Abstract
Introduction The number of individuals called upon to provide support for cancer patients within their personal networks is steadily increasing. Prior studies show that caregiver screening rates either decrease due to caregiving demands and associated fatalism or increase due to risk perception and healthcare involvement. However, there remains a gap in research in understanding how cancer screening relates to extent of caregiving and relationship to cancer patient, particularly non-family/spouse. This study aims to assess the impact of degrees of relationship to cancer patient(s) and extent of caregiving on adherence to breast, cervical, prostate, and colorectal cancer screening guidelines.
Methods Participants of Bronx, New York were recruited online or through community events to complete a set of core items adapted from the NCI Health Information National Trends Survey and other sources. Logistic regression analyses identified factors associated with variation in screening.
Results Analyses were based on 1430 participants (73% female, mean age 50 years, 43% Hispanic). An unexpectedly large proportion had cancer within their families (72%) and/or provided some support to a cancer patient (79%). Four support patterns were found-none, emotional support only, less intensive, and more intensive support. Women who provided emotional support only were less likely to be screened for breast and cervical cancer. Among more active caregivers, cervical and prostate cancer screening were greater among those who provided more intensive support.
Conclusions Having family or friends with cancer is a normative experience that affects one’s own preventive care. Those with a family member or spouse with cancer have increased personal screening. As involvement in cancer caregiving deepens, one’s own screening rises. Caregivers that provided only emotional support are the exception, potentially because they have greater distance from the cancer experience. Findings suggest the potential value of tailoring interventions to address personal experiences with cancer caregiving.
Introduction
The incidence of cancer continues to rise and with it, the number of individuals in the community called upon to provide support and care for cancer patients. Cancer remains the second leading cause of death in the United States, and 15 million people are projected to be diagnosed by the year 2020 [1,2]. Due to advances in treatment, diagnosed patients now have the chance to live longer, though not without a lengthy cancer experience. As such, exposure to cancer in one’s personal networks and provision of support for family and friends with cancer have become normative experiences [3]. Indeed, over the adult lifespan, individuals may provide support and care for multiple family members and friends affected by cancer. Cancer caregivers assume significant supportive roles, ranging from emotional support, meal provision, financial assistance, housing accommodation, household management, treatment coordination, and accompaniment during hospital visits [4]. The extent of cancer caregiving often depends on the complexities surrounding the care of a particular patient. The unique nature of the diagnosis, including cancer type, stage, and treatment, in part determines the duration and intensity of caregiving needs [5]. Extent of caregiving is also influenced by patients’ needs and capacities, particularly for spouses, children and others who remain with the patient throughout their cancer journey. When patients require daily assistance, caregiving can become a full-time job of coordinating care and meeting the escalating demands of decision-making and problem-solving. In low-income communities, social determinants of health can exacerbate stress associated with cancer caregiving [6,7]. Even caregivers less directly involved in personal or practical aspects of care can face challenges helping patients cope with emotional needs. Given this broad range of caregiving experiences, it is worthwhile to consider how the varying challenges and demands faced by cancer caregivers affects their own cancer screening behavior.
Caregiver stress and burden
It has been well-documented that caregivers’ own self-care and health seeking behavior may be affected by the level of physical and psychological burden associated with the role [8,9]. Studies have paralleled caregivers’ slow deterioration of physical health with the accumulation of chronic stress, which in turn is associated with inflammatory processes and increased relative risk of cardiovascular disease [10,11]. Beyond the amount of time devoted to caregiving or the severity of patient symptoms, caregivers’ experience of burden is also associated with feelings of lower self-efficacy and inadequacy to perform what is expected [12,13]. The loss of predictability that is often tied to cancer can overwhelm caregivers’ perception of control and lead to anxiety and depression. Caregivers may also become socially isolated, worsening mental health symptoms as well as sleep disturbances and fatigue [14,15]. The stress of caregiving can exacerbate health-risk behaviors and pose as a perceived barrier in the ability to preserve one’s health [7,8]. Although adopting healthy behaviors could safeguard against these detrimental health effects, caregivers often have little time or energy to focus on their own well-being.
