An Environmental Scan of Biopsychosocial and Clinical Variables in Cohort Studies of Cancer Survivors

According to the National Cancer Institute's (NCI) Office of Cancer Survivorship, “an individual is considered a cancer survivor from the time of diagnosis, through the balance of his or her life (1).” Recent estimates indicate that 15.5 million people in the United States are cancer survivors, and the number of cancer survivors will increase over the next few years as the probability of survival after a cancer diagnosis continues to improve (1, 2). As the number of cancer survivors increases, it is becoming more important to assess the short- and long-term consequences of cancer and its treatment, including the emotional, behavioral, economic, lifestyle, and psychosocial ramifications (3). In addition, a new body of research has begun to examine how biomarkers (e.g., cortisol, inflammatory markers) relate to these psychosocial outcomes (4, 5).

Although medical research is expanding to address the psychosocial and behavioral challenges experienced following a cancer diagnosis, these data are often collected from cross-sectional and retrospective studies, limiting the inference that can be derived from them. As summarized by Elena and colleagues (6), few existing cohort studies were designed with the ability to examine the role of behavioral, psychologic, and social contextual factors on health outcomes after a cancer diagnosis. Therefore, a rigorous, multidisciplinary approach is needed to capture the full breadth of cancer survivors' experiences in prospective longitudinal studies and identify the gaps among these studies.

A 2011 paper by Harrop and colleagues (7) provided a review of current cancer survivorship research efforts and a summary of current survivorship research at NCI-designated Cancer Centers. They found that all cancer types required more attention from the research community and encouraged focus on late effects of treatment among cancer survivor cohorts (7). In 2015, the American Society of Clinical Oncology (ASCO) Survivorship Committee identified the need to better understand the current scope of cancer survivorship research to identify future research opportunities (8). They surveyed ASCO members and found key gaps in current cancer survivorship research including several common cancers (e.g., colorectal, prostate, lung), long-term survivors, older and younger survivors, and on survivorship care delivery and quality (8).

A centralized inventory of information gathered from cancer survivor epidemiology cohorts is needed to provide researchers with the behavioral and psychosocial antecedents associated with short- and long-term outcomes after the completion of cancer treatment. The first step is to examine the domains of measures and use of validated scales in cancer survivor cohorts. Thus, we conducted an environmental scan of cancer survivor cohorts, to determine the scope and scale of information collected on demographic, biopsychosocial, and selected clinical variables from cancer survivors. Environmental scans utilize a variety of approaches (e.g., literature review, a targeted survey) to determine and interpret the scope of the pertinent literature on a particular health outcome (9–12). This methodology is beneficial in the study of cancer survivors cohorts given the breadth of variables considered in cancer survivorship. These results will aid researchers in understanding the types of information currently available from cancer survivors, help identify gaps that future studies could address, and encourage future research opportunities for collaborative data pooling projects.

Cancer epidemiology cohort studies with data collected on cancer survivors were identified and surveyed to assess which psychologic characteristics, social context, patient demographics, patient clinical and diagnostic information, comorbidities, medication use, lifestyle, biomarkers, biospecimens, patient-reported symptoms, and clinical treatment information have been collected.

Cohort studies were included if they met the following criteria: (i) data for cancer survivors were collected at least once in the time period following cancer diagnosis, and (ii) the follow-up period for cancer outcomes was at least one year. Only studies conducted in the United States and published in English were considered. Cohort studies that were (i) solely focused on cancer screening or cancer etiology and risk factors of cancer incidence, or (ii) a registry database, biorepository, consortium, or web-based support forums/programs/message boards without follow-up data were excluded. Eligibility assessment of these cohorts was performed by two reviewers (J.L. Krok-Schoen and B.M. Bernardo).

Using these eligibility criteria, we conducted a literature review of the cancer survivorship literature using the PubMed and MEDLINE databases, limited to original research in English and cohorts that were active as of October 30, 2015. Reference lists of identified articles were reviewed for potentially eligible studies. We searched the PubMed database using key words “cancer survivor” combined with each of the following terms: “psychologic,” “psychosocial,” “health behavior,” “health outcomes,” “quality of life,” “social determinants,” “lifestyle,” “late effects,” and “long-term effects”. A separate PubMed and MEDLINE search was conducted for the key words “cancer survivorship” in the title or abstract.

