Data source

This is a retrospective cohort study using Texas Cancer Registry (TCR)-Medicare linkage data with cancer patients diagnosed from 1999–2017. TCR is one of the largest cancer registries in the U.S; data collected by TCR include patient demographics, primary tumor site, stage, first course of treatment, tumor morphology, cause of death, and survival. TCR-Medicare linkage was performed by the National Cancer Institute (NCI) and the CMS. The Medicare claims data include billing information on hospital stays, physician services, and hospital outpatient visits. For this study, data were extracted from Medicare beneficiary summary files, Medicare Provider Analysis and Review files, Outpatient Standard Analytical Files, and Medicare Carrier files. The study involving human subjects was reviewed and approved by IRB (#21–0311) of the University of Texas Medical Branch at Galveston and another IRB (#22–0031) for Medicare data overall. This data is acquired through the Data Use Agreement (DUA) with the NCI and individual investigators are not required to obtain participant consent. Medicare Beneficiary Identifiers (MBI) were used to uniquely identify Medicare patients and link their clinical claims across multiple years.

Study cohort

This study identified cancer patients first time diagnosed with breast, prostate, or colorectal cancer at any time during the period of 1999–2014 and who had survived over five years since diagnosis following the methodology used in our previous study [19]. A yearly cohort from 2006 to 2019 was generated (14 cohorts in total). To account for at least 5-year survival patients who were diagnoses with cancer between 1999 and 2001 were included in the 2006 cohort; patients who were diagnosed with cancer between 1999 and 2002 were included in the 2007 cohort, and vice versa until they dropped out of Medicare. The sample collection was repeated every year and the 2019 cohort included patients who were diagnosed with cancer between 1999 and 2014.

At each study year, individuals who were under 66 years old, had passed away, were living in a nursing home, or had not enrolled in both Medicare Parts A & B for 12 consecutive months in the year were excluded. The study cohort was only limited to cancer survivors with at least two outpatient claims, or one inpatient claim related to diabetes (ICD-9 and ICD-10: 250.x, E10.x, E11.x) in the period of 1999 to 2014. For instance, the 2006 cohort included patients with diabetes claims between 1999 and 2006 and the 2007 cohort included patients who had diabetes-related claims between 1999 and 2007. The same process was repeated to select cohorts that met the inclusion criteria. However, beneficiaries with multiple cancers, who lived in long-term care facility, who were diagnosed with end-stage renal disease or kidney transplantation, or who had enrolled in Medicare Advantage were excluded. S1 Fig in S1 File provides a flow chart of eligible cohort in 2006 as an example.

Dependent variables

The primary outcome was the first-time use of DSMES per 1000 in each year from 2006 to 2019 (see Formula 1). In the multivariable model, the outcome was a dichotomous variable including 1 = yes (first time) and 0 = no (not the first-time use). The secondary outcome was the number of minutes of DSMES that a patient has received (see Formula 2). The service can be covered by Medicare Part B using either Healthcare Common Procedure Code System [HCPCS] code G0108 or G0109 [20]. The HCPCS code G0108 refers to diabetes outpatient self-management training services for an individual per 30 minutes and G0109 refers to DSMES for a group (two or more people) per 30 minutes. Within the first 12 months, Medicare beneficiaries can attend no more than 10 hours of training in a group setting focused on increasing patients’ self-care knowledge and skills. CMS reimburses no more than two hours of follow-up training for subsequent years. Based on the formula below, 28 rates were generated and used to constitute two trends from 2006 to 2019.

2006 cohort: First-time DSMES per 1000 = (# of first-time DSMES users in 2006) / (# of all cancer survivors in 2006) *1,000 (1)

2006 cohort: Mean DSMES utilization time in minutes = [(sum of # of minutes of first-time users) + (sum of # of minutes of follow-up users in 2006)] / (# of first-time users + # of follow-up users in 2006) (2)