Cancer fatalism
An examination of the 2007 National Cancer Institute’s Health Information National Trends Survey (HINTS) data found that those with personal or familial exposure to cancer are more likely to worry about getting cancer, to agree that they will develop cancer in the future, and to disagree that cancer is most often caused by individuals’ behavior or lifestyle [16]. A driver of these attitudes towards cancer are fatalistic beliefs that events are inevitable, and human behavior cannot influence health outcomes. This may be further reinforced by the belief that genetic factors determine cancer risk, not lifestyle factors [17]. A recent study found that women held a strong misconception that heredity was the most important factor in developing breast and colorectal cancer [18]. Lower educational levels and lower income are also correlated with lower levels of health literacy, which in turn is associated with fatalistic cancer beliefs [19,20]. Helplessness associated with this sort of fatalism leads to reduced motivation to maintain health behavior and to adhere to preventive screening [21,22].
Caregiver risk perception and rates of screening
Along with the stresses associated with caregiving, studies have also shown that caregivers can become more cognizant of their own risks. This awareness of risk may either exacerbate or offset distress associated with demands of caregiving. For example, first-degree familial relations of cancer patients may worry that they are also at risk for cancer due to family history. Even without explicit familial risk factors or genetic syndromes, supporting a family member with cancer can lead caregivers to become increasingly conscious of their own risk [23]. This heightened sense of risk may be adaptive, leading caregivers to seek cancer information [24]. Prior studies have found that greater family history predicts breast and colorectal cancer screening [25,26]. They suggest that individuals who perceived that they were at greater risk of developing cancer themselves are more aware of recommended guidelines and more motivated to be screened. This also pertains to the greater use of preventive services, as well as efforts to achieve healthy lifestyles with diet, exercise, smoking cessation, and decreased alcohol consumption [9,27,28,29].
Situational influences on caregivers’ health behavior
For spousal and familial relations of cancer patients, the shift in health behavior may result from shared social context. Recent studies suggest that as the cancer patient changes his or her health behaviors, the partner is also likely to do so [27]. As the main confidante in a family member’s journey to survivorship, caregivers themselves are presented with more opportunities to reflect upon the consequences of not pursuing screening. They also spend more time around providers, with ready access to preventive services [30]. However, this may be counterbalanced by demands on caregivers’ time, finances, and motivation to take on another health issue. Cancer caregivers may be more inclined than others to consider cancer screening and early detection in terms of financial implications or cost-benefits [18]. Those exposed to cancer patients may observe firsthand the significant economic and physical burden of the disease and desire to prevent the same hardship for themselves and their families. Rather than becoming fatalistic, it is ideal if these caregivers can become proactive regarding their own health.
Study goals and hypotheses
Earlier studies support two competing hypotheses regarding caregivers’ adherence to recommended cancer screenings: caregivers either have lower than expected rates of screening due to the demands of caregiving, perceived lack of control and fatalistic beliefs, or caregivers increase their rates of screening due to proximity to the cancer patient, direct awareness of the consequences of cancer, and involvement with the healthcare system.
However, few studies have considered caregivers’ cancer screening in light of the interplay between caregivers’ extent of caregiving and their relationship to patient. Research considering the full range of caregiving roles, including involvement of non-family/non-spousal caregivers, is particularly limited. Prior studies have grouped them with relatives and partners as informal caregivers (persons who are unpaid and without formal health care education) but have not studied how they may differ in their health behavior [31]. An investigation of these relationships can produce a more comprehensive understanding of how to address prevention and planning of care for cancer caregivers [9].
This observational study draws upon a purposive sample of the population of the Bronx, New York. The Bronx is the poorest county in New York State, designated as a persistent poverty community by the USDA and a health services shortage area by HRSA. Residents are predominately Latinx and non-Hispanic Black. Due to low socioeconomic status and associated social determinants of health, Bronx residents are already vulnerable to factors that discourage self-care. In these analyses, we examine how caregivers’ relationship to cancer patient(s) and extent of caregiving is associated with challenges with adherence to age- and sex-specific guidelines for breast, cervical, prostate, and colorectal cancer screening. The study also examines whether cancer attitudes, perceived risk of developing cancer, and number of barriers to care were associated with the caregiving relationship and extent of caregiving.
We hypothesize that respondents’ screening will be positively associated with their closeness to cancer patients (e.g., spouse vs family vs friend vs none) and with the intensity of support provided to cancer patients (e.g., more intensive support, less intensive support, emotional support only, no support). However, we will also examine the competing hypothesis that greater caregiving demands may be associated with fatalistic attitudes and beliefs about cancer and negatively impact screening behavior. Our analyses will examine whether and how these factors compete and combine to influence screening.
