The Women’s Circle of Health Follow-Up Study: A population-based longitudinal study of Black breast cancer survivors in New Jersey

Abstract

Purpose:

The Women’s Circle of Health Follow-Up Study is an ongoing longitudinal study of African American/Black breast cancer survivors in New Jersey, specifically designed to evaluate the impact of obesity and related comorbidities on breast cancer survival and health-related quality-of-life in this understudied population. Here, we describe our recruitment and data collection methods, and compare characteristics of the overall cohort and the subcohort with follow-up data.

Methods:

Newly diagnosed breast cancer cases have been recruited into the study since 2006. Pre-diagnosis data on relevant factors and a saliva sample are collected during an in-person interview within 12 months from diagnosis. In 2013, we began active follow up by recontacting participants annually, including two home visits at approximately 2 and 3 years post-diagnosis, during which blood samples are collected. Mortality outcomes (all-cause and breast cancer-specific mortality) are ascertained through linkage with New Jersey State Cancer Registry files. We expect to assemble a cohort of over 2,000 Black breast cancer survivors with at least 800 of them having detailed post-diagnosis data.

Results:

Distribution of sociodemographic characteristics, body mass index, comorbidities, clinicopathologic characteristics, and treatment modalities were very similar between those in the full cohort and the subset with follow-up data and blood samples. Obesity (>50%), hypertension (>58%) and diabetes (22%) were common in this population.

Conclusions and Implications for Cancer Survivors:

This ongoing longitudinal study represents a unique resource to better understand breast cancer outcomes, patient-reported symptoms, and health-related quality of life among Black breast cancer survivors.

INTRODUCTION

Even after controlling for stage and grade, African American/Black (referred to hereafter as Black) women have higher breast cancer mortality than any other racial/ethnic group (1). Poorer prognosis after a breast cancer diagnosis among Black women is likely multifactorial, including diagnosis with tumors that exhibit features indicative of more aggressive phenotypes and more advanced stage at diagnosis and less access and adherence to optimal care (2). In addition to those factors, obesity and related-comorbidities have been shown to be important contributors to racial disparities in breast cancer survival (3, 4). While Black women are more likely to be diagnosed with triple negative breast cancer, which is aggressive and difficult to treat, most of the observed disparities in outcomes were observed for hormone receptor (HR) positive tumors (2), which can be treated with endocrine therapy. Notably, HR+ breast cancer is more strongly related to obesity among postmenopausal women (5).

While obesity is a global concern, Black women have higher prevalence of obesity (body mass index, BMI ≥30 kg/m²) and central obesity (waist circumference, WC ≥88 cm) than any other racial/ethnic group in the U.S. (6, 7), with prevalence rates projected to reach 70.7% and 90.8%, respectively, by 2020 (8). There is growing evidence, mostly from studies in White women (9), showing that obesity increases breast cancer mortality (10), risk of recurrence (9), contralateral breast cancer and second cancers (11), as well as treatment-related side-effects such as lymphedema (4, 12). Obese breast cancer patients tend to present with more advanced disease, are more likely to receive suboptimal treatment, and are more likely to be chemo-resistant (3, 4). Obesity can also contribute to diabetes, hypertension, and cardiovascular disease, which are more common among Black cancer survivors (13). Diabetes and hypertension may be particularly important in explaining disparities in all-cause mortality among Black breast cancer survivors (14–16). Although obesity-related co-morbidities may explain 30–50% of the survival disparity in breast cancer patients (14–16), management of these conditions may be overlooked in survivorship care, leading to poorer outcomes (13, 17). Previous studies examining the role of comorbidities on breast cancer outcomes in Black women have been retrospective and based on surveillance and healthcare administration data (14–16), and have not assessed the impact of management and control of comorbid conditions on breast cancer treatment and outcomes. This is particularly important given emerging evidence that specific medications used to treat common obesity-related comorbidities may have independent associations with improved breast cancer outcomes (18–20).

To address gaps in the current literature on these important issues in Black breast cancer survivors, we designed the Women’s Circle of Health Follow-Up Study. Here we discuss our methodology, our experience with recruitment, data and biospecimen collection, and follow-up of this unique ongoing cohort of Black breast cancer survivors. We also present descriptive characteristics for the study population, as well as the sub-cohort for whom active follow-up has been conducted.

METHODS

History of the study and overall study design.

The Women’s Circle of Health Follow-Up Study was built upon the infrastructure of two related studies (Figure 1). The first one, the Women’s Circle of Health Study (WCHS), was a population-based case-control study established in 2003 that recruited newly diagnosed Black and White women with breast cancer and matched controls. Participants were initially recruited from the New York City metropolitan areas, and beginning in 2006, from 10 counties in New Jersey (21–36). The WCHS aimed to evaluate risk factors for aggressive and early onset breast cancer. The second study started in 2007, as a linked study with a goal to acquire and abstract medical record data from WCHS cases in New Jersey to study breast cancer treatment disparities (37–41). Leveraging these two studies, and with additional funding to expand data collection and conduct follow-up, the Women’s Circle of Health Follow-Up Study was launched in 2013, as an ongoing population-based longitudinal study of Black breast cancer survivors in 10 counties in New Jersey (Figure 2). As mentioned in more detail below, after 2013, we followed similar methods to recruit and interview participants but added data elements relevant to survivorship and study goals. We also added annual follow-up, including two follow-up home visits after the initial interview, blood pressure measurements in each visit, and blood collection in two follow-up visits, as well as data linkage with the New Jersey State Cancer Registry (NJSCR) files. Specifically, data and biospecimens (saliva) are collected during home visits at the initial interview, within 12 months after diagnosis (median: 9 months) and at two follow-up interviews with collection of blood samples, approximately 2 and 3 years after diagnosis. After each visit, participants are provided with a $50 gift card. Annual phone calls at 4 and 5 years post-diagnosis are conducted to update information; those completing calls receive a $20 check by mail. As an additional incentive, cohort participants are mailed annual newsletters with updates about study personnel, findings, and relevant information about breast cancer that could be of interest to them. In addition to keeping participants engaged in the study, mailed newsletters help to keep participant addresses up to date, minimizing loss to follow-up.

