^34	81.6^b	78.4
FIT	64.7^35	81.6^36^,^c	78.4^36
CY	64.0^37	64.0^d	N/A
(b) CRC epidemiology^e.
Age band (years)	Baseline CRC incidence per 100,000		Resulting incidence per 100,000 per modality (assuming 100% compliance)^f
	No screening^5	mt-sDNA	FIT	Colonoscopy
50–54	51.4	19.0	16.9	9.8
55–59	87.6	32.4	28.9	16.7
60–64	137.4	50.8	45.3	26.1
65–69	202.6	74.9	66.8	38.5
70–75	291.4	103.3	92.1	53.1
(c) Sensitivity^g.
Age band (years)	mt-sDNA^19^,^38			FIT^19			CY^5			Diagnostic CY^5,^h
	CRC	Adv adn	Nonadv adn	CRC	Adv adn	Non adv adn	CRC	Adv adn	Non adv adn	CRC	Adv adn^5	Non adv adn
50–54	92.3%	38.0%	17.2%	73.8%	23.8%	7.6%	95.0%	95.0%	80.0%	95.0%	95.0%	80.0%
55–59		38.0%
60–64		42.1%
65–69		41.5%
70–75		46.8%
(d) Specificity.
Age band (years)	mt-sDNA^i (%)^38	FIT (%)^19	CY (%)^5	Diagnostic CY (%)^j
50–54	92.2	94.9	86.0	86.0
55–59	92.2
60–64	89.0
65–69	85.7
70–75	81.7
(e) Adverse event rates due to colonoscopy^18.
Event	CY without polypectomy (%)^k	CY with polypectomy (%)
Perforation	0.06	0.07
GI bleeding	0.21	0.87
MI	0.24	0.52
Arrhythmia	0.56	1.25
Cardiac arrest	0.15	0.25

Abbreviations. mt-sDNA, multi-target stool DNA; FIT, Fecal Immunochemical Test; CY, colonoscopy; CRC, colorectal cancer; Adv adn, advanced adenoma; and, Nonadv, nonadvanced adenoma.

^a If persons repeated screening with the same modality, repeat screening rates were applied; however, if persons switched modalities, the primary screening compliance was applied.

^b Assumed same as FIT.

^c The FIT repeat screening estimate was calculated as the average of Rounds 2–4 from Table 2 in Jensen.

^d Assumption that it is the same as first round.

^e For illustrative purposes only, the modeled population size and adjusted compliance incidences.

^f Reductions from the SimCRC model reported in Knudsen supplement are applied to the No screening incidence. These reductions (Colonoscopy: 81%, FIT: 67%, mt-sDNA: 63%) assume 100% compliance and occur when ages of 50–74 years are screened.

^g Because the adenoma estimates come from cadaver data, it is assumed to be the underlying truth; therefore, sensitivities of advanced and nonadvanced adenomas per modality are applied to the calculated incidences in order to determine the number of them detected. Because the calculated CRC incidences are not the underlying truth to which modality CRC sensitivity measures are typically applied, CRC sensitivity estimates are applied only to these incidences in order to determine the number of all patients who require biopsies.

^h Assumed to be the same as a screening colonoscopy.

ⁱ Age-adjusted specificity for mt-sDNA and is related to increased shedding of methylated DNA in stool [38].

^j Assumed to be the same as screening colonoscopy from Knudsen et al. [5].

^k Calculation based on Warren et al. [18] and the simplifying assumption that AE rates of screening colonoscopies map to colonoscopies without polypectomies.

Population

We assigned age-specific risk profiles for incidence and prevalence of CRC, adenomas, and death, by dividing the model population into 5-year age bands (See Supplementary Table S1 for age distributions) at the beginning of each cycle. Annual all-cause mortality was based on the US life tables within 5-year age bands. In each cycle, 20% of each surviving subpopulation age into the subsequent age group; those above 75 years age out of the model. Beginning at the second cycle, ∼12,600 49-year-olds age into the screening eligible population as 50-year-olds each year.

As in the previous evaluations, propensity to adhere to screening was included in the specification of health states^6,47,48. The ‘never-screener’ cohort comprises individuals who remain unscreened throughout the time horizon, and the ‘potential screener’ cohort comprises individuals who have a probability of adhering to recommended screening guidelines over 10 years. Based on a 10-year longitudinal retrospective claims analysis of over 150,000 subjects, 24% of patients remained unscreened throughout the time horizon³⁷. Our approach, reflecting adherence heterogeneity, prevented the entire cohort from eventually being screened.