Materials and Methods
Participant selection
This survey was performed as part of a larger observational study by the Montefiore Einstein Cancer Center (MECC) to assess the cancer attitudes, health risks and health behavior evident in the diverse population of Bronx, New York. It was part of a larger NCI effort to collect local data from designated cancer centers’ catchment areas to compile and compare with the national HINTS survey.
The Bronx is the catchment area served by MECC, characterized by very low socioeconomic status, marked ethnic and racial diversity, and poor health outcomes. The collection of data was initiated in May 2017 and completed in February 2018. Inclusion criteria included adults 18 to 74 years of age, who speak English or Spanish, and who reside in the Bronx at the time of survey. Survey participants were either recruited online, through email addresses provided to our health system or in-person, at community events. Email addresses were obtained through the hospital Electronic Health Record (EHR). To increase likelihood of connecting with the intended participants, we limited the email list to individuals who had a provider visit at least once in the prior three years. Prior to initiating email waves, we stratified the list according to Bronx community district of residence, age, and sex. We monitored responses in real time and adjusted subsequent waves of email to ensure our sample included adequate representation of males and younger respondents. We sent approximately 130,000 email invitations to complete the survey. Of this number, 6497 (5%) opened the email and clicked on the REDCap survey link or contacted our toll-free number to schedule a telephone interview. An additional 44 individuals were recruited at events held at community venues (e.g., churches, schools, medical centers, homeless shelters) or from locations within the community with heavy foot traffic (e.g., shopping centers). After deduplication and removal of mostly empty records, 1883 (29%) provided usable data. These individuals could complete the survey using REDCap, by telephone or in person with paper and pencil. A total of 1430 respondents (76%) provided complete data on family caregiving necessary for the present study.
Survey development
All participants completed a set of core items determined by a group of investigators conducting this survey at 15 different cancer centers participating in the NCI’s Population Health Assessment Initiative [32]. Core survey domains included demographics, health information seeking, access and barriers to health care, cancer screening knowledge and behavior, tobacco usage, cancer beliefs, awareness of risk, and barriers to care. Subsets of study participants were also randomly assigned to complete one of two supplemental modules, either to describe a recent health care episode or to assess social determinants of health, chronic conditions, and quality of life. The variables included in these analyses were drawn from the core set of items completed by all patients.
Measures
Independent variable: extent of cancer caregiving
Our primary independent variable is the extent of cancer caregiving. To assess caregiving, we asked respondents to indicate the types of support they had ever provided to family or friends with cancer on a 12-item checklist-personal care, household chores, accompaniment to appointments and other errands, help making decisions about their care, help with money, rent, or other expenses, provision of a place to stay, and emotional support. Participants were able to make as many selections as applicable or indicate that they had not provided any support.
Dependent variables: adherence to cancer screening
Our main dependent variables addressed adherence to age and gender appropriate cancer screening recommendations. All items of the cancer screening modules were adapted from HINTS. Binary variables indicating adherence were derived for each screening modality, as described below. Note that individuals who responded that they ‘don’t know or don’t remember being screened’ were coded as being unscreened.
Survey questions for colorectal cancer screening were only presented to men and women over the age of 50, in accordance with the US Preventive Services’ guidelines in place at the time of the survey. Participants were asked if they have ever had a blood stool test, colonoscopy, or sigmoidoscopy, and how long it had been since their last home kit test or exam. We chose to analyze whether the individual had ever had one of the approved screens.
The survey module for breast cancer screening was limited to only female participants ages 40 and above. This section assessed whether the respondent had ever had a mammogram, and if so, the time since their last mammogram. For this analysis, we examined factors associated with mammography completion in the past two years.
The module for cervical cancer screening was identical to the breast cancer screening module but administered to all women participating in the survey. Adherence to cervical cancer screening guidelines was assessed in terms of completion of a pap test within the past two years.
Inclusion criteria for prostate cancer screening was limited to men aged 40 and over. Given the USPSTF’s Grade C recommendation regarding PSA testing among men ages 55 to 69, participants were asked if a health care provider ever talked about the pros and cons of having a PSA test to look for possible signs of prostate cancer, as well as whether they had ever received the test. These two variables were nearly identical, so our analysis included whether men had ever had a PSA test.