Figure 1.

The Women’s Circle of Health Follow-up Study: Study design and data collection.

Figure 2.

First primary, invasive breast cancer among Non-Hispanic Black women, age 20–74 years by county (New Jersey State Cancer Registry, 2008–2013) and counties included in the Women’s Circle of Health Follow-up Study.

Primary factors of interest in the study are shown in Figure 3. Details on recruitment, data collection, and follow-up are below. Based on current accrual rates, we expect to enroll over 2,000 women into the study, with a subset of at least 800 participants with detailed follow-up information, including patient-reported symptoms and health-related quality of life (QoL) measures, as well as detailed clinical and treatment information based on medical and pharmacy record review. Ascertainment of mortality outcomes will be essentially complete through linkage with the NJSCR files. During home interviews, women are asked to sign informed consents including medical and pharmacy records releases and permission to conduct data linkage with the NJSCR. Participants who indicate a breast cancer event (e.g., contralateral breast cancer, recurrence, metastasis) during annual phone calls are also asked to provide updated medical and pharmacy records releases. The study is approved by Institutional Review Boards at all participating institutions.

Figure 3. Primary factors under consideration in the Women’s Circle of Health Follow-Up Study.

Legend: Study aims to examine the impact of obesity and obesity-related related comorbidities, including the role of comorbidity management, medication use, and control, on breast cancer treatment, health-related quality of life (QoL) outcomes, obesity-related biomarkers evaluating mechanistic pathways, and survival outcomes within the context of multi-level factors that can affect breast cancer diagnosis, treatment, and outcomes among Black breast cancer survivors.

Recruitment and Follow-up

Black women newly diagnosed with breast cancer in 10 counties in New Jersey are identified by rapid case ascertainment by the NJSCR. The counties were selected for having the largest number of breast cancer cases in Black women, while being within a 2 hour driving distance to the Rutgers Cancer Institute of New Jersey, where data and biospecimen collection are based (Figure 2). Eligibility criteria include women who self-identify as Black, have incident histologically confirmed ductal carcinoma in-situ (DCIS) or invasive breast cancer, are aged 20–75 years at diagnosis with no cognitive incapacity, are able to speak and understand English, and have no prior history of cancer except non-melanoma skin cancer. Physician permission to contact potential participants is requested by mail for each newly identified case. Passive approval is assumed if the physician does not respond within three weeks. NJSCR staff contacts potential participants first by mail, followed by a telephone call to explain the study and to request permission to release their names to staff at the Rutgers Cancer Institute of New Jersey, who then contact each case by phone to schedule an initial interview.

As mentioned earlier, recruitment of cases has been ongoing in New Jersey since 2006 (with the first diagnosis in June 2005) for the WCHS. As shown in Figure 4 (top panel), of the 5,802 total cases identified by the NJSCR through March 2019, 813 (14.0%) women were ineligible, and 165 (2.8%) were deceased leaving 4,824 (83.1%) potentially eligible women for recruitment into the study. Physician permission to contact was available for 4,764 (98.8%) of these women. Of these, 2,437 (51.2%) agreed to be contacted by a WCHS interviewer, with 2,185 (45.9%) actively or passively refusing, and only a small proportion who could not be located (2.2%, n=105,) or interviewed (<0.1%, n=36), with one withdrawal prior to contact by WCHS staff. Of the 2,437 women who agreed to be contacted by WCHS staff, 43 (1.8%) were found to be ineligible for the study, and a final number of 2,015 (82.7%) provided written consent for participation and completed the initial interview as of March 2019.

Figure 4. Women’s Circle of Health Follow-up Study Recruitment and Follow-up: Progress as of March 2019.

^aWomen ineglible for a specific follow-up were those who have died, moved out of the NJ area, or had not yet reached their follow-up window. Across F/U 1 to F/U 4, a total of n=66 women died and n=18 moved out of the NJ area.

In 2013, we began to invite all women to participate in the Women’s Circle of Health Follow-Up Study (Figure 4, bottom panel). Of the 1,075 participants consented to the Follow-up Study by March 2019, 71.6% (706/986) and 65.0% (472/726) of those eligible for their follow-up (F/U) 1 and F/U 2 visits (± 3 months of 24 and 36 months post-diagnosis), respectively, completed their interviewed. Of those eligible for follow-up, 20.6% of F/U 1 (n=203) and 20.8% (n=151) of F/U 2 participants were passive refusers, with less than 5% (F/U 1: n=40, 4.1%; F/U 2: n=25, 3.4%) of participants actively refusing to be interviewed. At F/U 3 and F/U 4, which were telephone interviews conducted 4 and 5 years (± 3 months) post diagnosis, respectively, the interview completion rates were 72.7% (350/481) and 71.4% (237/332), respectively, with combined active and passive refusal rates of 15% (72/481) and 4.5% (15/332), respectively. Our loss to follow-up rates were below 5% at F/U 1 and F/U 2, and were slightly higher at later follow-ups, ranging between 7 and 10%. Approximately 6% of participants were deceased so they could not be interviewed, and a handful of participants (n≤18) had moved out of the NJ catchment area for the study. The remaining eligible participants are still in the process of being contacted for their interviews, and thus with increased follow-up time the total proportion interviewed is expected to be higher. Regardless of the completion status of follow-up visits or calls, linkage with the NJSCR will provide mortality outcomes on essentially the full cohort, even for those moving out of New Jersey.

