The HealthStreet Cancer Survivor Cohort: a Community Registry for Cancer Research

Abstract

Purpose:

This report describes a cancer survivor cohort from a community engagement program and compares characteristics and willingness to participate in health research between the cancer survivors and non-cancer community members.

Methods:

Among 11,857 members enrolled in HealthStreet at the University of Florida (10/2011–03/2020), 991 cancer survivors were identified and 1:1 matched to control members without cancer on sex, age, and zip code. Demographics, body weight, height, social determinants of health, history of cancer, and willingness to participate in research were recorded by Community Health Workers as a part of the baseline Health Needs Assessment.

Results:

Among the cancer survivors, 71.6% were female and 19.2% lived in rural areas with a mean age of 56.7 years in females and 60.8 years in males. At baseline, 44.7% received a cancer diagnosis within 5 years, while 15.8%, more than 20 years. Cancer survivors (vs. matched non-cancer controls) were less likely to be Black (31.1% vs. 63.6%) but more likely to be divorced, separated, or widowed (49.5% vs. 41.2%), be normal/underweight (34.0% vs. 25.6%) and have health insurance (80.0% vs. 68.6%; all p<0.05). Cancer survivors versus matched controls reported higher rates of ever being in a health research study (32.4% vs. 24.9%) and interest in participating in studies ranging from minimal risk to greater-than-minimal risk.

Conclusions:

Cancer survivors from this community engagement program agnostic to cancer types and treatment are diverse in geography, race, and social determinants of health and can be a valuable resource for observational, interventional, and biospecimen research in cancer survivorship.

INTRODUCTION

Cancer survivorship has been recognized as a significant research field [1-3]. Because of the improvement of cancer treatment and survival in the past three decades, the number of people living with cancer has been increasing in U.S. communities. There were approximately 17 million cancer survivors in 2019, and the number is expected to reach 22.1 million in the next decade [4]. Various efforts have been made to form cohorts to study cancer survivorship and outcomes [5-7]. As the U.S. population is highly diverse, a better understanding of underserved populations, such as people who are minorities in race/ethnicity and those living in rural and non-metropolitan areas, is needed. In particular, social determinants of health defined by the World Health Organization as “conditions in which people are born, grow, work, live and age, and the wider set of forces and systems shaping the conditions of daily life” [8] are crucial for the goal setting, design, and implementation of cancer survivorship research [9].

Understanding research perceptions and willingness to participate in health research among community members is key to recruiting and enrolling them to research studies and trials. Consistently low minority participation in cancer research is a concern for research, public health, and equity issues [10, 11]. It has been hypothesized that people who are minorities in race/ethnicity and live in rural areas distrust research and are less willing to participate in studies [12]. However, more recent research challenged the ideology, and multiple studies showed that people minorities in race, such as African Americans, were more willing to participate in research than non-Hispanic White individuals [12-15]. In addition, a higher proportion of African Americans living in rural areas reported a high level of beliefs and trust in research and researchers than those living in urban areas [16]. The data on cancer survivors are sparse [17], particularly from minority and rural community members.

In this report, we describe the diagnoses and treatments of a cancer survivor cohort to better understand the characteristics of cancer survivors in the communities of Northern Florida. We matched them with community members without a history of cancer to compare the demographics, social determinants of health, and willingness to participate in research studies between these two groups. We also discuss the opportunities and the challenges that the cohort provides for future research.

METHODS

The HealthStreet Program, community engagement strategies, and Community Health Workers

Briefly, HealthStreet is a community engagement program launched in October 2011 (and prior, in 1989 in St. Louis, MO) with the Department of Epidemiology and is part of the Clinical Translational Science Institute (CTSI) at the University of Florida, with a mission to reduce disparities and improve the health of our community by bridging gaps in health care and health research [18, 19]. The backbone of the program is the Community Health Workers (CHWs), who are lay community members sharing a common language and culture with the people they serve [20]. CHWs assess needs and concerns, refer community members to low or no-cost medical and social services, and offer opportunities to learn about and participate in research relevant to them. Community members are engaged at various recruitment sites, including recreational parks, local libraries, laundromats, churches, bus stops, clinics, and public events. Participants provide written informed consent and complete a Health Need Assessment via interview. As of the end of 2019, HealthStreet CHWs had conducted over 11,000 assessments, navigated over 5,800 individuals to research studies, and provided over 23,900 medical and social referrals.