The investigators then reviewed the studies to identify active cancer survivor epidemiology cohorts collecting behavioral, psychologic, and social context data. Queries were also made using the NIH RePORTER (https://projectreporter.nih.gov/reporter.cfm), the Cancer Epidemiology Descriptive Cohort Database (CEDCD; https://cedcd.nci.nih.gov/), and NCI Cohort Consortium (http://epi.grants.cancer.gov/Consortia/cohort.html) tools to identify additional cancer survivor epidemiology cohorts that may not have been captured in the literature review, by searching for currently funded projects in the United States using the key word, “survivor” (Fig. 1).

A web-based REDCap (13) survey was developed to assess behavioral, psychologic, social, biomarker, biospecimen, patient symptoms and comorbidities, and clinical treatment information collected in the identified cohorts, as well as the method of measurement and source of metadata for these variables. The web-based survey was modeled after an earlier survey (9) used to collect information about cancer survivor cohorts for a different purpose.

Principal investigators (PI) or project representatives (identified by the PI as the staff person who is most knowledgeable about the project) of the identified cancer survivor epidemiology cohorts were sent an email invitation to participate in October 2015. The invitation introduced the study, provided contact information for a study representative about questions or concerns, and provided a link to the web-based survey. Potential project representatives were asked to complete and return the web-based survey within 1 month. A reminder e-mail was sent two weeks after the initial study email, a third reminder e-mail in December 2015, and a final fourth e-mail 1 month later. Two telephone calls were also made to reach the nonresponders. After the sixth attempt to contact, those potential respondents were considered “nonresponders.” Study information was gathered for 24 nonrespondents from online searches of PubMed articles and websites including NIH RePORTER, clinicaltrials.gov, epi.grants.cancer.gov, grantome.com, fundedresearch.cancer.gov, epi.grants.cancer.gov, pcori.org, and the nonresponder's institution/study website. Sources of abstracted study information were the abstracts of grants, methods, and results sections of peer-reviewed publications, and study summaries posted on study websites.

The REDCap survey asked whether the following domains were collected in the cohort: psychologic characteristics, social context, patient demographics, patient clinical and diagnostic information, patient-reported symptoms, comorbidities, medication use, complementary and alternative medicine (CAM) use, lifestyle, biospecimens, and biomarkers processed from biospecimens. Each domain is further described below.

Respondents to the online surveys were asked to identify (yes/no) which psychologic constructs and common validated scales were collected in their respective cohort study. Surveyed participants were provided with validated scales measuring specific psychologic constructs and were asked to select any validated scales used to collect information from cancer survivors. Examples of psychologic constructs and validated scales used to measure each construct include: anxiety (14), body image (15), depression (16), fear of cancer recurrence/cancer worry (17), loneliness (18), mood/affect (19), neuroticism (20), optimism (21), perceived cognitive ability (22), perceived stress (23), posttraumatic stress disorder (24), quality of life (25), satisfaction (26), self-efficacy (27), sexual functioning/fertility issues (28), and well-being PROMIS (29).

Social constructs were identified by asking respondents to select (yes/no) which common validated scales associated with those items were collected from their respective cohort study. Surveyed participants were provided with a number of validated scales measuring these social constructs and were asked to select any validated scales they used when collecting information from cancer survivors. Social constructs included but were not limited to discrimination (30), financial hardship, natural, and built environment characteristics, religiosity (31), social support (32), and survivor caregiving issues.

The administered surveys assessed types of demographic information collected within cohorts. Participants were asked (yes/no) which survivor demographic data were collected in their cancer survivor cohort study, including survivor age, country of origin, educational attainment, ethnicity, family history of cancer, health insurance status, household size, income, marital status, number of children, occupation, primary language, race, sex, sexual orientation, time spent at current residence, type of residence, U.S. citizen status, and veteran status.

To determine which patient clinical characteristics and diagnostic information were collected in the cohort studies, the survey asked which of the following clinical information was collected: cancer stage, grade and type, receipt of chemotherapy, diagnosis date, diagnostic tests, follow-up period, functional status, palliative/end-of-life care, provider information, recurrence, second cancer, time from diagnosis to treatment, treatment type, tumor location, markers, node, size, HPV vaccine receipt, and use of health care services.

The survey also evaluated the types of symptom data that cohort studies collected from participants. Symptom data included a myriad of symptoms that are commonly reported among cancer survivors, included but not limited to aching joints, appetite change, coughing/wheezing, difficulty breathing, concentrating, sleeping, and swallowing, drowsiness, fatigue, fever, hair loss, headaches/migraines, hearing change, irritability, nausea, night sweats, nerve problems, neuropathy, pain, problems with sexual functioning/interest/activity, swelling of limbs/hands/feet, urinary/bowel incontinence, vaginal/genital irritation/dryness, vision problems, vomiting, and weight gain/loss.