DATA AND BIOSPECIMEN COLLECTION

We prioritized hiring experienced Black interviewer/phlebotomists to conduct the home visits. All interviewers undergo extensive standardized training supported by procedure manuals for each component of the home interview. In addition they complete training on Human Subjects research, research integrity and the responsible conduct of research, and on safe biospecimen collection, handling, and shipping practices before they are allowed to start conducting interviews. Procedures are reviewed periodically at staff meetings for quality control. Computer-assisted interviewing is used to facilitate data collection by supporting better interview flow with programmed skip patterns, prompting possible errors in data entry, and minimizing missing values. As an additional quality control measure, interviews are audio recorded with participant permission.

Initial Home Visit

The questionnaire administered during the home interview collects information on a wide range of factors, including country of origin, race/ethnicity, education, income, health insurance, as well as established and suspected risk factors for breast cancer, including family history, reproductive and menstrual history, hormone use, diet and dietary supplement use, alcohol intake, physical activity, smoking history, medical history, and medication and supplement use one year before diagnosis. Physical activity and sitting time are assessed using the Black Women’s Health Study (BWHS) Physical Activity Questionnaire (42, 43). Similarly, dietary intake is assessed using the modified NCI-Block Food Frequency Questionnaire (FFQ) used by the BWHS (44, 45), with minor modifications. For most of these factors, participants are asked to report their behavior/experiences one year before diagnosis. Women are also asked to report their weight and height one year before diagnosis, and at several time points during their life, including at age 20 years.

With the expansion of WCHS in 2013 to include a focus on survivorship, we started to collect detailed information on relevant obesity-related comorbidities, which are important among breast cancer survivors, and/or more prevalent among Blacks (e.g., hypertension, diabetes, cardiovascular disease, asthma, sleep apnea, and osteoporosis). For each condition, we ask about age at diagnosis, medications used, adherence to medications, and reasons for not taking prescribed medications. We also added questions about experience with discrimination (46), sleep quality before diagnosis using the Pittsburgh Sleep Quality Index (47), and health-related QoL as assessed by the SF-36 Questionnaire (48).

During the interview, the names of all healthcare providers who provided care from one year prior to diagnosis to the time of the interview are obtained, along with names of hospitals where care was provided, and pharmacies used to fill prescription medications. Signed releases for medical and pharmacy records are collected, and sent to all providers and pharmacies requesting a copy of the patient’s records. Records are requested from primary care physicians, specialists being seen to manage comorbid conditions (e.g. cardiologists, endocrinologists, etc.), as well as surgeons and medical and/or radiation oncologists involved in breast cancer care. A major advantage of conducting in-person home interviews is that if participants take prescription medications and/or dietary supplements, this information can be directly obtained from medication and supplement bottles. This is a major strength as it improves the quality of these data.

Anthropometric measurements are taken at each home interview following a standardized protocol (24). Participants are asked to wear light clothing. Interviewers are trained to measure the following: weight, standing height, neck, waist and hip circumferences. Body composition (lean and fat mass) is measured using a bioelectrical impedance analysis machine Tanita® TBF-300A scale. Waist and hip circumferences are recorded twice and if the difference between the measures is more than 2 cm, a third measurement is taken. Results are averaged for analyses. Height and neck circumferences are recorded only once to reduce subject discomfort. BMI calculated from self-reported weight and height data and from measured weight and height data were found to be highly correlated both when considered as continuous variables (Intraclass correlation coefficient=0.96) and when standard BMI categories were used (kappa=0.81) (49).

Blood pressure measurements are taken during home visits according to the standardized protocols provided by the American Heart Association (50, 51). In brief, blood pressure is measured in a seated position after at least 5 minutes of rest using a clinically validated automated blood pressure monitor (Omron HEM-907XL). A set of 2 blood pressure readings are taken and averaged for analyses. Women are excluded from a blood pressure measurement if they have a condition that could potentially cause them harm or discomfort or would prevent accurate blood pressure measurement. Participants are asked about conditions that might potentially affect their blood pressure measurement, such as recent consumption of food, coffee, cigarettes, alcohol, and recent vigorous activity.

Saliva samples are collected using Oragene Kits (DNA Genotek, Inc). Methods are described below.

Follow-up Home Visits

Participants who consented to be interviewed annually are sent a reminder letter when they are eligible for their Follow-Up 1 (F/U 1) or Follow-Up 2 (F/U 2) home visit. Study staff reach out to them via telephone shortly after the reminder is sent to schedule a date and time for their follow-up study interview. F/U 1 and F/U 2 home visits take place at approximately 2 and 3 years (± 3 months) post-diagnosis, respectively. Similar to the initial home visit, questionnaires are administered using computer-assisted interviewing, and anthropometric measurements (including body size and body composition) and blood pressure readings are taken following the same procedures as in the initial home visit. New medical and pharmacy records releases are obtained. Blood samples are also collected at F/U 1 and F/U 2 visits, as explained in more detail below.

Questionnaires.