Utilization of screening modalities in the primary analysis

The primary evaluation was between a status quo scenario and an ‘increased mt-sDNA’ utilization scenario. In the status quo scenario, we assumed that the screening mix remained constant throughout the time horizon as follows: 83% colonoscopy, 11% FIT, and 6% mt-sDNA, based on publicly available data on current utilization of screening modalities in the United States (Supplementary Table S2). In the ‘increased mt-sDNA’ scenario, the proportion utilizing mt-sDNA increased to 28% and the proportions utilizing FIT and colonoscopy decreased to 9 and 63% over 10 years.

Primary cross-sectional compliance, repeat screening compliance, and adherence

We used the following definitions for compliance and adherence: cross-sectional screening compliance—uptake of screening when due; repeat screening compliance—uptake of the same screening modality if it is repeated; screening adherence—evaluated according to CRC screening guidelines² over a period of time. We used literature-based sources for cross-sectional compliance, and calculated the resulting adherence rates for all persons entering the model.

To calculate the number of individuals screened per modality per cycle, cross-sectional screening compliance proportions are applied to the number of eligible individuals allotted to each screening modality in each year. Cross-sectional screening compliance was 64%³⁷, 64.7%³⁵, and 67.49%¹³⁴ for colonoscopy, FIT, and mt-sDNA, respectively. Repeat screening compliance was 64% for colonoscopy³⁷ and 81.6% for FIT³⁶. Repeat screening compliance for mt-sDNA and FIT was assumed to be the same. Diagnostic colonoscopy compliance subsequent to a positive stool-based screening test was 78.4%³⁶ for FIT, and was assumed to be the same for mt-sDNA. As in a previous study, surveillance colonoscopy compliance was assumed to be 100%⁷. We did not evaluate scenarios with surveillance more or less frequently than it is recommended by guidelines.

Adenoma and CRC epidemiology

The model is clinical in nature and is driven by observable epidemiological parameters. Therefore, unobserved adenoma progression was not simulated. Distributions (incidence) of adenoma number and size per age band were calculated based on the literature estimates^5,49. Size determined whether an adenoma was advanced (≥10 mm) or nonadvanced (<10 mm)⁵ Surveillance colonoscopy frequencies were simulated according to AGA guidelines, which recommend surveillance every 3 years for patients with ≥1 advanced adenoma or ≥3 nonadvanced adenomas²⁸

CRC and adenoma epidemiology estimates for this analysis were based on models, published in the peer-reviewed literature, which simulated underlying disease^5,21,50. The Surveillance, Epidemiology, and End Results Program (SEER) CRC incidence estimates from 1975 to 1979, prior to widespread CRC screening, were applied to patients who remained unscreened throughout the time horizon⁵. In addition, SEER estimates were used as baseline incidence rates for potential screeners. Relative risk incidence reductions per screening modality were based on the CISNET SimCRC colorectal model and were 63% for mt-sDNA² (reflecting a 3-year screening interval), 67% for FIT³ (reflecting an annual screening interval), and 81% (reflecting a 10-year screening interval) for colonoscopy⁴⁵. These reductions were applied to baseline incidence estimates to reflect the number of detected cancers for individuals undergoing screening.

Because the SimCRC model assumes 100% adherence⁵ applying those reductions in imperfect compliance scenarios would underestimate CRC incidence. Therefore, incidence reductions were weighted among adherent potential screeners (to whom the full SimCRC reduction per modality are applied) and noncompliant potential screeners (to whom the 1975–1979 SEER CRC incidence rates are applied) per annual cycle. We believe that this adjustment generates a more precise approximation of overall CRC incidence among screeners in real-world scenarios.

Test sensitivity and specificity per screening modality were extracted from published sources^5,19 and FDA safety and effectiveness data³⁸. Related to increased shedding of methylated DNA in stool, age-adjusted specificity was applied to mt-sDNA⁴⁹. These performance characteristics were used to determine the number of biopsies during screening and diagnostic colonoscopies, as well as the number of detected cancers and advanced or nonadvanced adenomas. For colonoscopies, the number of non-neoplastic findings was determined by the calculation of false positives based on test specificity⁵¹; however, for, stool-based tests, the number of false positives was calculated based on the diagnostic colonoscopy false positives for the screening continuum. One non-neoplastic finding per false positive was assumed.