Surveyed PIs or study representatives were asked (yes/no) which patient comorbidity data were collected in their cancer survivor cohort study. Comorbidity data included conditions commonly reported among cancer survivors, including but not limited to angina, diabetes, emphysema, frequent infections, generalized anxiety disorder, heart or coronary artery disease, high cholesterol, hypertension, kidney disease, major depressive disorder, osteoarthritis, osteoporosis, rheumatoid arthritis, second primary cancer diagnosis, skin disorders, and vascular disorders.

The survey assessed whether medication data were collected from PIs or study representatives. Information regarding names of medications, dosage information, immunization history, medication adherence, medication side effects, prevalence/mean number of any medications and prescription medications, therapeutic category, and types of medications were collected.

Respondents to the surveys indicated whether information on complementary and alternative medicine (CAM) therapies were collected in their cancer survivor cohort study. Several CAM therapies were considered, such as acupuncture, aromatherapy, biofeedback, chiropractic or deep breathing exercises, diet-based therapies, hypnosis, massage, meditation, music therapy, natural products, prayer/religious practices, progression relaxation, qi gong, reflexology, supplements, tai chi, and yoga.

The survey also determined whether cohort studies collected various lifestyle information from cancer survivors. Lifestyle information included ability to perform daily activities (33), alcohol use (34), body mass index, diet/nutrition, physical activity/exercise (35), risky sexual behaviors, self-rated status, sleeping habits (36), smoking/tobacco use, stress management (37), sun exposure, traumas/periods of hardship (38), and weight change.

To determine the extent of biospecimen data collection among cancer survivors, survey respondents indicated whether or not their study collected biospecimens from participants, and if so, which specimens were collected. Potential biospecimens could be blood, buccal cells, buffy coat, DNA, expired carbon monoxide and salivary cotinine, feces, organ tissues samples, saliva, tumor tissue samples, and urine.

Finally, the surveys asked participants to indicate which biomarkers were processed from collected biospecimens within their cohort study. Processed biomarkers included adipokines, androgen, C-reactive protein, carcinoembryonic antigen, cortisol, cotinine, epidermal growth factors receptor, estrogen, fibrin, glucose, HER2/NEU, hormones, human chorionic gonadotropin-β, inflammatory markers, insulin, insulin-like growth factors, interleukin, leptin, prostate-specific antigen, serum tumor markers, sex hormone, and TNF.

To determine the frequency of collected domains and the common validated scales, descriptive frequency tables were generated using IBM SPSS Statistical Program (Version 23).

The PubMed and MEDLINE searches for the key words resulted in 318 articles, 26 (8.2%) of which met the inclusion criteria. Of these, 15 (57.7%) were duplicated from what was found in the online database searches. Searches on NIH RePORTER and the NCI Cohort Consortium produced 395 and 42 studies, respectively. Out of 437 studies in these databases, 120 (27.5%) studies met the study criteria. Thus, 11 studies were identified through PubMed and MEDLINE, 120 were found in the NCI Consortium and NIH RePORTER resulting in a final sample of 131 studies identified as active cancer survivor epidemiology cohorts.

A total of 131 studies were invited to participate, of which 13 (9.9%) were ineligible for participation (e.g., study was not an epidemiologic study on cancer survivors). An in-depth review of the 118 studies was conducted and an additional 56 (47.5%) studies were excluded due to being duplicate subsets of survivor cohorts and intervention studies rather than cancer survivor epidemiology cohorts (Fig. 1). After the second review of studies, the final sample included 62 cancer survivor epidemiology cohort studies, 38 (61%) of which responded directly to the study survey and 24 (39%) of which were abstracted from online sources described above.

Table 1 provides a breakdown of studies (N = 62) by type(s) of cancer survivors studied. Out of the 62 studies that provided cancer type, 39 (62.9%) studies studied multiple types of cancer and 23 (37.1%) were single site studies. Most studies included breast cancer survivors (n = 34, 54.8%), which is considerably more than the number of studies that included other cancer sites, lymphoma and leukemia, which consisted of 23 (37.1%) and 22 (35.5%) studies, respectively. Survivors of colon or rectal cancers or Hodgkin disease were the next most researched with 21 (33.9%) and 20 (32.3%) studies, respectively. Eight (12.9%) studies focused on childhood cancer survivors.