At each follow-up visit, information is updated on lifestyle factors (alcohol intake, smoking, physical activity, diet and supplement use), and weight changes. Questions are also asked about receipt of a survivorship care plan and schedule for followup care at the end of treatment, medical history, including medication use, any other cancer event (e.g., breast cancer recurrence, contralateral breast cancer, or other cancers over the past year), patient-reported symptoms, and health-related QoL measures. Questions on whether participants received a schedule for their follow-up and/or a survivorship care plan were taken from the National Health Interview Survey (NHIS) Cancer Control supplement on survivorship (https://www.cdc.gov/nchs/nhis/index.htm). The Functional Assessment of Cancer Therapy, Breast (FACT-B+4), is used to assess QoL (52). Questions on financial stress and access to health care are included, which ascertains the type of health insurance coverage, reasons for delay of medical care, and the degree that cancer caused financial problems. We use the Quick Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire for lymphedema (53–55). Because obesity and chemotherapy can both increase levels of proinflammatory cytokines (56, 57) thought to be responsible for sickness symptoms reported by breast cancer patients, including pain (58), fatigue (56), and sleep disturbance (59–63), we use several validated instruments to assess cytokine-related symptoms. These include the 13-item Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue scale (64), the Pittsburgh Sleep Quality Index (47), and the 10-item Cohen’s Perceived Stress Scale (65, 66) to measure psychological distress. We also assess health-related QoL again using the SF-36 scale (48). Given that Black cancer survivors report spirituality and social support as important determinants of QoL (67–69), we administer the FACIT- Spiritual Well-being questionnaire (70), and include questions about church attendance and participating in cancer support groups. Spirituality may be particularly relevant among Black women because of its potential contribution to treatment delays (71, 72). Questions assessing fear of recurrence are taken from the 2010 NHIS supplement on survivorship, and resilience is assessed using the Brief Resilient Coping Scale (73).

Data Collection for Annual Follow-Up Calls

At approximately 4 and 5 years post-diagnosis, we conduct shorter computer-assisted interviews by telephone to update information on weight changes, diet and supplement use, smoking, alcohol, physical activity, medical history, including comorbidities and medication use, new cancer events and treatments, adherence to endocrine therapy if prescribed, survivorship follow-up plan, and access to care. We also assess health-related QoL again using the SF-36 scale (48). These interviews are also recorded for quality control.

Biospecimen Collection and Processing

Saliva samples are collected using Oragene™ DNA Self-Collection Kits (DNA Genotek, Inc) during the initial home visit. Each specimen is labeled with the participant’s unique study ID number and kept at room temperature at the Rutgers Cancer Institute of New Jersey until shipped on a monthly basis to Roswell Park Comprehensive Cancer Center Laboratory. Upon receipt at Roswell Park, saliva samples are stored at room temperature until processed for DNA extraction. DNA is kept at −80°C for future analysis. We have requested saliva samples since 2006 when we started recruiting for the WCHS in New Jersey. Only a small proportion of participants have refused to donate a sample (2%), therefore, DNA is available for almost the full cohort.

With the expansion of the study in 2013 to start collecting follow-up data, we added blood collection to F/U 1 and F/U 2 visits after treatment with chemotherapy and/or radiation had been completed. At the end of these visits, our trained phlebotomist interviewers collect non-fasting blood samples, drawn into 2 red tops and 2 lavender top vacutainers. Blood samples labeled with the participants’ study ID numbers are transferred shortly after the end of the interview to Rutgers Cancer Institute of New Jersey Laboratories (Biospecimen Repository Service, BRS) for processing. Blood tubes are centrifuged and separated into serum, plasma, and formed elements (i.e. buffy coat and red blood cells), and aliquoted into labeled color-coded cryovials and stored at − 80°C. Samples are sent monthly to the Roswell Park Comprehensive Cancer Center DataBank and Biorepository for storage (74) until they are assayed. F/U 1 collections will be assayed for obesity-related biomarkers, while samples collected at F/U 2 will be stored for future use. As of March 2019, 73% of the women who completed F/U 1 (512/706) and 66% (309/472) of those who completed F/U 2 donated a blood sample.

Breast Cancer Treatment Information

Information on breast cancer treatment is obtained from several sources, including medical and pharmacy record review, patient’s self-report during in-person interviews, and linkage with the NJSCR files.

Medical records acquisition and review.

Medical records related to breast cancer diagnosis and treatment have been obtained since study recruitment began in NJ in 2006 as part of our study on treatment disparities (37–41). Virtually all participants (98%) agree to sign the medical records releases. There has been also great cooperation from healthcare providers, as medical records have been received for approximately 87% of participants as of March 2019, with 9% of records still pending, and a refusal rate of less than 4%. Data abstraction includes details on diagnosis, type of surgery, surgical margins, chemotherapy (regimen, dose schedule, agents, number of cycles, dosing, dates of administration), radiation therapy (number of sessions, dose per session and dates), endocrine and monoclonal antibody therapy (agent, dose and date per cycle), comorbidities and treatments prescribed, clinical and pathological characteristics from pathology reports, as well as recurrence, metastasis and vital status.

Information from medical records will be used to ascertain adherence to breast cancer treatment guidelines as defined by the National Comprehensive Cancer Network (NCCN), as well as chemotherapy dosing and dose reduction, treatment delays, and early discontinuation. We also collect information on treatment-related toxicities, focusing on myelosuppression (neutropenia), which is the major dose-limiting factor for chemotherapy (75) and use the NCI’s Common Terminology Criteria for grading chemotherapy-related toxicity.