Cost data

Although epidemiological and population-based parameters and assumptions did not vary between perspectives, costs included in each perspective did. The IDN and payer perspectives both incorporate direct medical costs, which were micro-costed and included procedural, AE, and CRC treatment components. Procedural and AE treatment costs were derived from Centers for Medicare and Medicaid Services (CMS) CPT codes⁴² and the Healthcare Cost and Utilization Project (HCUP)⁵² respectively. Colorectal cancer treatment costs were derived from the literature estimates⁴.

As a nonpayer owned delivery system, it was assumed that an IDN incurs neither bowel preparation nor stool-based screening test costs. Therefore, those costs were unique to the payer perspective and were derived from the literature and CMS³⁹. In order to estimate the full cost of screening programs, additional cost components were included in the IDN analysis. Direct nonmedical costs such as navigation services for patients and provider support⁴⁶ and indirect nonmedical costs were attributed to elements such as screening program management and administration services⁴⁶. For mt-sDNA, applicable direct nonmedical and indirect costs were assumed to be the same as for FIT. The aggregate estimates were applied annually per screening modality (Table 2).

Table 2. Economic Inputs.

(a) Screening test costs^a.
Screening modality	Medicare patients	Notes
mt-sDNA	$508.87	[39], CPT 81528
FIT	$18.05	[39], CPT G0328
(b) Colonoscopy procedure^a.
Medicare patients			Notes
Anesthesia	$465.32		[40]^b
Bowel prep	$78.83		[41]^c
CY type	Facility	Non-facility
CY (without biopsy)	$213.01	$361.44	[42], CPT 45378
CY (with biopsy)	$231.12	$464.20	[42], CPT 45380
Pathology	$70.28	$76.78	[42], CPT 88305
(c) Adverse events costs^43^,^d.
Event	Medicare patients	Notes
Perforation	$23,499.37	[44], ICD9 569.83
GI bleeding	$8,242.98	[44], ICD9 578.9
MI	$12,429.36	[44], ICD9 410.9
Arrhythmia	$8,237.96	[44], ICD9 427.9
Cardiac arrest	$12,898.06	[44], ICD9 427.5
(d) Direct nonmedical costs^e^,^f.
Item	mt-sDNA	FIT^46	CY^46	Notes on mt-sDNA assumptions
Manage provider contracts, billing systems, and other procedures	$0.00	$105.28	$63.99	If IDN has EHR integration of mt-sDNA, there is no cost
Provide patient support and navigation services	$0.00	$70.19	$104.25	Navigation services included in mt-sDNA price/cost
Support providers with screening and diagnostic services	$17.55	$17.55	$77.41	Same value as FIT; navigation for diagnostic colonoscopy after +  result, or followup with patient after negative result
Ensure appropriate treatment for complications and cancers	$22.71	$22.71	$48.51	Same value as FIT
(e) Indirect nonmedical costs^e^,^f.
Item	mt-sDNA	FIT^46	Colonoscopy^46	Notes on mt-sDNA assumptions
Program management	$233.79	$273.52	$194.05	Average of FIT and colonoscopy
Quality assurance and professional development	$113.54	$163.08	$63.99	Average of FIT and colonoscopy
Partnership development and maintenance	$0.00	$32.00	$46.45	Included as part of the manufacturer’s web portal and laboratory services
Data collection and tracking	$0.00	$139.34	$114.57	Included as part of the manufacturer’s web portal and laboratory services
Program monitoring and evaluation	$0.00	$75.35	$85.67	Included as part of the manufacturer’s web portal and laboratory services
Administration	$111.99	$102.18	$121.80	Average of FIT/ Colonoscopy
(f) Cancer treatment^4^,^g.
CRC stage	Medicare patients
Initial phase
Stage I	$27,935
Stage II	$38,829
Stage III	$47,501
Stage IV	$62,253
Continuing care costs
Stage I	$2,280
Stage II	$2,126
Stage III	$3,038
Stage IV	$9,419
Final year of life costs
Stage I	$51,385
Stage II	$51,240
Stage III	$53,991
Stage IV	$72,460

Abbreviations. mt-sDNA, multi-target stool DNA; FIT, fecal immunochemical test; CY, colonoscopy.

