The Impact of Race–Ethnicity and Diagnosis of Alzheimer’s Disease and Related Dementias on Mammography Use

Simple Summary Analyzing real-world data from the OneFlorida+ Clinical Research Network, we examined the impact of ADRD diagnosis and race–ethnicity on mammography use in breast cancer screening (BCS)-eligible women. We found that BCS-eligible women with Alzheimer’s disease (AD) and AD-related dementias (ADRD) were more likely to undergo a mammography than the BCS-eligible women without ADRD. Stratified by race–ethnicity, BCS-eligible Hispanic women with ADRD were more likely to undergo a mammography, whereas BCS-eligible non-Hispanic black and non-Hispanic other women with ADRD were less likely to undergo a mammography. Abstract Breast cancer screening (BCS) with mammography is a crucial method for improving cancer survival. In this study, we examined the association of Alzheimer’s disease (AD) and AD-related dementias (ADRD) diagnosis and race–ethnicity with mammography use in BCS-eligible women. In the real-world data from the OneFlorida+ Clinical Research Network, we extracted a cohort of 21,715 BCS-eligible women with ADRD and a matching comparison cohort of 65,145 BCS-eligible women without ADRD. In multivariable regression analysis, BCS-eligible women with ADRD were more likely to undergo a mammography than the BCS-eligible women without ADRD (odds ratio [OR] = 1.19, 95% confidence interval [CI] = 1.13–1.26). Stratified by race–ethnicity, BCS-eligible Hispanic women with ADRD were more likely to undergo a mammography (OR = 1.56, 95% CI = 1.39–1.75), whereas BCS-eligible non-Hispanic black (OR = 0.72, 95% CI = 0.62–0.83) and non-Hispanic other (OR = 0.65, 95% CI = 0.45–0.93) women with ADRD were less likely to undergo a mammography. This study was the first to report the impact of ADRD diagnosis and race–ethnicity on mammography use in BCS-eligible women using real-world data. Our results suggest ADRD patients might be undergoing BCS without detailed guidelines to maximize benefits and avoid harms.


Introduction
Breast cancer is the most prevalent cancer and the second leading cause of death by cancer in women in the United States (US) [1]. It is estimated that approximately 287,850 women will be diagnosed with and 43,250 women will die from breast cancer in the United States in 2022 [1]. Breast cancer screening (BCS) with mammography is a crucial method for improving cancer survival. Undergoing mammography helps detect breast cancer in its early stages, and thus provides the best opportunity for successful treatment and prognosis. The benefits of screening mammography have been well-documented in many randomized trials and observation studies [1][2][3][4][5][6][7][8][9]. In the report from the International Agency for Research on Cancer (IARC), mammography reduced the mortality rate of breast cancer by 40% among women of 50-69 years of age [9]. The proven benefits of mammography led to the creation of BCS guidelines by national professional associations, such as the mentia, drug-induced persisting dementia), or conditions that could cause dementia (e.g., Parkinson's disease, paralysis agitans, temporal sclerosis, etc.; see Appendix A Table A1 for ICD codes) between the ages of 50 and 73 years from the OneFlorida+ EHRs. To these women, we applied the same exclusion criteria that were applied to the case cohort. Next, each case woman with ADRD was matched to three women without ADRD based on age and calendar year. For the comparison cohort, the index date was defined as the date of a random inpatient or outpatient encounter in the matched calendar year.

Study Outcome and Exposures
The primary outcome of the study was mammography use (yes or no) within two years (i.e., 730 days) after the index date ( Figure 1). Since the USPSTF recommended that women aged 50 to 74 years should get a screening mammography every two years, we considered a woman in our study population compliant with the USPSTF guideline if she underwent a mammography within two years after the index date. Mammography use within the outcome observation window was identified using Current Procedural Terminology (CPT) codes 77065, 77066, and 77067. The primary exposures were having an ADRD diagnosis or not (i.e., belonging to the case or comparison cohort) and raceethnicity (non-Hispanic white [NHW], non-Hispanic black [NHB], non-Hispanic other [NHO], Hispanic, or unknown).
Cancers 2022, 14, x FOR PEER REVIEW 3 of 16 excluded case patients who were previously diagnosed with breast cancer prior to the index date, as well as patients without a valid ZIP code. To create the comparison cohort, we extracted all women without any ADRD diagnosis, dementia (e.g., alcohol-induced persisting dementia, drug-induced persisting dementia), or conditions that could cause dementia (e.g., Parkinson's disease, paralysis agitans, temporal sclerosis, etc.; see Appendix A Table A1 for ICD codes) between the ages of 50 and 73 years from the OneFlorida+ EHRs. To these women, we applied the same exclusion criteria that were applied to the case cohort. Next, each case woman with ADRD was matched to three women without ADRD based on age and calendar year. For the comparison cohort, the index date was defined as the date of a random inpatient or outpatient encounter in the matched calendar year.