Pharmacy records acquisition and abstraction.

At the initial home visit, participants are asked to provide the names and addresses of all pharmacies and/or mail order prescription services used to fill their prescription orders from one year before their breast cancer diagnosis to the date of the interview. Permission is requested to contact the participant’s individual pharmacy and/or mail order company to release their prescription refill information. Pharmacy information and release forms are updated at each home visit. The pharmacy refill data contains information on service date, National Drug Code (NDC), drug name, drug class, and drug strength, pill dose, quantity dispensed, and days’ supply, which will be used to assess medication adherence for comorbidities. Most participants completing home visits agree to sign pharmacy records releases (96%) and records are obtained for most participants signing release forms, with 87%, 92%, and 87% of requested pharmacy records received (as of March 2019) for women completing their baseline, F/U 1, and F/U 2 interviews, respectively.

Assessing Management and Control of Obesity-Related Comorbidities

Information on obesity-related comorbidities (primarily focusing on diabetes, hypertension, hypercholesterolemia, and heart disease) and their management and control is obtained through several sources, including through in-person interviews with participants to collect information on existing diagnoses and age at diagnosis of any of the comorbidities of interest, medication being taken for each comorbidity, adherence to treatment, and reasons for not taking the prescribed treatment. In addition, medical records are obtained from multiple providers, as described above, allowing us to capture major comorbid conditions, treatment approaches, and control of symptoms. Pharmacy records are also obtained and abstracted. Information is being collected from one year before diagnosis through the corresponding follow-up visit. This information is complemented by our blood pressure measurements at each home visit, as described above.

To assess disease severity for diabetes and hypertension, we use measures developed and used in the Medical Outcomes Study (MOS) (76). Guidelines from the Eighth Joint National Committee (JNC 8) and The American Diabetes Association’s Standards of Medical Care in Diabetes is used to assess management of hypertension and diabetes, respectively (77, 78). For hypercholesterolemia, we use the NHLBI Adult Treatment Panel III (ATP III) classification (79). Disease-related medical complications are recorded. Severity of chronic renal disease, a potential complication of diabetes that can affect chemotherapy dosing and toxicity, is evaluated by categorizing disease into five stages based on glomerular filtration rate (GFR, mL/min per 1.73m²) according to clinical guidelines (80). We also review records for markers of physiologic control, including HbA1c levels for diabetes, systolic and diastolic blood pressure measurements for hypertension, blood lipid levels for management of hypercholesterolemia, and serum creatinine levels for renal disease. Recommended and prescribed treatments are also recorded.

Outcomes ascertainment.

Major outcomes of interest include breast cancer-specific mortality, competing cause mortality, and all-cause mortality, which will be available for the full cohort. In the full cohort (which includes women diagnosed as early as 2005), vital status, causes and date of death will be ascertained through linkage with NJSCR files, which are updated annually using various sources including state and the National Death Index, hospital discharge files, Medicare and Medicaid files, and Social Security Administrative Data. Hospitals also submit annual vital status updates, and NJSCR uses information from e-path to update vital status. A potential issue in cohort studies can be loss to follow-up. However, as mentioned earlier, linkage with the NJSCR will allow for essentially complete ascertainment of mortality outcomes, even if cases move to other states.

RESULTS

While the study is ongoing, we now have collected sufficient data to be able to compare distributions for major prognostic factors for women in the full WCHS cohort compared to the subset who participated in the follow-up study. Table 1 includes data up to March 2018 given delays in cleaning and compiling data from the various sources, including linkage with the NJSCR files. Details of the groups compared in Table 1 are described in detail below.

Table 1.

Selected characteristics at initial visit for participants in the Women’s Circle of Health Follow-Up Study (as of March 2018)