^a These costs are applied to patients in the payer perspective only.

^b The cost from the source ($589) was based on 45 min of physician time and was assumed to be Commercial. The Medicare cost was calculated as $589/the Medicare to Commercial cost index.

^c Calculated as the weighted average from branded (42%) and generic (58%).

^d 2014 and 2015 USD adjusted to 2019 USD using medical CPI data [45].

^e Screening program costs are applied only to the IDN perspective. Costs from the literature (assumed 2017 USD) were updated to 2019 USD with medical CPI data [45].

^f All mt-sDNA program costs are assumed.

^g CRC treatment costs were extracted from Medicare treatment costs in Table 2 of Goede in 2015 USD [4]. Costs were adjusted to 2019 USD with medical CPI data [45]. Commercial and Medicaid costs are calculated from Medicare costs based on cost indices.

* Note that the cost indices for the Commercial and Medicaid populations are multiplied by Medicare costs above.

Cost indices were calculated in order to convert Medicare costs to Commercial and Medicaid costs, and from Commercial costs to Medicare and Medicaid costs (See Supplementary Data). Because costs are reported on a cumulative basis over a relatively long time horizon (10 years), they were discounted at an annual rate of 3%⁵³. All costs were either extracted from 2019 data or were updated to 2019 USD based on the medical CPI⁴⁵.

Alternative assumptions

We modeled additional assumptions to test the robustness of the budget impact results. In alignment with American Cancer Society (ACS) recommendations, we extended screening eligibility to 45- to 75-year olds⁸. We also ran a simulation in which stool-based tests and bowel preparation costs were included in the IDN perspective. Finally, in order to compare our results with the USPSTF economic analysis⁵ we ran scenarios in which all potential screeners receive a single modality throughout the time horizon, in conjunction with assumptions of 100% compliance and the inability to switch modalities over time.

Sensitivity analyses

For the deterministic sensitivity analysis (DSA), all clinical (such as incidence, incidence reduction, cross-sectional compliance, etc.) and economic parameters (such as CRC screening program line item costs, colonoscopy costs, AE costs, and CRC treatment) were varied by ±20%, and keeping all else constant. The probabilistic sensitivity analysis (PSA), however, was conducted utilizing the Monte Carlo technique⁵⁴. The model was replicated 10,000 times during which all parameters could vary across a predetermined range, as established by draws from a beta distribution. The range for costs and resource use was determined using a gamma distribution. In all cases, we used an assumed standard error of 10%.

Results

In year 1 of the horizon, a total of 302 K eligible and 229 K patients potentially screened each year for the 50- to 75-year-old cohort. Compared to the status quo, an additional 6,412 screenings occurred in the increased mt-sDNA utilization scenario (resulting from an additional 22,068 mt-sDNA screenings, 13,793 fewer colonoscopies, and 1,863 fewer FIT screenings), when using real-world compliance rates drawn from the literature and manufacturer supplied data. Overall, increased use of mt-sDNA resulted in 257 fewer surveillance colonoscopies, as would be expected when increasing the utilization of stool-based tests, but an increase in diagnostic colonoscopies (an additional 2,037) occurred. An additional 5 cases of CRC and 1,603 fewer non-neoplastic findings were detected. At the end of 10 years, overall screening adherence was 80%, both for the status quo and with increased utilization of mt-sDNA (Table 3).

Table 3. Incremental health outcomes for the IDN and payer perspectives.

Health outcomes
	Ages 50–75 years (302K eligible patients, 229K potential screeners)	Ages 45–75 years (360K eligible patients, 280K potential screeners)
Total screens	6,750	7,038
# Screening colonoscopies	–13,793	–13,645
# Surveillance colonoscopies	–257	–228
# Diagnostic colonoscopies	2,037	2,005
Detected CRC cases	5	4

Abbreviations. mt-sDNA, multi-target stool DNA; FIT, fecal immunochemical test; CY, colonoscopy.