Moderator: relationship to cancer patient(s)
Respondents’ degree of exposure to friends and family with cancer was treated as both a covariate and a potential moderator of the impact of caregiving experiences on screening. Two binary items examined the participants’ exposure to cancer: (1) Have any of your blood relatives ever been told by a doctor that they had cancer? This includes parents, grandparents, brothers, sisters, children, or grandchildren, and 2) Has your spouse or partner ever been told by a doctor that she or he had cancer? In addition, we also controlled for whether respondents themselves had ever been diagnosed with cancer.
Covariates: cancer attitudes, beliefs, and awareness
Six items from HINTS were used to evaluate cancer attitudes and beliefs: (1) It seems like everything causes cancer, (2) There’s not much you can do to lower your chances of getting cancer, (3) There are so many different recommendations about preventing cancer, it is hard to know which ones to follow, (4) When I think about cancer, I automatically think about death, (5) Cancer is most often caused by a person’s behavior or lifestyle, and (6) I’d rather not know my chances of getting cancer. All items were assessed using a 4-point scale with selections ranging from ‘strongly agree’ to ‘strongly disagree.’
Covariates: chances of getting cancer
One item from HINTS was included to examine how relationship to cancer patient and extent of caregiving roles affected perception of personal risk: ‘Compared to other people your age, how likely are you to get cancer in your lifetime?’ Respondents indicated their likelihood of getting cancer on a 5-point scale, with selections ranging from ‘much less likely’ to ‘about the same’ to ‘much more likely.’ In order to fully explore risk perception, we derived three variables from this item. First was a linear measure of perceived risk. To avoid missing data on this measure, people who responded that they did not know their perceived risk were assigned the midpoint score of 3, “about the same risk as others my age.” Second, we included a binary variable to distinguish people who indicated that they did not know their risk compared to other respondents. In addition to providing a statistical correction for imputation of the missing value, this variable also allowed us to consider whether unwillingness or inability to judge one’s own cancer risk was a meaningful predictor of behavior. Third, we included a quadratic (squared) risk score to address the possibility that extreme high and extreme low perceived risk could each discourage screening.
Covariates: barriers to care
Participants chose all that applied from the following measures for why they could not see a doctor when they needed to within the past 12 months: (1) Cost, (2) Lack of time, (3) Transportation, (4) Caring for children, (5) Caring for other family members, (6) Work or school responsibilities, (7) Family disagreements about what to do, (8) Emotional upset or stress, (9) Illness or fatigue, (10) Difficulties related to housing, (11) Legal difficulties, or (12) Other. Given the relatively small number of people reporting five or more barriers, we rescaled this count variable to indicate the number of barriers as either zero, one, two, three, or four or greater.
Data analysis
Describing relationship to cancer patient(s) and patterns of cancer support
All statistical analyses were implemented using SPSS version 27.0. Descriptive statistics were performed to characterize the study sample, including relationship to people with cancer and extent of support provided to cancer patients. We divided the sample into three groups to describe presence of cancer in the family: those with no family indicated, with family but no spouse, and spouse with/without family. We derived categorical variables to describe different patterns of caregiving support provided by the respondent, which are defined below. Note that this included support for cancer patients outside the family.
Evaluating the effects of provision of cancer support on caregivers’ adherence to cancer screening
We conducted a series of logistic regression analyses to examine the effects of caregiving for cancer patients on respondents’ cancer screening. Dependent variables for these analyses included screening measures described above. The sample size differed for each logistic regression analysis and was based on respondents with complete data eligible for a given screening modality by virtue of age and sex. We conducted separate analyses of colorectal screening within sex, to allow comparison with the other screening modalities that are sex specific. To isolate the independent contributions of family involvement on screening, a variety of covariates were entered into the regression model. Covariates were selected a priori because of their anticipated relationships with cancer screening adherence. Covariates were permitted to enter regression models at p<0.10 using forward selection and were considered in the following hierarchical order:
Demographic measures: Age, Race (Black, White, Other), Ethnicity (Hispanic Origin)
Personal and Familial Exposure to Cancer: Personal (1,0), Family Member (1,0), Spouse/Partner (1,0)
Number of Barriers (from 0 to 12)
Perceived Personal Risk of Cancer: Rating of Risk (1 to 4), Squared Risk, Don’t Know Risk (1, 0)
Cancer Attitudes: Six items, summarized into three orthogonal dimensions using principal components analysis: Pessimistic/Fatalistic about Cancer Outcomes; Preference to Not Learn One’s Own Risk of Cancer; Belief that Cancer is Mostly Caused by One’s Behavior or Lifestyle (standardized scores)
On each step, we retained only covariates that contributed significantly to the model. This hierarchical order of entry prioritized variables referring to personal and family history of cancer before considering perceptions and attitudes. Instances of multicollinearity among covariates are noted below in our description of regression results.
