Development of a Follow-Up Measure to Ensure Complete Screening for Colorectal Cancer

Key Points Question What are the supporting evidence, feasibility, reliability, and validity of a quality performance measure on follow-up colonoscopy after an abnormal result of a stool-based screening test (SBT) for colorectal cancer (CRC)? Findings In this quality improvement study including 20 581 adults at 38 health care organizations, 48% of patients received a colonoscopy within 6 months after an abnormal result of an SBT for CRC. A quality measure that tracks follow-up rates within 6 months of SBT is feasible, valid, and reliable. Meaning These findings suggest that a measure on follow-up colonoscopy after an abnormal result of an SBT for CRC is warranted based on low current performance rates and high feasibility, validity, and measurement reliability.


Introduction
Colorectal cancer (CRC) is the third leading cause of cancer deaths in men and women in the US, and it is estimated that there will be approximately 53 000 deaths due to CRC in 2023. 1 Screening for CRC is widely recommended [2][3][4][5] because, when identified early, CRC is one of the most treatable forms of cancer, with a 5-year survival rate of 90%. 6,7In contrast, the 5-year survival rate among those diagnosed with late-stage disease is only 14%. 7][10] Diagnosis at an early stage will reduce neoplastic progression of disease [8][9][10] and therefore increase the possibility of curative treatment. 9It follows that early diagnosis may also decrease downstream complications, improve outcomes, and minimize costs to the patient and the health care system. 11Further, early diagnosis helps reduce racial and ethnic disparities in outcomes: when CRC is diagnosed at a localized stage, survival rates are comparable across racial and ethnic groups. 12tients and health systems are increasingly using SBTs for convenience and patient preference 13,14 and to maximize population-level screening rates without a substantial increase in costs. 15[18][19] In a study using the Optum Labs Data Warehouse (OLDW), follow-up rates at 3 and 6 months were low, at 43% and 51%, respectively. 20Some health systems have been able to achieve effective follow-up of 85% of patients within 6 months, 21 but substantial variation exists. 20ality measures can lead to improved performance, [22][23][24][25][26][27] but the current CRC screening Healthcare Effectiveness Data and Information Set (HEDIS) measure 28 is incomplete for individuals receiving SBTs for screening.The CRC screening measure is numerator compliant if an SBT is performed but does not account for the result of the test.Only when the SBT result is negative (or normal) is the screening process complete.If the SBT result is positive (or potentially abnormal), a follow-up colonoscopy is required to detect the presence of neoplasia.For these patients, the existing HEDIS measure only captures the first step in the screening process.
To close this gap, we developed and tested an additional measure to track the completion of screening for patients with abnormal SBT results.While the proposed measure would complement the existing HEDIS CRC screening measure used in health plans, we propose this measure for reporting by health systems due to the availability of test results in EHR data.The CRC screening completion measure will assess the rates of timely (within 6 months) follow-up colonoscopy for adults aged 45 to 75 years who completed an initial stool-based CRC screening test (fecal occult blood test, fecal immunochemical test, or multitarget stool DNA) with an abnormal result.A follow-up window of 180 days was supported by studies that demonstrated an increased risk of any stage of CRC when colonoscopies are conducted at least 6 months following an abnormal SBT result. 1,8,9The measure is further stratified by race and ethnicity to gauge disparities in CRC screening and follow-up.This addresses the growing trend by quality measure stewards to institute stratification by proxy variables for at-risk populations to address health equity and adjust for patient mix across organizations.As part of the measure development process, we engaged 4 national experts to guide the development and specification of the measure.We then tested the robustness and reliability of the measure using an existing database of electronic health record (EHR) and adjudicated claims data.
The proposed CRC screening completion measure is a novel, innovative measure concept that builds on and addresses an important shortcoming in an existing measure.The proposed measure, combined with the current HEDIS measure, ensures that patients have received a complete screening, which includes a follow-up colonoscopy after an abnormal stool-based test result.This report will describe the measure development and testing process, as well as results, for the CRC screening completion measure.

JAMA Network Open | Public Health
Development of a Follow-Up Measure to Ensure Complete Screening for CRC

Methods
This study was deemed exempt from institutional review board approval and the need for informed consent by the Institutional Review Board Affairs Department of WCG because the research was limited to interactions involving educational tests, survey procedures, interview procedures, or observations of public behavior.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and Standards for Quality Improvement Reporting Excellence (SQUIRE) reporting guidelines.

Data Source
This study primarily used deidentified EHR data-including clinical, demographic, medication, laboratory, and utilization or visit data-from January 1, 2016, to December 31, 2020, available in the OLDW.These data are sourced from more than 50 US health care organizations (HCOs) and have been normalized and standardized into a single database.The OLDW contains longitudinal health information on enrollees and patients, representing a diverse mixture of ages and geographical regions across the US.Adjudicated, deidentified administrative claims data were also accessed for a sensitivity analysis of received services, such as follow-up colonoscopies, provided outside of the index health system where the index CRC screening occurred.Approximately 10% of patients in the OLDW also have claims data, and this overlap was used exclusively in the sensitivity analysis.These claims data include medical and pharmacy claims, laboratory results, and enrollment records for commercial and Medicare Advantage enrollees.Study data were accessed using techniques compliant with the Health Insurance Portability and Accountability Act of 1996.