In the primary analysis from the IDN perspective, the estimated cumulative 10-year costs were $1.001 B for the status quo scenario and $988.9 M in the increased mt-sDNA scenario, which led to a cost savings of $19.6 M and a per-person-per month (PPPM) savings of $0.16. The distribution of total savings was attributable to decreases in the following costs: 1% in surveillance colonoscopies, 69% in screening colonoscopies, 10% in AEs, and 19% for direct non-medical costs. From the payer perspective, the aggregate 10-year costs were $659 M for the status quo and $655 M for the increased mt-sDNA scenario, which led to an incremental cost savings of $4.4 M. The distribution of total savings was attributable to decreases in the following costs: 4% in surveillance colonoscopies, 52% across all screening modalities, and 31% in AEs. As a result, the PPPM savings was $0.04 (Table 4).

Table 4. Incremental economic outcomes for the IDN and payer perspective.

(a) Incremental economic outcomes for the IDN perspective
Costs	Status quo		Increased mt-sDNA		Incremental costs compared to status quo
	6% mt-sDNA/ 11% FIT/ 83% CY		28% mt-sDNA/ 9% FIT/ 63% CY
	Ages 50–75	Ages 45–75	Ages 50–75	Ages 45–75	Ages 50–75	Ages 45–75
CRC screening	$345,607,562	$392,018,192	$329,724,337	$376,362,864	–$15,883,225	–$15,655,327
Surveillance CY	$16,927,962	$18,935,870	$16,618,601	$18,662,285	–$309,361	–$273,586
Diagnostic CY	$5,299,238	$5,888,874	$7,690,457	$8,234,466	$2,391,219	$2,345,592
Adverse events	$64,762,696	$73,020,075	$62,381,640	$70,688,413	–$2,381,056	–$2,331,663
CRC treatment	$179,103,417	$182,730,298	$179,686,708	$183,248,638	$583,291	$518,340
Direct non-medical CRC screening cost	$118,746,426	$134,573,117	$114,456,141	$130,354,200	–$4,290,286	–$4,218,917
Indirect CRC screening cost	$278,096,567	$315,242,675	$278,350,951	$315,675,205	$254,384	$432,529
Total cost	$1,008,543,868	$1,122,409,102	$988,908,834	$1,103,226,070	–$19,635,033	–$19,183,032
PPPM	$8.40	$9.35	$8.24	$9.19	–$0.16	–$0.16
(b) Incremental economic outcomes for the payer perspective
Costs	Status quo		Increased mt-sDNA		Incremental costs compared to status quo
	6% mt-sDNA/ 11% FIT/ 83% CY		28% mt-sDNA/ 9% FIT/ 63% CY
	Ages 50–75 years	Ages 45–75 years	Ages 50–75 years	Ages 45–75 years	Ages 50–75 years	Ages 45–75 years
CRC screening	$391,661,901	$444,144,314	$386,770,051	$439,518,233	–$4,891,850	–$4,626,081
Surveillance colonoscopies	$18,570,932	$20,767,798	$18,231,889	$20,467,915	–$339,043	–$299,883
Diagnostic colonoscopies	$5,806,603	$6,451,959	$8,424,774	$9,020,673	$2,618,172	$2,568,714
Adverse events	$64,762,696	$73,020,075	$62,381,640	$70,688,413	–$2,381,056	–$2,331,663
CRC treatment	$179,103,417	$182,730,298	$179,686,708	$183,248,638	$583,291	$518,340
Total cost	$659,905,548	$727,114,444	$655,495,062	$722,943,871	–4,410,486	–$4,170,573
PPPM	$5.50	$6.06	$5.46	$6.02	–$0.04	–$0.03

Abbreviations. Mt-sDNA, multi-target stool DNA; FIT, fecal immunochemical test; CY, colonoscopy; CRC, colorectal cancer; PPPM, per-person-per-month.

Compared to the primary results, when extending screening eligibility to 45- to 75-years, the overall cost savings slightly declined ($19.1 M for the IDN, a 2.5% decrease, and $4.2 M for the payer, a 4.5% decrease). Furthermore, by including stool-based tests and bowel preparation costs for screeners between ages 50 and 75 years in the IDN perspective, the savings decreased (65%) to $6.7 M. For the scenarios in which all eligible screeners receive a single modality throughout the time horizon, assuming 100% compliance and the inability to switch modalities over time, the aggregate 10-year costs from the IDN perspective were $1.19B for mt-sDNA, $979M for colonoscopy, and $2.29B for FIT. From the payer perspective, the cumulative costs were $470 M for FIT, $764 M for mt-sDNA, and $661 M for colonoscopy (Figure 2).