The primary independent variable, extent of cancer caregiving, was entered next after all significant covariates. This variable was represented by using orthogonal contrast codes to distinguish the four levels of support provided to cancer patients that were observed in the data. In the final step of the regression, we entered interaction effects to determine whether and how the impact of caregiving on respondents’ own cancer screening differed, depending upon whether respondents’ closest relationship with a cancer patient was with a spouse/partner, family member, friend, or no caregiving relationship. As the primary independent variable, all main effects and interactions involving patterns of cancer caregiving were tested at p<0.05 level of significance.
Results
Sample demographic characteristics
A total of 1430 participants with complete data met our study entry criteria. The participants averaged 50 years (±15 years) and were 73% female. The survey participants were 32% non-Hispanic Black, 21% non-Hispanic White, 43% Hispanic any race. Both the More Intensive and Less Intensive support groups (52 years) were older than Emotional Only (50 years) and No Support (46 years) (p<0.001). Both Blacks and Whites were more likely to have family with cancer than to have no family indicated (54.5% vs. 38.6%; p<0.001; 62.9% vs. 28.1%; p<0.001). Hispanics were also more likely to have family with cancer than have no family indicated (58.9% vs. 32.4%; p<0.001). Whites were more likely to provide More Intensive support compared to Black and Other race (41.6% vs. 34.3%, 31.6%; p=0.05). Blacks provided More Intensive support than any other level of support (34.3% vs. 29.1%; p=0.05).
Provision of cancer support
Patterns of provision of support were determined using descriptive statistics and cluster analysis. Two groups were immediately apparent: 402 (28.1%) respondents indicated that they had not provided support to a cancer patient, and 251 respondents (17.6%) provided emotional support only. On the 12-item checklist, the remaining 777 participants indicated that they had provided several types of support. K-means cluster analysis was used to determine how to best characterize this subsample. We divided these caregivers into two groups-260 (18.2%) who provided Less Intensive support and 517 (36.2%) who provided More Intensive support.
Relationship to cancer patient(s)
When stratified by the relationship to cancer patient(s), 527 (36.9%) indicated no family member with cancer, 792 (55.4%) with family, and 111 (7.8%) with spouse with cancer. Of those with spouse with cancer, 86 (78%) reported that they also had additional family members with cancer.
Personal history of cancer
Participants with a past diagnosis of cancer were more likely to also have family with cancer (60.6% vs. 55.0%; p=0.03) and to provide some form of support, whether it was Emotional Support Only (18.2% vs. 17.3%; p=0.003), Less Intensive support (26.3% vs. 17.3%; p=0.003), or More Intensive support (39.4% vs. 36.0%; p=0.003).
Barriers to care
The most prevalent barriers to care were family disagreement about what to do (327) and illness or fatigue (336). The mean number of barriers to care was highest for those without family with cancer (1.04; p=0.008), followed by those with family with cancer (0.93; p=0.008) and those with spouse and family with cancer (0.71; p=0.008).
Main results – associations of caregiving to caregivers’ own cancer screening
Women breast cancer screening in the past two years
Among 653 women over age 40, 541 women (82.8%) were adherent while 112 (17.2%) reported not having had a mammogram in the past two years. In unadjusted analysis, adherence to breast screening was unrelated to women’s caregiving roles (χ2 = 4.19, ns). However, planned comparisons showed that individuals who provided Emotional Support Only were less likely to be screened than those who provided additional forms of support (OR = 0.60; 95% CI = 0.36, 0.99; p=0.04). Significant covariates in logistic regression related to screening in the past two years included older age (adj OR = 1.06; 90% CI = 1.04, 1.08) and fewer reported barriers to care (adj OR = 0.83; 90% CI = 0.71, 0.97). After taking these covariates into account, we found that adherence to mammography was no longer associated with cancer caregiving differences. Examination of interaction effects demonstrated that the association of caregiving role with mammography depended on respondents’ relationships with cancer patients (Fig 1). Among the group that indicated having no family member or spouse with cancer (blue line), rates of mammography were slightly higher among those who had provided More Intensive support to non-family member. In the group that had provided support to family members (orange line), mammography screening was markedly lower among those who provided Emotional Support Only. In contrast, among those who had provided care for spouses, screening was notably lower among those who provided More Intensive levels of support (grey line).