As part of the HealthStreet workforce training, all CHWs complete at least 20 hours of shadowing to learn the skills of rapport building and administering informed consent and the Health Needs Assessment. Additionally, the first three assessments done by a newly trained CHW are observed, and feedback is given by an experienced CHW with input from one of the faculty supervisors. Weekly supervisory meetings facilitate fidelity to all protocols and ensure continuous quality assurance. The protocol was approved by the University of Florida Institutional Review Board; all participants provide their written consent.

Study participants

All participants were HealthStreet members. As of March 2020, 1,080 out of 11,857 members reported a history of cancer and were part of the HealthStreet Cancer Survivor Cohort (HST-CSC). Nine-hundred and ninety-one cancer survivors with complete information were individually matched with 991 members without a history of cancer on sex (male or female), age at enrollment (<20, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79, ≥80 y), and residential zip code. We performed matching on age and sex for proper comparisons between the two groups because cancer survivors in HealthStreet were more likely to be older and to be women than members without a history of cancer. Also, because neighborhood-related factors have been suggested as a determinant of cancer survivorship [21], matching by residential zip code would reduce confounding by neighborhood factors [22].

Data collection

CHWs conducted face-to-face interviews (Health Needs Assessment) lasting approximately 20 minutes; interviews elicited information on social determinants of health, which includes aspects of social environment (e.g., education level, marital status, economic stability, and social support), physical environment (e.g., place of residence), and health services (e.g., insurance status, medical history, and health concerns) [23], substance use, and willingness to participate in different types of health research studies.

Participants who self-reported cancer history were asked the type of cancer(s) they were diagnosed with, year of diagnosis, and types of treatment received. Current weight and height were self-reported, and body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. We defined BMI <25.0 as normal/underweight, 25.0- 29.9 as overweight, ≥30.0 as obese [24]. Race was recorded as Asian, Black/African American, White, and Others, including American Indian/Alaskan native. Latino or Hispanic ethnicity was recorded separately. For the county of residence, a rural county was determined as that it has been classified as rural by Human Resources & Services Administration [25] or by the Florida Department of Health using the criterion of 100 persons or less per square mile [26], both based on Census 2010 data. We collapsed marital status into three categories: never married, married, and divorced/separated/widowed. Information on education was elicited by the last grade completed then stratified into less than 12 years versus 12 years or more (equivalent to high school degree or GED and beyond). Employment was elicited through the question, "Are you employed full-time or part-time?" (Yes or No). The employment status was classified as unemployed if the answer was no. Food insecurity was determined by the answer to the question, "Have there been times in the last 12 months when you did not have enough money to buy food that you or your family needed?". Current smokers were defined as individuals who ever smoked cigarettes in the last 30 days; former smokers were defined as individuals who ever smoked cigarettes but did not smoke in the last 30 days.

Data on perception and participation of research studies were also assessed from HealthStreet members; the methods are detailed elsewhere [17, 27]. For this publication, we included data from two questions: "Have you ever been in a health research study?" (Yes or no), and "How interested are you in being in a research study?" (Definitely, maybe, or not at all). To assess willingness to participate in health studies, members were asked whether they would volunteer for various study types that require different levels of participants’ involvement. We grouped the study types into two categories: “minimal risk studies” that survey participants' health, request consent to review medical records, and solicit a blood sample or “greater-than-minimal risk studies” that require participants to take study medications or use medical equipment, stay overnight in a hospital/clinic, and give a sample for genetic studies [28].

Statistical Analysis

The distributions of cancer site (by sex and age group), number of cancers, calendar year of cancer diagnosis and treatment, years since cancer diagnosis and treatment, and types of cancer treatment were described. Years since cancer diagnosis and treatment were calculated as the time from receiving diagnosis or treatment to the time of enrollment. Chi-square tests and two-sample t-tests were used to compare the characteristics and willingness to participate in research of cancer survivors with the matched non-cancer members. SAS version 9.4 (SAS Institute, Cary, NC) was used in conducting all statistical analyses.

RESULTS

In the sample of cancer survivors, 71.6% was female (mean age = 56.7 ± 14.1 y) and 28.4% was male (mean age = 60.8 ± 12.6 y); 19.2% lived in rural areas. Figure 1 shows the counties (n = 19) of residence of the cancer survivors. The leading sites of cancer diagnosis were skin cancer in men 18-64 years of age and women ≥65 years of age (Table 1). Other leading cancer sites were cervical cancer in women 18-39 years of age, breast cancer in women 40-64 years of age, and prostate cancer in men ≥65 years of age. Cervical cancer was less prevalent as age increased among women, but it remained the 3^rd and 4^th largest cancer diagnosis in women 40-64 and ≥65 years of age, respectively. Other cancers commonly diagnosed in U.S. men and women, including colon cancer, lung cancer, and lymphoma, were also seen in HealthStreet cancer survivors.