Figure 2. Cumulative total cost (in millions) by category for the IDN and payer perspectives.

The DSA suggested that both perspectives were most sensitive to the number of covered lives, proportion of covered lives at average risk, specificity of colonoscopy, compliance with colonoscopies and the cost of anesthesia (Figures 3 and 4). The IDN perspective included screening program costs and these pertinent non-medical costs played a key role in its budget impact results. The payer perspective included mt-sDNA and FIT test costs, and its simulation was sensitive to the mt-sDNA cost. Among 10,000 iterations, the PSA demonstrated that 100 and 98.4% of the IDN and payer perspective simulations were cost saving and the number of incremental cancers detected ranged from 11 to 44.

Figure 3. Top 10 Results of the DSA for the IDN perspective. Note. The incremental BIM for the primary analysis from the IDN perspective is −$19,635,033.

Figure 4. Top 10 Results of the DSA for the payer perspective. Note. The incremental BIM for the primary analysis from the payer perspective is −$4,410,486.

Discussion

We used a Markov cohort simulation model to project and contrast the budgetary impact of increasing utilization of mt-sDNA for CRC screening, relative to colonoscopy and FIT, for IDNs and payers. Our results show that increased utilization of mt-sDNA generates significant cost savings for both perspectives, with the IDN perspective realizing greater savings, as compared to the status quo, with similar clinical outcomes as the payer perspective.

Given that the majority of screening colonoscopies in individuals at average risk for CRC are negative, the study suggests that screening with a highly sensitive and specific stool-based test (e.g. mt-sDNA) will result in an enriched (i.e. a higher likelihood of precancer or cancer) population going on to have a colonoscopy. Colonoscopy is a more expensive screening option and carries significantly more risk of harms for patients¹⁵ including bleeding and perforation.

The cost savings for both perspectives can be explained principally by the reduction in the number of colonoscopies, resulting in savings on anesthesia, procedures, and AEs due to colonoscopy. For the IDN perspective, a large proportion of the cost savings is driven by the reduction in the number of FIT screenings, associated with a reduction in direct nonmedical spending. Several items, such as providing patient support and navigation services, were assumed to be $0 for mt-sDNA as they are included in the test cost. Although relatively inexpensive as a direct medical cost at ∼$18, IDNs may fail to consider the relatively expensive potential FIT screening program costs^46,55,56. The difference in degree of cost savings between the payer and IDN perspectives is due to the fact that screening program costs are included in the IDN perspective only, as well as the assumption that IDNs incur neither FIT nor mt-sDNA test costs.

The model results suggest that it is important to consider the incremental number of non-neoplastic findings between the status quo and the increased mt-sDNA utilization scenarios. Although the detection and removal of an adenoma would constitute ‘positive’ spending in that it would lead to CRC prevention, the detection and removal of a non-neoplastic finding could be considered wasteful spending among IDNs and payers.

Silva-Illanes and Espinoza⁵⁷ found that only 26% of 34 state-transition CRC screening models reviewed included administration costs related to CRC screening programs and only 12% included indirect costs. In order to estimate screening program costs incurred by IDNs, literature-based estimates were employed, or assumptions made for the various screening modalities (Table 1). Because several direct nonmedical and indirect costs (e.g. patient navigation and provider support) are provided by the manufacturer and thus included in the mt-sDNA cost, they were assumed to be $0 (Table 1). Similarly, several indirect costs, such as partnership development, data collection and tracking, and program monitoring and evaluation were also assumed to be $0 with mt-sDNA, as those services are provided through the manufacturer’s web portal and laboratory support services.