Measure Specification
For the study, the measure specified was the percentage of adults aged 50 through 75 years who completed a colonoscopy within 180 days following an abnormal SBT result for CRC within the

Measure Testing Procedures
The specified measure performance rates were calculated and stratified by several patient characteristics, including race and ethnicity.One-way analysis of variance (ANOVA) was used when necessary to determine whether measure performance differed across patient strata.Measure performance was also compared across 38 HCOs.A repeated-measures ANOVA was used to test the change in HCO performance over time.To assess reliability, a β-binomial distribution 29  between-category differences vs variability within a category and is a quantification of how well a measure can distinguish 2 or more strata.Reliability of at least 70% is considered sufficient when comparing groups of individuals. 29Two sensitivity analyses were also conducted.The first compared a 180-day follow-up with a shorter follow-up period of 90 days.The second used adjudicated claims data to identify colonoscopies that occurred outside of the health system where the index SBT result was returned.
Feasibility testing included assessments of face validity as determined by the national expert advisors during 6 monthly 1-hour meetings and external feasibility field testing to assess the plausibility of collecting all required data elements for a performance measure and automatically calculating the measure in an EHR system for e-measurement.The National Quality Forum's feasibility scorecard was used to field test the measure.The scorecard assessed data availability, accuracy, data standards, and work flow related to capturing the data required to calculate the measure by which each data element was scored.Descriptions of the data element domains in the scorecard are provided in eTable 1 in Supplement 1.

Measure Testing Population or Participants
The measure testing population included patients who were at average risk and were eligible for stool-based CRC screening at the time of the study.This selection was based on the US Preventive Services Task Force recommendations for CRC screening in 2020 for patients aged 50 to 75 years.In 2021, the US Preventive Services Task Force revised the recommendation to include individuals aged 45 to 75 years, as supported by studies reporting increased rates of CRC among adults aged 45 to 49 years. 1,14Because fecal occult blood tests are also used diagnostically to test for blood in the stool unrelated to CRC, particularly in the inpatient and emergency care setting, we excluded test results that were associated with either of these settings (within 14 days of index).We selected 2018 as the primary year for measure evaluation because it was the most recent available year for which follow-up was not affected by the COVID-19 pandemic.Data were available for all years from 2016 through 2020, though this larger dataset was used only for the year-by-year comparisons.An attrition diagram is shown in Figure 1.

Statistical Analysis
Data were analyzed from June to estimate β-binomial distributions.A priori significance levels were set at a 2-sided P < .05,and 2-sided tests were used for all comparisons.Uncertainty was quantified using 95% CIs for patient level characteristics or using the IQR when comparing across HCOs.

Measure Population Description
Using the eligible population criteria, the EHR-derived population for measure evaluation included 20 581 patients with an abnormal SBT result in the 2018 measurement year.Of these, 2163 had overlapping adjudicated administrative claims data that were used in a sensitivity analysis to examine colonoscopy receipt outside of the index health system.Twenty-two of the HCOs were integrated delivery systems, representing 17 181 patients (83.5%) in the sample.Table 1 provides patient characteristics of EHR-derived and overlapping claims-based populations.
Other b or unknown 1077 ( b Specific race categories were unavailable from the Optum Labs Data Warehouse.
c Metropolitan urbanized area indicates population of or greater; large urban cluster, population of 10 000 to 49 999; small urban cluster, population of 2500 to 9999; and rural, population of less than 2500.
Variation in measure performance across HCOs is indicative of measure feasibility, 31 as is the ability to move or change rates over time.Colonoscopy follow-up rates differed significantly across years (1-way repeated-measures ANOVA, P < .001)and increased by approximately 33% between 2016 and 2019, indicating the ability of HCOs to improve performance rates.Rates subsequently declined by 14% through 2020, presumably due to the COVID-19 pandemic (Figure 3).
measurement year.At the time of the study, the guideline-recommended lower age bound was 50 years.The denominator was defined as the number of adults aged 50 through 75 years with an abnormal CRC screening SBT result within the measurement year.The numerator was defined as the number of individuals in the denominator who received a colonoscopy within 180 days of the positive test result.The 180-day follow-up window extended into the subsequent measurement year to account for patients who received an abnormal SBT result in the second 6 months of the measurement year.Patients were excluded if they had a previous CRC diagnosis (International Classification of Diseases, Ninth Revision [ICD-9], codes 153.8 or 153.9;International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] codes C18.0 or C18.1); had a history of total colectomy (ICD-9 Clinical Modification codes 45.81 or 45.82, ICD-10 Procedure Classification codes 0DTE0ZZ or 0DTE4ZZ; Current Procedural Terminology [CPT] code 44150 or 44151); had a diagnostic SBT (CPT code 82271 or 82272); or were receiving palliative or hospice care (ICD-9 code V66.7, ICD-10 code Z51.5, CPT code 99377 or 99378, or Healthcare Common Procedure Coding System code G0182).Health care organizations with less than 20 eligible index patients and those with less than 100 colonoscopies performed in the measurement year were excluded, leaving 38 HCOs in the analytic dataset.The measure schematic is provided in eFigure 1 in Supplement 1.
was fit to quantify the variance within and across strata of interest (HCO, race, ethnicity, and year).The reliability statistic represents the percentage of variance across the population that is due to JAMA Network Open | Public Health Development of a Follow-Up Measure to Ensure Complete Screening for CRC JAMA Network Open.2024;7(3):e242693.doi:10.1001/jamanetworkopen.2024.2693(Reprinted) March 25, 2024 3/12 Downloaded from jamanetwork.comby guest on 03/30/2024

Figure 2 .
Figure 2. Measure Performance, by Health Care Organization and Volume of Tests With Positive Results, 2018 5000 4000 3000 2000 1000 0

Table 1 .
Patient Characteristics in 2018 Follow-up colonoscopy within 180 days of abnormal stool-based test ranged from 13.1% to 66.9% (median, 47.9%) across 38 health care organizations.