Mammography in the past two years – relationship with cancer patient by level of support.
Women cervical cancer screening in the past two years
Among 1005 women, 737 (73%) were adherent while 268 (27%) report not having had cervical cancer screening in the past two years. In unadjusted analysis, adherence to pap testing was unrelated to women’s caregiving roles (χ2 = 2.53, ns). Significant covariates in logistic regression related to screening in the past two years included younger age (adj OR= 0.96; 90% CI = 0.95, 0.97), being Hispanic (adj OR= 1.57; 90% CI = 1.17, 2.12) and fewer barriers to care (adj OR= 0.82; 90% CI = 0.74, 0.91). Women who described their race as either Black or White (versus Other) reported higher rates of adherence (adj OR=1.51; 90% CI = 1.08, 2.12). Additionally, women who had pap screening tended to believe that they were at lower risk for cancer (adj OR= 0.84; 90% CI = 0.73, 0.98). After taking all these covariates into account, we found a trend that women who reported providing Emotional Support Only to a cancer patient were less likely to be screened, compared to those who had provided More Intensive support (adj OR= 0.71; 90% CI = 0.51, 0.98, p=0.08).
Women colorectal cancer screening – lifetime
Among 596 women over age 50, 488 (82%) were adherent while 108 (18%) report never having had colorectal screening. In unadjusted analysis, adherence to colorectal screening was unrelated to women’s caregiving roles (χ2 = 4.34, ns). In terms of covariates in logistic regression, older age was positively associated with colorectal screening (adj OR = 1.03; 90% CI = 1.08, 1.12). Alternatively, women who identified as Hispanic (adj OR = 0.44; 90% CI = 0.26, 0.73) or who reported more barriers to care (adj OR = 0.73; 90% CI = 0.57, 0.93) were less likely to be screened. Among those women who identified as Other race, non-Hispanics were older than and significantly more adherent to screening than Hispanics (63 years vs. 57 years, 100% vs. 66%). After taking these variables into account, we found that colorectal screening was related to active cancer caregiving. Women who reported carrying out More Intensive involvement as a cancer caregiver were more likely to be screened than those who performed fewer supportive functions (adj OR = 0.46; 95% CI = 0.21, 1.03).
Men colorectal cancer screening – lifetime
Among 314 men over age 50, 278 (89%) were adherent while 36 (11%) reported never having had colorectal screening through either endoscopy or stool sample. In unadjusted analysis, adherence to colorectal screening was unrelated to men’s role as cancer caregivers (χ2 = 4.91, ns). Significant covariates predicting men’s adherence to CRC screening included older age (adj OR = 1.16; 90% CI = 1.08, 1.24) and fewer barriers to care (adj OR = 0.70; 90% CI = 0.50, 0.98). Men with a prior history of any cancer in our sample were less likely to report colorectal cancer screening (adj OR = 0.10; 90% CI = 0.25, 0.61). Greater pessimism about cancer prevention and treatment was also associated with reduced likelihood of screening (adj OR = 0.38; 90% CI = 0.60, 0.96), while the belief that cancer is primarily caused by personal behavior or lifestyle was associated with greater likelihood (adj OR = 1.64; 90% CI = 1.07, 2.52). After adjusting for these covariates, men’s role as cancer caregivers remained unrelated to adherence to colorectal screening.
Men prostate cancer PSA discussion and testing– lifetime
Among 342 men over age 40, 158 (46%) were had discussed PSA with a health care provider while 184 (54%) had not. Of the men who had discussed PSA with a provider, almost all (96%) had received the test. Unadjusted analysis revealed that men who had provided the More Intensive support as a cancer caregiver were significantly more likely than other men to have had a PSA screen (χ2 = 8.23, p=0.016). Significant covariates in our logistic regression model to predict PSA screening included older age (adj OR = 1.10; 90% CI = 1.07, 1.13) and identifying as either Black or White (versus Other) (adj OR = 2.01; 90% CI = 1.18, 3.42). There was also a U-shaped effect involving perceived risk, with PSA testing being more likely among men who perceived that they were at either a higher or lower risk of getting cancer compared to those with either average or unknown risk (adj OR = 1.26; 90% CI = 1.06, 1.50). After adjusting for these covariates, provision of More versus Less Intensive support was associated with a trend of greater likelihood of PSA testing (adj OR =1.97; 90% CI =1.05, 3.70; p=0.07). Further analysis suggested that the influence of experience as a family caregiver depended upon the respondents’ relationship to the cancer patient. After removing this main effect from the model, the interaction between intensity of support provided to family versus non-family was significant (adj OR = 9.44; 95% CI =1.39, 64.12). The plot of this interaction effect demonstrates that PSA testing was greater for individuals who had a family member or spouse with cancer (orange line) compared to those who did not (blue line), and that this difference increased with the intensity of support provided (Fig 2).