	Group 1: Full cohort (2005–2018) n=1719	Group 2: Invited and consented to the follow-up study (2013–2018) n=956	Group 3: Completed F/U 1 visit (2013–2018) n=530	Group 4: Donated blood during F/U 1 visit (2013–2018) n=400
	n (%)	n (%)	n (%)	n (%)
Age at diagnosis (years)
20–40	175 (10.2)	85 (8.9)	51 (9.6)	39 (9.8)
41–60	1011 (58.8)	542 (56.7)	304 (57.4)	230 (57.5)
60–75	533 (31.0)	329 (34.4)	175 (33.0)	131 (32.8)
Missing	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Education
≤High school graduate	668 (38.9)	336 (35.2)	192 (36.2)	141 (35.3)
≥Some college	1049 (61.0)	618 (64.6)	338 (63.8)	259 (64.8)
Missing	2 (0.1)	2 (0.2)	0 (0.0)	0 (0.0)
Insurance status (before diagnosis)
Medicaid	218 (12.7)	129 (13.5)	73 (13.8)	53 (13.3)
Medicare	267 (15.5)	173 (18.1)	95 (17.9)	77 (19.3)
Private	943 (54.9)	499 (52.2)	275 (51.9)	201 (50.3)
Other	93 (5.4)	50 (5.2)	28 (5.3)	25 (6.3)
Uninsured	188 (10.9)	97 (10.2)	56 (10.6)	42 (10.5)
Missing	10 (0.6)	8 (0.8)	3 (0.6)	2 (0.5)
Family history of breast cancer (first-degree relative)
No	1414 (82.3)	784 (82.0)	434 (81.9)	330 (82.5)
Yes	305 (17.7)	172 (18.0)	96 (18.1)	70 (17.5)
Missing	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Smoking status (before diagnosis)
Never	1009 (58.7)	566 (59.2)	307 (57.9)	232 (58.0)
Former	421 (24.5)	231 (24.2)	130 (24.5)	100 (25.0)
Current	288 (16.8)	158 (16.5)	93 (17.6)	68 (17.0)
Missing	1 (0.1)	1 (0.1)	0 (0.0)	0 (0.0)
BMI (kg/m2 - measured during home visit at initial interview)
Underweight (<18.5)	8 (0.5)	3 (0.3)	3 (0.6)	2 (0.5)
Normal (18.5–24.9)	269 (15.7)	141 (14.8)	80 (15.1)	61 (15.3)
Overweight (25.0–29.9)	473 (27.5)	267 (27.9)	149 (28.1)	123 (30.8)
Obese I (30.0–34.9)	434 (25.3)	246 (25.7)	132 (24.9)	97 (24.3)
Obese II (35.0–39.9)	278 (16.2)	163 (17.1)	90 (17.0)	64 (16.0)
Obese III (≥40.0)	217 (12.6)	110 (11.5)	66 (12.5)	45 (11.3)
Missing	40 (2.3)	26 (2.7)	10 (1.9)	8 (2.0)
Comorbidities before diagnosis (self-report)
Hypertension	740 (43.1)	557 (58.3)	310 (58.5)	231 (57.8)
Diabetes	280 (16.3)	212 (22.2)	118 (22.3)	86 (21.5)
Heart disease or angina	84 (4.9)	71 (7.4)	39 (7.4)	27 (6.8)
Osteopenia or osteoporosis	45 (2.6)	45 (4.7)	23 (4.3)	17 (4.3)
Arthritis	267 (15.5)	252 (26.4)	111 (20.9)	76 (19.0)
Asthma	147 (8.6)	136 (14.2)	68 (12.8)	46 (11.5)
Sleep apnea	83 (4.8)	80 (8.4)	35 (6.6)	26 (6.5)
TUMOR CHARACTERISTICS
Behavior
DCIS	334 (19.4)	199 (20.8)	113 (21.3)	95 (23.8)
Invasive	1385 (80.6)	757 (79.2)	417 (78.7)	305 (76.3)
Unknown/missing	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Stage
In situ	285 (16.6)	150 (15.7)	107 (20.2)	91 (22.8)
Localized only	674 (39.2)	315 (33.0)	234 (44.2)	182 (45.5)
Regional	449 (26.1)	221 (23.1)	146 (27.5)	99 (24.9)
Distant site(s)/node(s) involved	64 (3.7)	35 (3.7)	17 (3.2)	11 (2.8)
Unknown/missing	247 (14.4)	235 (24.5)	26 (4.9)	17 (4.3)
Grade
I-Well differentiated	168 (9.8)	110 (11.5)	70 (13.2)	56 (14.0)
II-Moderately differentiated	617 (35.9)	356 (37.2)	181 (34.2)	132 (33.0)
III-Poorly differentiated	722 (42.0)	397 (41.5)	227 (42.8)	170 (42.5)
IV-Undifferentiated/anaplastic	7 (0.4)	1 (0.1)	0 (0.0)	0 (0.0)
Unknown/missing	205 (11.9)	92 (9.6)	52 (9.8)	42 (10.5)
ER
Positive	1224 (71.2)	690 (72.2)	382 (72.1)	292 (73.0)
Negative	474 (27.6)	256 (26.8)	145 (27.4)	106 (26.5)
Unknown/missing	21 (1.2)	10 (1.1)	3 (0.6)	2 (0.5)
PR
Positive	947 (55.1)	538 (56.3)	295 (55.7)	228 (57.0)
Negative	746 (43.4)	406 (42.5)	230 (43.4)	169 (42.3)
Unknown/missing	26 (1.5)	12 (1.3)	5 (0.9)	3 (0.8)
HER2
Positive	290 (16.9)	160 (16.7)	88 (16.6)	61 (15.3)
Negative	1146 (66.7)	645 (67.5)	365 (68.9)	275 (68.8)
Unknown/missing	283 (16.5)	151 (15.8)	77 (14.5)	64 (16.0)
Triple negative (ER-/PR-/HER2-)
No	1127 (65.6)	636 (66.5)	354 (66.8)	262 (65.5)
Yes	307 (17.9)	167 (17.5)	97 (18.3)	73 (18.3)
Unknown/missing	285 (16.6)	153 (16.0)	79 (14.9)	65 (16.3)
TREATMENT INFORMATION (Self-report)
Type of surgery
Lumpectomy	951 (55.3)	510 (53.4)	329 (62.1)	247 (61.8)
Mastectomy	604 (35.1)	275 (28.8)	175 (33.0)	136 (34.0)
Missing	145 (8.4)	132 (13.8)	9 (1.7)	7 (1.8)
Chemotherapy
Yes	851 (49.5)	429 (44.9)	272 (51.3)	202 (50.5)
No	723 (42.0)	395 (41.3)	249 (47.0)	191 (47.8)
Missing	145 (8.4)	132 (13.8)	9 (1.7)	7 (1.8)
Radiation therapy
Yes	879 (51.1)	455 (47.6)	284 (53.6)	223 (55.8)
No	694 (40.4)	368 (38.5)	237 (44.7)	170 (42.5)
Missing	146 (8.5)	133 (13.9)	9 (1.7)	7 (1.8)
Endocrine therapy
Yes	739 (43.0)	397 (41.5)	267 (50.4)	210 (52.5)
No	830 (48.3)	423 (44.3)	252 (47.6)	182 (45.5)
Missing	150 (8.7)	136 (14.2)	11 (2.1)	8 (2.1)
FIRST COURSE TREATMENT (Cancer Registry data)
Type of surgery
No surgery	130 (7.6)	74 (7.7)	36 (6.8)	27 (6.8)
Lumpectomy	978 (56.9)	572 (59.8)	316 (59.6)	238 (59.5)
Subcutaneous or total mastectomy	361 (21.0)	200 (20.9)	121 (22.8)	92 (23.0)
Radical mastectomy	242 (14.1)	105 (11.0)	56 (10.6)	43 (10.8)
Mastectomy, NOS	1 (0.1)	1 (0.1)	0 (0.0)	0 (0.0)
Surgery, NOS	1 (0.1)	4 (0.4)	0 (0.0)	0 (0.0)
Unknown/missing	6 (0.4)	74 (7.7)	1 (0.2)	0 (0.0)
Chemotherapy
Yes, received	793 (46.1)	442 (46.2)	246 (46.4)	174 (43.5)
Not received	776 (45.1)	433 (45.3)	243 (45.9)	193 (48.3)
Recommended, not received	59 (3.4)	41 (4.3)	22 (4.2)	19 (4.8)
Recommended, unknown if received	41 (2.4)	20 (2.1)	5 (0.9)	4 (1.0)
Unknown/missing	50 (2.9)	20 (2.1)	14 (2.6)	10 (2.5)
Radiation therapy
Yes	777 (45.2)	460 (48.1)	255 (48.1)	194 (48.5)
No	819 (47.6)	408 (42.7)	235 (44.3)	179 (44.8)
Recommended, but unknown if received	78 (4.5)	63 (6.6)	22 (4.2)	14 (3.5)
Unknown/missing	45 (2.6)	25 (2.6)	18 (3.4)	13 (3.3)
Endocrine therapy
Yes	617 (35.9)	406 (42.5)	228 (43.0)	171 (42.8)
No	877 (51.0)	431 (45.1)	241 (45.5)	186 (46.5)
Recommended, unknown if administered	141 (8.2)	86 (9.0)	39 (7.4)	28 (7.0)
Unknown/missing	84 (4.9)	33 (3.5)	22 (4.2)	15 (3.8)