Figure 1:

County of residence of HealthStreet Cancer Survivor Cohort members in Northeast Florida

* Exact sample size for Duval County

* Exact sample size for Alachua County

The inset shows the location of the counties in Florida

(inset picture credit: https://commons.wikimedia.org/w/index.php?curid=749776 ).

Table 1.

Leading cancer diagnosis by sex and age group in the HealthStreet Cancer Survivor Cohort¹

Rank	18-39 y		40-64 y		65+ y
	Male	Female	Male	Female	Male	Female
	N=17	N= 92	N=148	N=399	N=111	N=214
	Site (%)2	Site (%)	Site (%)	Site (%)	Site (%)	Site (%)
1	Skin (25.0)	Cervical (52.2)	Skin (39.2)	Breast (27.6)	Prostate (46.8)	Skin (47.7)
2	Lymphoma and Neuroendocrine (12.5)	Skin and Ovarian (8.7)	Prostate (18.9)	Skin (25.1)	Skin (34.2)	Breast (32.7)
3	Spinal, Soft Tissue, Head/Neck, Blood, Pancreatic and Testicular (6.3 each)	Breast (6.5)	Colon (12.8)	Cervical (17.0)	Colon (9.0)	Colon (5.1)
4	NA	Thyroid (5.4)	Lymphoma (6.1)	Ovarian (6.8)	Lymphoma (2.7)	Cervical (4.7)
5	NA	Lymphoma (4.3)	Lung (4.7)	Uterine (6.3)	Lung, Kidney, Bladder, Breast, Leukemia, Oral (1.8 each)	Uterine (4.2)

NA: Not applicable

¹981 participants provided the response. N is an unduplicated number of participants in the sex and age group.

²Cancer diagnosis site and its percentage of all cancer sites in the group. Multiple cancers were counted in the percentage calculation if a participant reported more than one cancer type.

Among the cancer survivors, 88.9% had one cancer type, 8.3% had two cancers, and 2.8% had three cancers (Table 2). Approximately half (44.7%) of the cancer survivors received their cancer diagnosis fewer than 5 years prior to the baseline, while 17.2% received their cancer 5 to <10 years prior to baseline, 13.1%, 10 to <15 years, 19.2%, 15 to <20 years, and 15.8%, 20 or more years. The mainstay of cancer treatments, including surgery, radiation, and chemotherapy, were reported by 71.8%, 24.0%, and 23.1% of cancer survivors, respectively.

Table 2.

Characteristics of cancer diagnosis and treatment in the HealthStreet Cancer Survivor Cohort

Characteristics	%
Number of cancer type
One cancer type	88.9%
Two cancer types	8.3%
Three cancer types	2.8%
Year of cancer diagnosis 1
<2000	18.1%
2000-2009	26.1%
≥2010	55.8%
Years since cancer diagnosis 1
0 to <5	44.7%
5 to <10	17.2%
10 to <15	13.1%
15 to <20	19.2%
≥20	15.8%
Year of cancer Treatment 2
<2000	15.8%
2000-2009	22.2%
≥2010	62.0%
Years since cancer treatment 2
0 to <5	52.5%
5 to <10	14.9%
10 to <15	11.4%
15 to <20	7.0%
≥20	14.4%
Types of cancer treatment 3
Surgery	71.8%
Radiation therapy	24.0%
Chemotherapy	23.1%
Oral Medications	21.8%
Hormone Therapy	7.6%
Proton therapy	1.6%

¹Among 660 HealthStreet members who reported year of diagnosis

²Among 632 HealthStreet members who reported year of treatment

³Among 629 HealthStreet members who reported cancer treatment since March 2015. Multiple treatment responses are accepted.

Compared to the matched members without a cancer history on sex, age, and zip code (Table 3), cancer survivors were less likely to be Non-Hispanic Black (31.1% vs. 63.6%; p<0.0001 for testing all race/ethnicity groups), but were more likely to be divorced, separated, or widowed (49.5% and 41.2%; p<0.0001) and had higher educational attainment (85.6% vs. 79.1% with 12+ years of education; p=0.0002). Cancer survivors were also more likely to be normal or underweight (BMI <25 kg/m2, 34.0% vs. 25.6%; p=0.0002) and have health insurance (80.0% vs. 68.6%; p<0.0001). There was no statistically significant difference in smoking prevalence (27.5% vs. 31.8%, p=0.09), employment status (24.7% vs. 27.6%, p=0.15), or food insecurity (44.5% vs. 45.0%, p=0.84) between the cancer survivors and matched controls.

Table 3.

Demographics of cancer survivors and members without cancer history in HealthStreet. Numbers are percentages or mean (SD).