The majority of the cancer survivor epidemiology cohort studies (n = 59) collected information on clinical characteristics and basic diagnostic information. Studies also collected information on patient demographic characteristics (n = 57), patient-reported symptoms (n = 44), lifestyle (n = 45), and psychologic (n = 42) characteristics associated with health behaviors, health outcomes, and well-being. Less commonly collected information included biospecimens (n = 35), biomarkers (n = 35), CAM use (n = 19), and social characteristics (n = 27; Table 2). Survey respondents and nonrespondents' study information are combined to provide an inclusive summary of the information being gathered in currently funded cancer survivor cohorts.

In the psychologic domain, the most common information collected was quality of life (n = 33), depressive symptoms (n = 22), and anxiety symptoms (n = 20). The least common psychologic information collected was motivation (n = 0), neuroticism (n = 1), conscientiousness (n = 0), self-control (n = 1), and health literacy (n = 1). In the social domain, the most common information collected by the cancer survivor cohort studies was type of social support received (n = 17), financial hardship (n = 7), survivor caregiving issues (n = 6), religiosity (n = 4), and acculturation (n = 4). The least common social information collected within the studies was civic participation (n = 0), culture (n = 0), and sense of community (n = 0). Studies that confirmed the collection of demographic information most often gathered the age (n = 53), race (n = 47), ethnicity (n = 46), education (n = 38), marital status (n = 38), and sex (n = 35) of their participants. The least common demographic information collected was participants' citizenship status (n = 0), veteran status (n = 2), and sexual orientation (n = 2).

Of the cancer survivor epidemiology cohort studies that collected clinical information, the most common information gathered was on treatment type (n = 54), cancer stage and grade (n = 40), and receipt of chemotherapy (n = 41). The least common clinical information gathered was the receipt of palliative, end-of-life care (n = 3), tumor vaccine (n = 3), and clinical trial enrollment (n = 6). The most common reported symptoms collected included pain (n = 23), fatigue (n = 18), and sleep disturbance (n = 14). The least common symptom information gathered was radiation burns (n = 1), hair loss (n = 2), and bruising easily (n = 2). The most common comorbidity information collected was on diabetes (n = 30), heart/coronary artery disease (n = 29), and hypertension (n = 23). The least common comorbidity information gathered was about alcohol hepatitis (n = 2), nonalcoholic steatohepatitis or nonalcoholic fatty liver disease (n = 2), and mononucleosis (n = 2). Finally, the most frequently collected medication information in the cancer survivor cohorts included the names of medications (n = 21), medication types (n = 20), and therapeutic category (n = 22). The least frequently collected information in the medication domain included medication side effects (n = 4), medication adherence (n = 4), and immunization history (n = 2).

In the CAM domain, the most commonly collected information was about supplement use (n = 7), chiropractic (n = 5), massage (n = 5), acupressure (n = 4), acupuncture (n = 4), diet-based therapies (n = 4), and yoga (n = 4). Information about ayurveda, chelation, homeopathic treatment, labyrinth walking, movement therapy, progression relaxation, qi gong, reflexology, and traditional healers, were not collected by any study. In the lifestyle domain, studies most often collected information on exercise/physical activity (n = 30), BMI (n = 31), and smoking/tobacco use (n = 30). The least frequently collected information within the lifestyle domain was hydration habits (n = 1), mindfulness (n = 1), and stress management (n = 0).

Of the 35 studies that collected biospecimens, the most commonly collected specimens were blood plasma (n = 23) and tumor tissues (n = 14). The least common biospecimen collected was feces (n = 1) and toe/fingernails (n = 1). Of the 35 studies that explored biomarkers, the most common biomarkers examined were insulin (n = 6), C-reactive protein (n = 6), and glucose (n = 5). No studies reported assessment of α-fetoprotein, cancer antigens, catecholamines, chromosomes 3, 7, 9, 17, cotinene, cytokeratins, expired carbon dioxide, gonadotropin, human complement factor BTA, interleukin, KIT gene, nuclear matrix protein, estrogen receptor, and thyroglobulin.