Independent variable & covariates

The primary independent variable of this study is the year of attending DSMES. The covariates were categorized into two parts: demographic characteristics and health conditions. Demographic factors included each beneficiary’s age, sex (male/female), race/ethnicity (White, Black, Hispanic, and Other), Medicare-Medicaid dual eligibility (yes/no), education, poverty, and income at the zip code level, and residence (metro/ non-metro area). The encrypted zip code data was created by TCR Medicare using the Census 1990, Census 2000, and the American Community Survey. Metro/non-metro was classified based on the Rural-Urban Continuum codes for years 1993, 2003, and 2013, and more information can be found on the website [21]. Health conditions included medical comorbidities and type of diabetes. The study adapted the Elixhauser Comorbidity Index to exclude diabetes and assess the burden of comorbidities [22]. Next, functional status was defined using each patient’s Medicare original entitlement to disability insurance (disabled or not disabled). Finally, patient’s diagnoses were classified as Type 1 diabetes (ICD-9 250.x1, ICD-9 250.x3, and ICD-10 E10.xx), Type 2 diabetes (ICD-9 250.x0, ICD-9 250.x2, and ICD-10 E11.xx), and diabetes with complications [23, 24].

Statistical analyses

Trends (2006–2019) of DSMES utilization (first-time users and follow-up users) among yearly study cohorts including patients who survived cancer for at least five years and living with Type 1 or Type 2 diabetes were generated. For each year from 2006 to 2019, patients’ demographics, type of cancer survived, presence of incident Type 1 or Type 2 diabetes, presence of uncontrolled Type 1 or Type 2 diabetes, DSMES utilization (first-time users, follow-up users, time of utilization), Medicare original entitlement, dual eligibility, education, poverty, residence and Elixhauser comorbidity were summarized using frequency and percentage for categorical variables and mean and standard deviation for continuous variables. Multivariable logistic regression to predict the likelihood of first-ever DSMES utilization (yes = the first time/no = not the first time) was performed using the same set of the covariates but excluding the poverty level, which is highly correlated with the education level. The study has repeated the model predicting the first-time utilization by three types of cancers (S1-S3 Tables in S1 File). Because every cancer survivor could start DSMES at any given time from 1999 to 2019, the accumulated number of minutes of DSMES among all users was used as the secondary outcome variable. A multivariable gamma model was constructed to study the association between length of average time of DSMES and year after holding all covariates constant. A gamma regression model was chosen because the distribution of length of DSMES time was non-negative and positively skewed. Data management and analysis were conducted using SAS statistical software version 9.4 (SAS Institute Inc.). A p-value less than 0.05 was considered statistically significant.

Characteristics of cancer survivors

The study identified Medicare beneficiaries that met the inclusion criteria every year and accumulated eligible samples from 2006 to 2019. Given the nature of aging, the later sample contains older people with more comorbidities. Additionally, this study shows that the portions of cancer survivors who live in areas with very good education, very low poverty, and metropolitan areas have increased over 14 years. These findings are in agreement with a rich body of literature regarding socially induced disparities in cancer survival. Studies show that living in rural areas or having poor access to education, employment, healthcare, insurance, quality housing, food security, and social support adversely contribute to cancer survival rates [25, 26]. Cancer patients have complex healthcare needs, and it is critical to advance equity in cancer health through a better understanding of, and action to address, social determinants [27]. Such a blueprint offers insights on practice, research, and policy to reduce the burden of cancers by addressing income inequalities, supporting targeted policies with explicit benefits for disadvantaged groups, and incentivizing healthcare providers to adopt holistic healthcare approach. More longitudinal studies are also recommended to track the improvement in cancer survivorship of disadvantaged populations.

Trends of initial DSMES and DSMES follow-up

Since each cancer survivor in the study could join DSMES at any given time between 1999 and 2019, the number of first-time users provides insight on how available and accessible the DSMES program is for all cancer survivors. On the other hand, the accumulated number of all users each year indicates the retention of all qualified cancer survivors. This study discovered a rapid increase in the number of first-time DSMES users in 2015. This might be explained by the increase in the number of certified diabetes educators and CMS’s leading the trend of healthcare systems to shift from fee-for-service to value-based care over the past decade [28, 29]. However, more studies are needed to investigate the cause of the increase in 2015 in Texas particularly. The study findings show a sudden drop in the trend of initiation in 2016, whereas the retention has remained on a plateau following a substantial increase in 2013. This might be explained by the change in the Medicare reimbursement for Chronic Care Management (CCM) in 2016 to incentivize a spending on a longer time on CCM services [30]. As a result, DSMES providers (e.g., diabetes education specialists) might focus more on current users to maximize the amount of reimbursement instead of attracting new users. By learning that the DSMT utilization is very low among Medicare beneficiaries, strategies to encourage diabetes care and education specialists to improve the enrollment and retention among cancer survivors are highly important to ensure the successful implementation of the program. More research is needed to explain changes in both DSMES initiation and retention trends by Medicare beneficiary cancer survivors in order to inform reimbursement strategies.