	Cancer survivors	Members without cancer history2	p-value
	(n=991)	(n=991)
Sex
Female	71.6%	71.6%	-
Male	28.4%	28.4%
Age (y), mean (SD)
Female	56.7 (14.1)	56.3 (14.0)	-
Male	60.8 (12.6)	59.9 (12.8)	-
Living in rural counties 1	19.2%	19.2%	-
Race/Ethnicity
Asian	0.6%	1.2%	<.0001
Black/African-American	31.1%	63.6%
White	62.9%	33.3%
Other	5.4%	1.8%
Latino/Hispanic	4.4%	2.2%	0.0077
Marital Status
Never Married	20.6%	26.8%	<.0001
Married	29.9%	32.0%
Divorced/ Separated/Widowed	49.5%	41.2%
Body Mass Index (kg/m2)
<25	34.0%	25.6%	0.0002
25- <30	29.9%	32.6%
30+	36.1%	41.8%
Smoking
Current smokers	27.5%	31.8%	0.0913
Former smokers	32.2%	31.3%
Never smoker	40.4%	36.9%
Education attainment (≥12 years)	85.6%	79.1%	0.0002
Currently Employed	24.7%	27.6%	0.1450
No health insurance	20.0%	31.4%	<.0001
Food Insecure	44.5%	45.0%	0.8451

¹Rural counties were: Putnam, Taylor, Suwannee, Lafayette, Dixie, Levy, Gilchrist, Columbia, Union, and Bradford Counties [25, 26].

²1:1 matched with cancer survivors by sex, age, and zip code

Among cancer survivors, 32.4% reported ever participating in a health research study, and nearly all survivors (95.3%) were definitely or might be interested in participating in research (Table 4). These proportions were significantly higher than those reported by matched members with no cancer history (24.9% and 83.7%, respectively). Over 90% of the cancer survivors would volunteer for health research studies that are minimal risk, and their willingness was higher for studies reviewing medical records (90.3 vs. 82.4%, p<0.0001) and obtaining a blood sample (92.1% vs. 80.2%, p<0.0001), compared to matched non-cancer members. For studies that had greater-than-minimal risk, between 67% and 91% of the cancer survivors expressed their interest to participate, and all proportions were higher among the cancer survivors than among the matched non-cancer (all p<0.001).

Table 4.

Research perceptions of cancer survivors and members without cancer history in HealthStreet

Questions	Cancer survivors	Members without cancer history	P-value
	(n=991)	(n=991)
Ever been in a health research study?, Yes	32.4%	24.9%	0.0012
Interested in participating in research?, Definitely/Maybe	95.3%	83.7%	<.0001
Would you volunteer for a health research study…
Study with minimal risk
that only asked questions about your health?, Yes	95.4%	93.5%	0.0792
if researchers wanted to see your medical records?, Yes	90.3%	82.4%	<.0001
if you had to give a blood sample?, Yes	92.1%	80.2%	<.0001
Study with greater than minimal risk
if you might have to take medicine?, Yes	67.6%	59.6%	0.0002
if you were asked to stay overnight in a hospital or clinic?, Yes	79.3%	65.8%	<.0001
if you might have to use medical equipment?, Yes	91.7%	77.1%	<.0001
if you were asked to give a sample for genetic studies?, Yes	91.8%	76.6%	<.0001

DISCUSSION

Participants in HST-CSC had diverse social determinants of health, which presented unique opportunities for cancer survivorship research in underrepresented and underserved populations. As shown in our analysis, a substantial percentage of cancer survivors in HST-CSC were Black (31.1%), did not have health insurance (20.0%), and experienced food insecurity (44.5%). The HealthStreet sample overall has a higher proportion of Black members than the communities [17], reflected by the high proportion (63.6%) in our matched members without a cancer history. The geographic area covered by this cancer survivor cohort would have been underrepresented by other cohorts drawn from large cancer centers, which are mainly in urban areas. While approximately one in five cancer survivors in HealthStreet lived in rural counties, the majority of other counties home to cancer survivors were small metropolitans [29]. Also, several of the counties were among the poorest counties in Florida in 2019: Putnam (percent of the population below 200% of poverty level: 50.9%), Levy (48.0%), Dixie (46.6%), Lafayette (46.3%), and Gilchrist (43.9%) [30]. The proportion of rural residents in our cancer survivors (19.2%) is similar to the national estimate for cancer survivors (20.8%), which translates to a population of 2.8 million rural cancer survivors in the U.S. [31]. Disparities in cancer survivorship issues between rural and urban cancer survivors have been observed [32, 33]. The HST-CSC can be an important resource for recruiting cancer survivors, assessing cancer burden, and facilitating cancer survivorship research in rural and non-large metropolitan populations.