Group 1 is the full cohort recruited up to March 2018, including all breast cancer cases in Black women who completed the study since WCHS began recruitment in NJ (first diagnosis was in 2005 with ongoing recruitment). All participants in this group have pre-diagnosis data (one year before diagnosis) and some data at the time of interview (e.g, body measurements, self-reported breast cancer treatment). For most participants, DNA samples and detailed medical records related to their breast cancer diagnosis and treatment are available. Mortality outcomes and first-line cancer treatment will be also available for the full cohort through data linkage with the NJSCR.

Group 2 is a subset of group 1 and includes women who were invited and consented to being part of the follow-up study. All participants in the full cohort (group 1) within the eligible F/U1 window were invited to participate in the follow-up study starting in 2013, when we expanded the WCHS to include survivorship and survival outcomes. As stated above, more than 95% of them consented to participate in the follow-up study.

Group 3 includes women in group 2, who completed the first follow-up visit a year later after consenting to the follow-up study during the initial home visit. In addition to data and specimens collected in the full cohort (group 1) as described above, these participants have post-diagnosis data, including data from medical and pharmacy records on comorbidities and their management and control.

Group 4 includes participants in group 3 who in addition to having pre- and post-diagnosis data and detailed medical and pharmacy records, donated a blood sample during their first visit. Overall, distributions of demographic, lifestyle, BMI, prevalence of comorbidities, clinicopathological characteristics and treatment modalities are very similar when comparing the full cohort with the subcohorts who consented and participated in the follow-up study, and provided a blood sample. There were some indications that compared to the full cohort, women who completed the F/U 1 visit and donated blood were slightly older, more educated, more likely to be diagnosed with DCIS than invasive disease, less likely to have received chemotherapy, more likely to have received radiation, and more likely to be receiving endocrine therapy. The magnitude of these differences, however, were generally small, indicating that selection bias is not a major concern

It is also worth noting that while the prevalence of comorbidities appears to be lower in the full cohort (group 1, Table 1), this may be partly due to the fact that we added survey questions ascertaining comorbid conditions in WCHS while the study was already ongoing for over six years. Once the medical records data are fully abstracted and cleaned, we will be able to validate the self-reported data on medical history.

As shown in Table 1, prevalence rates of obesity and related comorbidities are high in the cohort, as we anticipated, confirming the need to address their impact in this population. Over 50% of participants were obese at the initial visit, and rates were essentially the same across the different subgroups in Table 1 (e.g., 54.1% of the full cohort, 54.4% of those completing the F/U 1 visit, and 51.6% of those donating blood in F/U 1 visit).

Due to the lack of informed consent among non-participants, we were unable to compare characteristics of non-participants with those in the study cohort. However, we were able to compare the distributions of tumor characteristics among those participating with all eligible breast cancer cases identified by NJSCR in the same counties up to March 2018 (Figure 5). Distributions were very similar for tumor stage and grade among women participating at each study visit, suggesting that breast cancer tumor characteristics among women in the WCHS cohort is representative of all Black women diagnosed with breast cancer in New Jersey.

Figure 5. Tumor stage and grade distribution (%) of women participating in the Women’s Circle of Health Follow-Up Study vs. all eligible in the target areas in New Jersey (as of March 2018)*.