A distinctive feature of our model is that it allows for migration from one screening modality to another each eligible cycle. In a retrospective cohort analysis of over 300,000 FIT users, 32% switched screening modalities after the first round,³⁶ indicating that economic analyses should account for the switching of modalities due to various factors including patient preference and provider recommendations during shared decision-making. Due to the lack of data for population-specific costs, we calculated cost indices for Medicare to Medicaid and Commercial to Medicare costs in order to estimate the relevant costs. Future models could incorporate recent claims and cost data from all three patient populations. In addition, the model included the cost of following up on only the most advanced finding during screening, diagnostic, and surveillance colonoscopies. As a result, pathology costs incurred by payers and IDNs may be underestimated; however, at approximately $70 per test, this is a relatively small cost and therefore would likely not significantly affect the directionality of the overall findings. Furthermore, direct nonmedical and indirect costs are applied on a per-person-per modality basis, whereas in reality some of these costs may be fixed, that is not proportional to the number of persons screened. Future analyses should measure and distinguish among these fixed and variable (per-person) costs to more accurately estimate screening program costs. However, this model is the first to implement a variety of screening program costs not usually reflected in health economic evaluations and will aid in the decision-making processes for IDN and payer stakeholders. Further research should be conducted regarding the epidemiological components of the model. For example, it is uncertain to what degree 1 year of overdue screening within a screening regimen increases CRC incidence per modality over the time horizon. The impact of missed screening events varies with age just as incidence varies with increasing age⁵⁸ Although we did not simulate the natural history of CRC parameters, such as adenoma size and distribution, baseline cancer incidence by 5-year age band, cancer stage distribution, and cancer life expectancy per stage were extracted from highly vetted natural history models or published literature^{5,18,19,38,59}. Future research should further examine the implications of real-world health consequences of varying and switching screening modalities as technology continues to develop. Furthermore, future analyses should examine screening costs and benefits in a high risk population as well as the health and economic consequences incurred by a self-insured US employer.

When determining the optimal stool-based screening strategy, Knudsen et al.⁵ conducted a non-cost based analysis that included efficiency ratios, defined as the incremental number of colonoscopies divided by the number of life years gained compared to colonoscopy screening. According to this method, FIT screening dominated among noncolonoscopy screening methods. Because our model does not incorporate CRC-related death rates, the USPSTF modeling study incorporated 100% compliance estimates, and individuals do not switch screening modalities in the CISNET models, we ran a scenario in which all potential screeners receive a single modality with 100% compliance and are unable to switch modalities over the time horizon. Although the CISNET cost-efficiency modeling results found that mt-sDNA generates the fewest number of harms among the three modalities in this analysis over a screening lifetime^2,5 our model suggested that FIT incurred a slightly lower frequency of AEs (5,000) compared to mt-sDNA (5,080) and both generated a large reduction in AEs compared to colonoscopy screening (22,000). However, when running this simulation, mt-sDNA is cost saving relative to the other modalities with respect to IDN costs.

In order to externally validate our model with CRC screening economic analyses incorporating natural history hypotheses, we searched the literature for populations that would incorporate similar epidemiological data. In 2001, with the incremental cost-per life year gained method, Vijan et al.⁷ found that colonoscopy screening was the most cost-effective among various modalities compared to no screening. As with the Knudsen et al.⁵ study, we are unable to draw a direct comparison; however, Vijan et al.⁷ suggested that this result depended on compliance estimates for noncolonoscopy modalities. Estimates for the compliance of stool-based screening incorporated in our model are more recent as well as higher than those cited in Vijan et al., which stem from survey data collected in 1987^60,61 and 1992⁶⁰.

With the SimCRC CISNET model,²¹ Goede et al.⁴ found that additional screening costs during pre-Medicare years led to long term (12–14 years) costs offsets in Medicare spending, driven by a reduction in cancer treatment spending⁴. However, with a 10-year time horizon versus a screening lifetime, our model demonstrated annual increases in CRC spending. This is perhaps due to differing methodology of comparing current screening rates to a future scenario in which the rate reached 70% as opposed to examining the impact of varying different screening modalities. Furthermore, our model had a shorter time horizon and Goede et al.’s model implemented FOBT or colonoscopy modalities only.

Choosing which modality with which to complete CRC screening should be based on risk for CRC and shared decision making with a healthcare provider and a patient² Current models represent colonoscopy with the highest sensitivity and specificity among all of the modalities, its incidence reduction is the greatest on average assuming 100% adherence⁵ Our model suggested a very slight increase in CRC incidence with a reduction in colonoscopies. However, on-time screening in accordance with USPTF guidelines will reduce CRC-related incidence and mortality compared with no screening. There is no “one size fits all” approach to CRC screening and patients at average risk for CRC should be offered options for screening and be engaged in discussion about the benefits and potential for harms for each option as well as accounting for values and preferences⁸.