Study Outcome and Exposures
The primary outcome of the study was mammography use (yes or no) within two years (i.e., 730 days) after the index date ( Figure 1). Since the USPSTF recommended that women aged 50 to 74 years should get a screening mammography every two years, we considered a woman in our study population compliant with the USPSTF guideline if she underwent a mammography within two years after the index date. Mammography use within the outcome observation window was identified using Current Procedural Terminology (CPT) codes 77065, 77066, and 77067. The primary exposures were having an ADRD diagnosis or not (i.e., belonging to the case or comparison cohort) and race-ethnicity (non-Hispanic white [NHW], non-Hispanic black [NHB], non-Hispanic other [NHO], Hispanic, or unknown).

Covariates
We included in our data analysis a number of covariates that could potentially impact the utilization of mammography. These covariates included age, comorbidity, rurality (i.e., urban vs. rural residency), and social vulnerability. Comorbidity was measured using the Charlson Comorbidity Index (CCI), with a higher CCI indicating worse baseline health [24]. To calculate the CCI, we extracted the following conditions at baseline from the EHRs (Figure 1): moderate/severe liver disease, cerebrovascular disease, peripheral vascular disease, renal disease, hemiplegia or paraplegia, dementia, mild liver diseases, congestive heart failure, chronic obstructive pulmonary disease, diabetes, and diabetes with complications. The ICD codes used to identify these conditions were listed in Appendix A Table A2. We included rurality in the study as previous studies showed disparities in medical resources accessibility between rural and urban patients that may affect their cancer screen use [25,26]. Rurality was measured using the ZIP code level ruralurban commuting area (RUCA) codes. RUCA codes are a census tract-based classification that characterize all U.S. census tracts with respect to their rural/urban status and

Covariates
We included in our data analysis a number of covariates that could potentially impact the utilization of mammography. These covariates included age, comorbidity, rurality (i.e., urban vs. rural residency), and social vulnerability. Comorbidity was measured using the Charlson Comorbidity Index (CCI), with a higher CCI indicating worse baseline health [24]. To calculate the CCI, we extracted the following conditions at baseline from the EHRs (Figure 1): moderate/severe liver disease, cerebrovascular disease, peripheral vascular disease, renal disease, hemiplegia or paraplegia, dementia, mild liver diseases, congestive heart failure, chronic obstructive pulmonary disease, diabetes, and diabetes with complications. The ICD codes used to identify these conditions were listed in Appendix A Table A2. We included rurality in the study as previous studies showed disparities in medical resources accessibility between rural and urban patients that may affect their cancer screen use [25,26]. Rurality was measured using the ZIP code level rural-urban commuting area (RUCA) codes. RUCA codes are a census tract-based classification that characterize all U.S. census tracts with respect to their rural/urban status and commuting relationships to other census tracts [27]. ZIP code level RUCA codes are approximations of that at the census tract level, in which each ZIP code is assigned with a primary code. The primary codes range between 1 and 10, with 1 being the most urbanized and 10 being rural. We linked the RUCA codes to our study population using the patients' latest ZIP codes in the EHRs. Social vulnerability was measured using the Center for Disease Control and Prevention (CDC)'s Social Vulnerability Index (SVI). The SVI uses U.S. census data to measure the social vulnerability of every census tract. Each census tract is ranked on 15 social factors and grouped into four related themes that include socioeconomic status (SVI-SS), household composition and disability (SVI-HCD), minority status and language (SVI-MSL), and housing type and transportation (SVI-HTT). A higher value of SVI indicates higher social vulnerability. We linked the SVI as the four SVI themes to our study population using the patients' latest ZIP code in EHRs.

Statistical Analysis
First, to characterize our study population, we calculated the means with standard deviations (SDs) and frequencies with percentages for the variables of interest stratified by ADRD diagnosis. Second, to examine the association between ADRD diagnosis and mammography use, we built a logistic regression model (base model) with mammography use being the dependent variable and ADRD diagnosis, race-ethnicity, and the covariates being the independent variables. Third, to examine whether race-ethnicity modified the association between ADRD diagnosis and mammography use, we extended the base model to include an additional ADRD diagnosis by race-ethnicity interaction term (interaction model). Results from both logistic models were reported as odds ratios (ORs) and the associated 95% confidence intervals (CIs). All statistical analyses were conducted using python 3.9.4 and SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Characteristics of Study Population
We summarized the characteristics of our study population in Table 1. Overall, we extracted 21,715 BCS-eligible women with ADRD and 65,145 matching BCS-eligible women without ADRD from OneFlorida+. The ADRD and non-ADRD patients had the same mean age (64.8 vs. 64.8 years, p = 1.000), the matching variable. Although the overall racial-ethnic makeup did not differ significantly between the two study cohorts (p = 0.114), a higher percentage of NHW (47.1% vs. 42.7%) and NHO (4.9% vs. 3.5%) patients were presented in the non-ADRD comparison cohort compared with the ADRD patient cohort. Compared to the non-ADRD patients, the ADRD patients were more likely to reside in urban areas (91.3% vs. 87.8%, p < 0.001), have worse baseline health (CCI > 0: 28.5% vs. 15.8%, p < 0.001), and associate with higher social vulnerability in terms of all four SVI themes (SVI-SS: 0.57 vs. 0.55, p < 0.001; SVI-HCD: 0.52 vs. 0.49, p < 0.001; SVI-MSL: 0.60 vs. 0.59, p < 0.001; SVI-HTT: 0.57 vs. 0.50, p < 0.001). Lastly, the BCS-eligible ADRD patients were more likely to undergo a mammography within two years of the index date than the BCS-eligible non-ADRD patients (10.0% vs. 8.0%, p < 0.001).
We summarized the social vulnerability of the study population by race in Table 2. . NHW, NHO, and Hispanic BCS-eligible women with ADRD were more socially vulnerable compared to those without ADRD, while NHB BCS-eligible women without ADRD were more socially vulnerable than those who had ADRD.