Ever had PSA test– family versus non-family relationship with cancer patient by level of support.
Discussion
A large proportion of the sample had cancer within their families (72%) and/or provided support to cancer patients (79%). This is noteworthy as it signifies the broad impact of cancer and the heavy burden of cancer caregiving within this community. Whether providing only emotional support or a greater intensity of support, most people were involved with people in their lives affected by cancer. In sum, cancer caregiving must be viewed as a normative experience, at least in this low-income inner-city community. As such, lifetime personal experiences with cancer have been largely overlooked as a key factor in efforts to improve adherence and support health decisions. Understanding the impact of caregiving on respondents’ personal screening behavior may elucidate strategies to increase screening for all groups.
Demographics
Our community sample was more likely to be older and female compared to the overall Bronx population. Older age is a strong correlate of screening adherence as well as the likelihood of knowing and caring for network members with cancer. Caregivers’ age was controlled in all analyses. Although caregiving is often understood to be a role more commonly served by women, most of both women and men in our sample indicated providing at least some support for cancer patients. In a descriptive analysis, we found that older adults (≥50 years) provided more support than younger adults. In particular, older women provided more personal care and help with household chores than older men. Given these differences, we conducted separate analyses of factors influencing colorectal screening for women and men, as we did for the other sex-specific screening modalities.
Choices in cancer screening measurement
Measurement of each cancer screening modality was carefully specified. Due to lack of information regarding family history of cancer, adherence measures were based on recommendations for individuals at average risk. Completion of mammography within the past two years reflected current USPSTF breast cancer screening guidelines, while allowing for flexibility in physician recommendations by retaining the earliest-recommended ages of 40 to 50. We lacked information to account for variability in Pap testing taking women’s HPV status, age, and prior testing into account, so we used an intermediate measure adherence, Pap smear within the past two years. Colorectal cancer screening for both men and women included ever having had an endoscopy or stool test to encompass the varied approaches to screening per ACS. Survey limitations in determining timing of tests led us to use ever having had colorectal screening as our variable. Male prostate cancer screening included both ever having discussed PSA testing with a provider and ever having had PSA testing. This method intended to account for PSA discussion without a decision to have the test as acceptable adherence to guidelines. However, only 4 men in the sample indicated that they had discussed PSA with their provider without testing, so outcomes reported here focus on receipt of the PSA test.
Screening findings
In this diverse, inner-city sample, people who provided intensive support to cancer patients were most likely to adhere to age and sex-specific cancer and prevention screening recommendations. However, there were variations in levels of adherence within all groups. As expected, adherence to all modalities was greater by age except Pap testing, which is initiated at an earlier age. Adherence to all modalities was associated with having a personal history of cancer and fewer barriers to care.
For both breast cancer and cervical cancer screening, those women who provided emotional support only were less likely to be screened than those who provided more support. This was at odds with both our primary and alternative hypotheses. We had expected that either more exposure to cancer would encourage screening, or that fatalism associated with observation of poorer cancer outcomes in this lower-income community would discourage screening. The pattern of results here suggests a more nuanced process: individuals who provided only emotional support had only a limited view of the prognosis and journey of a cancer patient. They likely were only present for the particularly difficult, emotional aspects of the experience. This may have shaped their perspective on the benefits of therapy and the potential for a favorable prognosis of those with cancer. In short, limited contact when patients were particularly distressed may have fueled more fatalistic beliefs, lessening screening adherence. Alternatively, caregivers who provided more tangible support may have stepped into a partnership with the cancer patient that led to a comprehensive understanding of the battle against cancer. They may have seen more of the ups and downs and became more aware of the resources available to support cancer treatment. They may also have gained a greater understanding of the importance of early detection. Whatever the reason, these hands-on caregivers were more likely to be screened than people who had no personal experience with caregiving, indicating that exposure alone was not leading people to forego screening.