Determinants of utilization

This study found that DSMES initiation is positively associated with Hispanic ethnicity, diabetes incidence, diabetes complications, and dual eligibility, but negatively associated with older age, living in metropolitan areas, and having more comorbidities. The finding about residence is inconsistent with the literature, noting that urban residents (areas >50,000 residents) have more DSMES participation [31]. However, this study defined “metropolitan” as an area with more than 250,000 residents, and the study was focused on cancer survivors, not Medicare beneficiaries in general [32]. Next, a 2012 report suggested that fewer African American and Hispanic beneficiaries know that Medicare helps pay for diabetes self-management education, yet this study pointed out that Hispanic ethnicity is a favorable factor for both initiation (first-time use) and retention (average minutes per use) [33]. A recent finding revealed that Latino participants are more likely to complete DSMES sessions because they are enrolled into culturally tailored programs [34]. Therefore, developing programs unique to each culture is highly recommended in enhancing DSMES participation among racial minorities with diabetes. Finally, the study results show that dually eligible individuals are more likely to be first-time and follow-up users. It might be explained by their better awareness and access to more providers and DSMES programs covered by both Medicare and Medicaid. This finding may guide future efforts to strengthen DSMES utilization by increasing availability and payment of evidence-based practices.

Different from the determinants of initiation, this study found that having disability, living in areas with poor education, and having multiple comorbidities are related to a longer time of DSMES use. One study pointed out that people with higher education are more likely to work full or part-time, that impeded their continuation in the program [35]. On the other hand, individuals with disability or complications might be more likely to have routine medical visits, that increases the likelihood to stay in health systems and use DSMT [36]. Because the former research drew conclusions from various samples, more research on factors to retain DSMES participants would be helpful to design strategies to support them to complete the program.

DSMT/DSMES has been adopted by Medicare since 1997 and it is an evidence-based program that educates people living with diabetes on how to mitigate the burden of this chronic condition and reduce the risk of complication [37]. However, its utilization among people living with diabetes generally and among those cancer survivors living with diabetes is considerably limited. On average, this study found that around 10 of 1000 cancer survivors per year started to participate in DSMES over 14 years. A 2015 study also shows that only 5% of Medicare beneficiaries attended DSMES [7]. The Centers for Disease Control and Prevention identified three barriers to DSMT or DSMES utilization including programmatic barriers to starting or sustaining services, provider barriers to referral, and barriers to access and participation [38]. In 2022, the reimbursement to a provider in Galveston, TX for a 30-minute DSMES individual session is only $56.88, and $16.22 per patient for a 30-minute group session [39]. While DSMES will greatly increase the value of diabetic care, it has not been reasonably reimbursed in any value-based care model [40]. Additionally, prostate, breast, and colorectal cancer survivors account for 60% of all survivors [41]. The evidence-based DSMES that combines coping skills and lifestyle changes is very likely to enhance cancer survivorship [15–18]. In response, this study highly supports better coordination between primary care practitioners, oncologists, and endocrinologists to follow up with cancer survivors in a course covering diabetes [42, 43]. Strategies to address the three aforementioned barriers are also critical such as (1) individual level: use of telehealth to overcome transportation barriers, (2) provider level: incentives for providers to refer eligible patients to DSMES programs, and (3) system level: a reimbursement scheme to incentivize healthcare institutions to include DSMES in survivorship care plans [40, 43].

Limitations

This study has several limitations. First, the study used administrative claims data of Texas Medicare beneficiaries who had survived breast, prostate, and colorectal cancers for at least five years. The results cannot be generalized to other types of cancer survivors or Medicare beneficiaries in other states. Second, to qualify for DSMES, an individual must not only have a diagnosis of Type 1 or Type 2 diabetes but also have a fasting blood glucose of 126 mg/dL on two separate occasions [44]. However, the Medicare claims data lack the information on the second criterion. As this study has assumed that all cancer survivors diagnosed with diabetes qualify for the program, the participation rate calculated in this study could seem too low. Third, the American Academy of Professional Coders argues that diabetes is frequently under-coded which might result in lack of referral to DSMES [45]. Accurate coding of diabetes and its complications will render necessary services to cancer survivors.