Discussion
In this study, we extracted a cohort of BCS-eligible women with ADRD and a matching comparison cohort of BCS-eligible women without ADRD in the OneFlorida+ CRN. We examined the association of ADRD diagnosis and race-ethnicity with mammography use adjusting for age, baseline health, rurality, and social vulnerability. We found that BCS-eligible women with ADRD were more likely to undergo a mammography than the BCS-eligible women without ADRD. The BCS rate among the control group was low compared to that of the general BCS eligible population. This difference was expected, as the control group was selected to match the age, encounter year, and month of the AD patients. The matching criteria made the BCS-use pattern among the control group different from that in the general population. Stratified by race-ethnicity, BCS-eligible Hispanic women with ADRD were more likely to undergo a mammography, whereas BCS-eligible non-Hispanic black and non-Hispanic other women with ADRD were less likely to undergo a mammography. Mammography use did not differ by ADRD diagnosis in BCS-eligible NHW women.
Unexpectedly, we found that BCS-eligible women with ADRD were more like to use mammography than those without ADRD. Although no studies have examined mammography use in BCS-eligible ADRD patients, previous research has shown that mammography use is lowered in BCS-eligible, cognitively impaired women [28]. Our finding of an overall positive association between ADRD diagnosis and mammography use appears to be inconsistent with research on the cognitively impaired. However, our interaction analysis revealed that this observation was driven by a similar positive association between ADRD diagnosis and mammography use in BCS-eligible Hispanic women only. Lower or same likelihood of mammography use was observed in NHW, NHB, or NHO women with ADRD. As undergoing BCS for ADRD patients is usually decided through a shared decision-making process involving the physicians, patients, and their caregivers, more future studies are needed to confirm the observed racial-ethnic and ADRD disparities in mammography use and explore the psychosocial mechanisms behind these disparities. Considering ADRD patients have a shortened life expectancy and reduced quality of life, they were less likely to benefit from BCS. Our study provided the first evidence that ADRD patients might be undergoing BCS without detailed guidelines to maximize benefits and avoid harms As a clarification, our study does not discourage the use of BCS among ADRD patients. We suggest further studies to improve the benefit of BCS for ADRD patients in different stages. On the one hand, ADRD patients were shown to have good health baselines, i.e., low CCI according to our study cohort, qualifying them as a beneficial group of BCS. On the other, the estimated survival of diagnosed ADRD patients was around 3 to 8 years [29,30], potentially making them less beneficial for BCS.
Our study has a few strengths. A major strength is the use of real-world data in a large CRN in the national PCORnet. The rich data in OneFlorida+ allowed us to accurately identify our study population of ADRD and non-ADRD patients, their detailed medical history, and mammography use. Further, we used a prospective study design that examined mammography use within two years after ADRD diagnosis. Our study also has a few limitations to note. Due to the observational nature of the study, our results do not support any causal relationship between ADRD diagnosis and mammography use. In addition, we were unable to control for some potentially important covariates such as education or income due to the limitations of the EHR data. However, the SVI contains components that cover many of these variables at the fine-grained ZIP code level. Moreover, our results may not reflect the BCS use for all ADRD patients. As ADRD is usually diagnosed at later stages, our results may not reflect BCS use among early staged ADRD patients.

Conclusions
In summary, this study was the first to report the impact of ADRD diagnosis and race-ethnicity on mammography use in BCS-eligible women using real-world data. Our finding that BCS-eligible women, especially Hispanic women, with ADRD were more like to use mammography suggests that ADRD patients might be getting screened for BCS despite a lack of clear benefits. More future studies are needed to explore the psychosocial mechanisms behind these disparities in mammography use.     Table A2. ICD code set for CCI calculation.

Disease Code Type Code
Any malignancy, includes leukemia and lympyoma ICD-10