*Based on NJSCR data for Black female breast cancer diagnosed 2014–2016, age at diagnosis 30–74, without prior cancers and residing in one of the 10 counties in the study.

DISCUSSION

The Women’s Circle of Health Follow-Up Study is a unique study specifically designed to comprehensively understand and identify issues highly relevant to Black women after a diagnosis of breast cancer. Based on current recruitment, we expect to assemble a cohort of over 2,000 Black breast cancer survivors, of whom at least 800 will have detailed follow-up data on lifestyle factors, medical history, access and coordination of care, and patient-reported symptoms and health-related QoL trajectories. A unique feature of the study is the comprehensive approach to building participants’ clinical histories through multiple sources of information (self-report through in-person interviews, detailed medical and pharmacy records, and data from the state cancer registry linkage). An additional important strength is that the study is population-based, covering a wide geographic area with diverse Black populations that involve hundreds of healthcare providers in 10 counties in New Jersey, rather than being limited to a single hospital or cancer center.

Recruiting and retaining minority participants in cohort studies can be challenging, particularly when trying to recruit Black women newly diagnosed with breast cancer, who tend to be diagnosed at a more advanced stage compared to White women and often have to deal with other coexisting conditions. In our study, over 50% of the cases identified by the NJSCR as eligible for participation agreed to be contacted by our interviewers, with 43% (2015/4721) of all eligible cases eventually consenting and completing the initial home interview as of March 2019. The participation rate in the Women’s Circle of Health Follow-Up Study is slightly higher than the overall participation rate of 37% observed in the multi-racial/ethnic Pathways Study, a similarly comprehensive breast cancer cohort designed to longitudinally evaluate the impact of lifestyle factors on breast cancer prognosis (81). In the Breast Cancer Family Registry in Northern California, which like the Women’s Circle of Health Follow-Up Study used population-based cancer registries to screen for potentially eligible participants, the enrollment rate of Blacks was slightly higher at 48% (831 enrolled out of 1,713 screened), but this participation rate did not include the ~15% of women originally identified by the cancer registry who did complete the screening interview (82). Accounting for this would bring the overall participation rate to 41%, similar to rates we observed for Black women exclusively. The Life After Cancer Epidemiology (LACE) Study, a multi-racial cohort of early stage breast cancer survivors focused on modifiable risk factors affecting quality of life and long-term survival, primarily recruited from the Kaiser Permanente Northern California Cancer Registry and the Utah Cancer Registry, reported a slightly higher enrollment rate of 46% among those identified as eligible, but did not ask participants to consent to multiple follow-up visits (83). Overall, the similar participation rates across studies designed to establish breast cancer cohorts focused on disease prognosis suggests that Black breast cancer survivors are at least as likely to participate in observational cohort studies as other racial/ethnic groups.

In this cohort, the overall initial willingness to donate a saliva sample was very high near the time of diagnosis (~98%). Blood collection rates were lower, being closer to approximately 73% and 66% at 2 and 3 years post-diagnosis, respectively. Similar to our biospecimen collection rates at F/U1, the population-based Breast Cancer Family Registry in Northern California, which also identifies eligible participants through cancer registries and collects biospecimens using home visits, reported a high level of willingness to donate biospecimens among Black women, with approximately 80% willing to provide a blood sample and 10% choosing to provide a saliva sample instead (82). Findings from both studies would suggest a high overall level of willingness to donate biospecimens when home visits are utilized, although willingness may be somewhat lower for blood compared to saliva samples (82). Based on data from our study, however, willingness to provide a blood sample among Black cancer survivors may decrease over time if repeated collections are sought.

Our study demonstrates the feasibility of conducting longitudinal studies in Black cancer survivors and, given the paucity of data in this population, it should be a priority. Through our efforts in recruiting Black breast cancer survivors for over 12 years we have learned a few lessons. Incentives work and while a few women were happy to participate without receiving incentives, many were clearly motivated to receive the gift card. Many expressed an interest in our newsletters and frequently asked us when we were going to send the next one as they wanted to be informed about the study. Study protocols have to be modified sometimes because strict protocols that work in a clinical setting do not always work in the field (e.g, not having a chair for the patient to sit to measure her blood pressure). It is important that interviewers are mature and adaptable enough to solve problems in the field and can conduct themselves professionally when faced with challenges and unanticipated circumstances. Cultural-competence/awareness is also very important in developing a relationship of mutual respect and trust with participants as they enroll into the study. Our interviewers were from the same communities where the interviews were being conducted and most of them were also Black women, which undoubtedly contributed to our success in recruitment and retaining participants. Overall, our study demonstrates that recruitment and biospecimen collection is feasible in a population-based study of Black breast cancer survivors.

In view of the rising rates of obesity, particularly among Black women, and persistent racial differences in obesity and obesity-related comorbidities in the United States, understanding how excess body fat and related comorbidities impact breast cancer care, patient-reported symptoms and health-related QoL among Black breast cancer survivors is of critical importance. Such data could represent major breakthroughs in breast cancer management and contribute to persistent inequities in breast cancer outcomes. Our study is generating important data to address these issues. It has also served as a springboard for innovative, new studies that are being led by several early stage investigators. These investigators are expanding the original objectives of the study to novel areas including neighborhood factors and the built environment in relation to cardiovascular health, molecular mechanisms contributing to more aggressive breast cancers, epigenetics and metabolomics, with an overarching goal of better understanding the multi-level, sociobiologic contributors to the increased mortality rates among Black breast cancer survivors.