Even among more active caregivers, we found that both cervical cancer screening and prostate cancer screening were greater among those who provided more (versus less) intensive support to a cancer patient. Again, we hypothesized that those who had more intensive involvement were more likely to be screened due to their close-up and intimate understanding of the cancer journey. Heightened involvement may have made them more acutely aware of the benefits of being screened. Our findings are consistent with Song’s study, which suggested that higher rates of spousal caregivers’ screening and use of preventive services may be associated with shared marital or family context [27]. By living together and eating meals together, caregivers may be inclined to shift their own health behavior to be consistent with that of their family members. More involved caregivers may also be presented more chances to reflect upon their own lifestyles and mortality. The additional exposure to providers and hospitals may promote greater knowledge, easier access, and more motivation to pursue preventive care.
Cancer attitudes also played a significant role in rates of male colorectal screening. Greater pessimism about cancer prevention and treatment, an underlying foundation of fatalism, drove down screening. A large contributor to these attitudes towards cancer is the fatalistic belief that events are controlled by external forces, and humans are powerless to influence them. The associated helplessness can be tied to reduced motivation and self-efficacy and overall poor health behavior [15,22]. Consequently, these individuals become less likely to perceive willpower as crucial in shifting their health behavior and health outcomes.
Incidental findings
The high proportion of men who did not receive prostate cancer screening compared to other screenings likely reflects mixed and unfavorable recommendation for this testing from the USPSTF. This suggests that physician counseling in initiating the PSA conversation is a pivotal step to effective screening. Prior studies have found that strong family history also predicts prostate cancer screening [32]. This suggests that individuals who perceived that they were at greater risk of developing cancer, by lifestyle or family history, were more aware of recommended guidelines and more motivated to be screened. Conversely, the group that felt themselves to be at low risk also received PSA testing, perhaps due to decreased fatalism, lack of concern about potential harms of testing, or overall high receptivity to preventive care.
Limitations
The results of this study had limitations. The survey did not characterize the type of cancer affecting the patient relationally connected to the participants. With lack of knowledge on the survivability and prognosis of the cancer, it is difficult to ascertain the attitudes toward cancer and screening that may have been shaped by the caregiving experience. Additionally, participants were not asked about their familial risk or genetic syndromes, which would heighten perceived personal risk and explain screening shifts. The exact relationship of caregiver to familial cancer patient was also not determined. Future studies can focus on the degree of relationship.
The subset of participants who had a spouse with cancer to whom they provided no support or only emotional support was very small. This limited the analyses possible for this subgroup.
Colorectal screening data was also restricted to screening within the last five years. Without a ten-year frame to include ACS colonoscopy guidelines, full analysis was limited.
Overall, this survey was collected from volunteers, and the non-random process and self-reported data may lead to selection bias within the sample. However, of note, a full range of behaviors and attitudes were discovered within the survey.
Conclusions
Having a family member with cancer is a normative experience. Being involved with someone with cancer, whether friend, family, or spouse, affects one’s own preventive care. Importantly, those with a family member or spouse with cancer typically have increased personal screening. And as involvement in cancer caregiving deepens, one’s own screening typically rises as well. Providing only emotional support is the sole exception, by limiting perspective on the cancer experience and contributing to fatalism.
Our findings suggest that health professionals would do well to ask their patients about the prevalence of cancer within the family and their provision of cancer support to their social circles. Education can shift to target the patient. The focus can particularly be on the discrepancies in screening in those who provide only emotional support, addressing their fatalistic beliefs by providing accurate and individualized knowledge about screening.
The study provides a strong rationale for creating standardized tools for primary care, preventive medicine, and public health interventions to elicit the extent of cancer caregiving in their communities, then provide targeted counseling and resources to individuals and groups to promote cancer screening.
Data Availability
All relevant data are within the manuscript and its Supporting Information files.
Acknowledgments
We would like to acknowledge the support of Nicole Harris-Hollingsworth, EdD (Hackensack Meridian Health) and Howard Strickler, MD, MPH (MECC) for their assistance with this study; Mindy Ginsberg, Director of the Epidemiology and Clinical Research Informatics Shared Resource, for assistance with obtaining our sample; Rita Rivera for assistance with Spanish language interviews.
References