Conclusion

Mt-sDNA provides a screening option for screening eligible adults at average risk for CRC with greater sensitivity and a longer screening interval (3 years) compared with other stool-based tests (1 year)⁸ Therefore, increasing mt-sDNA utilization in CRC screening programs may provide opportunities for IDNs and payers to increase screening uptake and adherence. Based on the model results, increasing mt-sDNA utilization among screening programs including mt-sDNA, FIT, and colonoscopy, creates the potential for cost savings. Our investigation also suggests that increases in mt-sDNA utilization compared to the status quo would lead to increased CRC screening rates and more detected colorectal precancer and cancer.

With this analysis, we sought to enhance the literature on CRC screening health economic evaluations by making the cost results more applicable to real-world decision makers. The model had a shorter time horizon (10 years vs. lifetime), which reflected real-world frequent migration of patients to and from various health plans and IDNs. The model incorporated imperfect compliance estimates, which allowed simulations of inadequate adherence to screening guidelines. Our model also incorporated the ability for individuals to switch screening modalities each cycle when eligible for rescreening. With a variety of screening options, and given the choices, individuals may switch modalities over their screening lifetime. This model and framework may serve as a useful tool for IDNs and payers when considering evidence-based approaches to achieve CRC screening program goals.

Transparency

Declaration of funding

The funding source is Exact Sciences Corporation.

Declaration of financial/other relationships

Joanne M Hathway, MPH: is an employee of Precision Health Economics and Outcomes Research, which provides consulting services to the pharma, biotech, and diagnostic industry including Exact Sciences. Precision Health Economics and Outcomes Research has received professional fees for the development of the economic model.

Lesley-Ann Miller-Wilson, PhD, MBA: Employment, Stock from Exact Sciences Corporation.

Ivar S Jensen, MBA: is an employee of Precision Health Economics and Outcomes Research which provides consulting services to the pharma, biotech, and diagnostic industry including Exact Sciences Corporation. Precision Health Economics and Outcomes Research has received professional fees for the development of the economic model.

Burak Ozbay, PhD: Employment, Stock from Exact Sciences Corporation.

Catherine Regan, BA: is an employee of Precision Health Economics and Outcomes Research which provides consulting services to the pharma, biotech, and diagnostic industry including Sciences Corporation. Precision Health Economics and Outcomes Research has received professional fees for the development of the economic model.

Anupam B Jena, MD, PhD: Dr. Jena reports personal fees from Pfizer, personal fees from Hill Rom Services, Inc., personal fees from Bristol Myers Squibb, personal fees from Novartis Pharmaceuticals, personal fees from Vertex Pharmaceuticals, personal fees from Precision Health Economics, personal fees from Amgen, personal fees from Eli Lilly, personal fees from Analysis Group, personal fees from Sanofi Aventis, personal fees from Celgene, personal fees from Tesaro, personal fees from AstraZeneca, personal fees from Biogen, outside the submitted work.

Milton C Weinstein, PhD: Dr. Weinstein reports personal fees from Precision Health Economics and Outcomes Research,, which provides consulting services to the pharma, biotech, and diagnostic industry including Exact Sciences Corporation, during the conduct of the study.

Philip D Parks, MD, MPH: Employment, Stock from Exact Sciences Corporation.

JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Author contributions

All authors were instrumental in drafting the manuscript and participated extensively in review and editing of the manuscript.

Acknowledgements

The authors would like to acknowledge Andrew Piscitello, PhD, Weiyu Yao, MSc, and Phil L Cyr, MPH.

Notes

Correction Statement

This article was originally published with errors, which have now been corrected in the online version. Please see Correction (http://dx.doi.org/10.1080/13696998.2021.1935473)

Notes

1 Mt-sDNA’s compliance rate represents the cumulative completed tests from kits shipped to patients during the 6-month period ending 12 months prior to the end of the Q3 2019, excluding program orders.

2 Knudsen AB, Zauber AG, Rutter CM, et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US Preventive Services Task Force. JAMA. 2016 Jun 21;315(23):2595–609, Supplement, Table 6a.

3 Knudsen AB, Zauber AG, Rutter CM, et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US Preventive Services Task Force. JAMA. 2016 Jun 21;315(23):2595–609, Supplement, Table 5a.

4 Knudsen AB, Zauber AG, Rutter CM, et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US Preventive Services Task Force. JAMA. 2016 Jun 21;315(23):2595–609, Supplement, Table 3a.