Compared with national data, cancer survivors from the HealthStreet cohort were younger; the percentage of age ≥65 years was 40.2% in males and 30.8% in females (data not shown), and the corresponding percentages were 68% and 60% in the U.S. [4]. The cancer survivors in HealthStreet were enrolled at the time relatively close to diagnosis: 44.7% had diagnosis ≤5 years, and 15.8% had diagnosis ≥20 years. The national data showed 33% of cancer survivors in communities had diagnoses ≤5 years, and 18% had diagnoses ≥20 years [4]. However, smoking was more prevalent among HST-CSC, as 27.5% were current smokers, compared to 14% in male and 18% in female cancer survivors included in the National Health Interview Survey (NHIS) 2008-2012 [34]. In addition, the prevalence of obesity was 36.1% in HST-CSC, compared to 28% in male and 31% in female cancer survivors from the 2008-2012 NHIS data. In addition, the percentage of no health insurance (20.0%) in HST-CSC was much higher compared to national data (5%) [35]. Lack of health insurance can be a major obstacle for access to cancer surveillance and treatment and even general health care for cancer survivors. These data indicate that smoking cessation, promoting energy balance to reduce obesity, and improving access to cancer care are among the priorities of reducing cancer burden in the target population and geographic area.

An important strength of our study is that unlike conventional approaches often restricting enrollment to cancer patients at specific clinics, the HealthStreet Registry recruits both healthy and diseased community residents regardless of history of diseases and where they received treatment. As we have demonstrated, a non-cancer comparison group can be readily selected from HealthStreet members who are from the same communities as age and sex matched cancer survivors to allow for conducting association studies. Another strength is that the CHW model ensures community engagement and trust because the CHWs were recruited from the same community as participants. They are culturally sensitive and engendering trust in intake interviews with community members. This is reflected by the very high overall consent rate (84.8%, as of November 8, 2020) of people recruited by HealthStreet. CHWs also assist in participant navigation to research. As of September 2020, a total of 119 cancer survivors have been navigated, and 39 (33%) have been enrolled in cancer-related studies. For community members without a history of cancer, 352 have been navigated, and 89 (25%) have been enrolled in cancer-related studies. The empirical data are consistent with our observation that a higher proportion of cancer survivors were more willing to participate in a research study than members without a history of cancer. These findings suggest that through proper navigation, such as the CHW model, and eliminating barriers, such as researchers’ awareness and transportation, underserved populations can be more represented in research studies.

In our assessment and experience, innovative research addressing gaps in cancer survivorship can be conducted in HST-CSC. Published studies analyzing the HealthStreet intake data included substance use (hazardous drinking and cocaine) [36, 37], prescription opioid use [38], aging-related diseases (Alzheimer's disease) [39], healthcare utilization [40], technology (text message) to increase service use referrals [41], and geospatial analysis [42]. Researchers will have the same opportunities to investigate these topics in HST-CSC. As the cohort is open and continues to enroll members, many more emerging topics, e.g., improving access to surveillance and continuing care after cancer [43, 44], preventing long-term or late effects of cancer treatment, and maintaining energy balance and healthy lifestyle [45], warrant investigations in HST-CSC.

We acknowledge several limitations of the design of the cancer survivor cohort. First, the cohort was agnostic to cancer types, a limitation resulting in a small number of survivors in cancer with high mortality or rare cancer. As the cohort continues to enroll cancer survivors and our CHWs can make their presence more visible in specific cancer care clinics, conducting research on specific types of cancer will also become feasible. Second, the intake data relied on self-report, and there are potential recall errors in cancer diagnosis and treatment. We were unable to differentiate important cancer subtypes, such as melanoma vs. non-melanoma skin cancers. More recently, we have started to consent HealthStreet members for reviewing and linking health records. From January 1, 2019, to November 8, 2020, 92.1% (841 out of 913) of newly enrolled HealthStreet members provided their consent of medical record access. The linkage of their medical records and validating the self-report intake data is ongoing and will provide needed information on the validity of the self-report.

In conclusion, HST-CSC is a community-based, research-participatory cohort providing valuable research resources in cancer survivorship. The cohort presents opportunities for research in diverse, non-metropolitan, and underserved cancer survivors. High-quality, engendering enrollment of the CHW model and effective retention approaches have contributed to the high level of trust and engagement to research among the members. As we continue to develop the cancer survivor cohort, it will be a unique resource for a broad range of survivorship research and bridging gaps between research and cancer survivors living in communities.