Validated scales were commonly used to collect information from cancer survivors in three domains—psychologic, social, and lifestyle. Forty-five cohort studies reported using validated scales within the lifestyle domain, 42 cohort studies and 27 cohort studies reported using validated scales within the psychologic and social domains, respectively. In the psychologic domain, the most common scales used were the 12 and 36-item short-form health surveys [SF-36 (n = 9); SF-12 (n = 6)], and the Centers for Epidemiological Scale-Depression [CES-D (n = 9)]. The most common scales used in the social domain were the Functional Assessment of Cancer Therapy [FACT (n = 2)], Medical Outcomes Study [MOS (n = 1)]. Within the lifestyle domain, the most common scales were the Medical Outcome Study (n = 10), Harvard Food Frequency Questionnaire (n = 6), Pittsburgh Sleep Quality Index (n = 6), in addition to NHANES (n = 6). Four scales (FACT, MOS, PROMIS, and SF-12) were used to collect information over multiple domains (Table 3). Study aims and general information for the cohorts can be found in Table 4. Finally, Table 5 reports scales included by cohort respondents.

The purpose of this study was to conduct an environmental scan of cancer survivor cohorts, to determine the scope and scale of information collected on demographic, biopsychosocial, and selected clinical variables from cancer survivors. The increase in the number of cancer survivors living longer has brought attention to the importance of understanding and addressing the needs of this population (9). Several key findings are apparent from this study. First, the majority of studies included breast cancer survivors, while other major cancer sites, including lung, colon, and prostate, are not as well-studied. Breast cancer is the most common cancer among women with favorable survival rates, increasing the need to address problems these survivors face (2). In addition, due to their large numbers, it may be easier to recruit breast cancer survivors compared with other cancer types. However, there are important research questions that need to be addressed for cancer survivors across cancer site. It may be more difficult to recruit cancer survivors with shorter life expectancies, yet there are many other cancers with favorable survival rates that can be recruited into cohort studies. For example, we found that approximately 23% of the studies focused on lung cancer survivors yet, in 2017, lung cancer has the second highest incidence rate than any other cancer in both men and women, and with targeted treatments, some survivors are living for months and years with significant symptoms (2).

Second, the majority of the studies examined clinical, demographic, psychologic, and lifestyle measures, whereas fewer studies gathered information about social factors, CAM use, and collected biospecimens. Although a clinical focus (e.g., cancer type, cancer stage) is inherent in cancer survivor studies, more effort should be made to not only collect more detailed and complete data on these clinical variables but also expand into symptoms, adverse outcomes/toxicities, and treatment received. Another important yet understudied area of cancer survivorship is the use of CAM. In this study, we found that 30.7% of the studies collected information about the use of CAM in their studies. However, previous studies have found up to 67% of cancer survivors reported CAM use within the past year (39, 40). Future studies should consider measuring CAM use because it is a common practice that could interact with treatment or disease progression, as well as affecting other psychologic, social, and clinical characteristics of cancer survivors.

Another research gap is the collection of information related to social domain among cancer survivors. In this study, we found that less than half of the studies explored social factors in cancer survivorship. However, social factors such as acculturation, culture, and discrimination have repeatedly been shown to influence health behaviors of cancer survivors (41–43) and are of particular interest to those interested in health disparities. Furthermore, individuals interested in health disparities should understand that where people live including, but not limited to, their environment, community cohesiveness, and healthcare access impact their health (44). Future research should take a broader, macro-level approach to cancer survivors' social environment, including sources of support, and geographic surroundings to develop strategies that can be implemented to better address individual needs accordingly.

An area of growth that was observed in this study is the proliferation of research using biospecimens and biomarkers. Biomarkers provide an objective approach to understanding the etiology of both side effects, long-term effects and survival with applications in observational and analytic epidemiology, randomized clinical trials, screening, diagnosis, and prognosis (45). Disadvantages of biomarker research such as cost, timing, and storage, may have deterred many cancer survivor cohort studies from collecting biospecimens for biomarker research. Transdisciplinary research is needed to link basic clinical research with psychosocial research to better understand how these areas are related. For example, which survivors are likely to respond more favorably to certain treatments or have side effects from therapies as determined (46). To address these research gaps regarding biospecimens, social factors, and CAM, future research should consider utilizing a multilevel approach, such as the Multi-level Biological and Social Integrative Construct (MBASIC; ref. 47), which encourages transdisciplinary collaborations to streamline intervention, implementation, and translation efforts.

Finally, there is a need for having the same types of data collected in a parallel sample of similar individuals without cancer and across studies for more efficient data pooling. Wider use of standardized instruments such as those used in population-based samples (e.g., NHANES) would support comparisons within and between survivor cohorts and population-based samples and help to put into perspective the needs/experiences of cancer survivors. In addition, efforts to harmonize validated measures and assessment tools (e.g., SF-36, CES-D) for use among cancer survivors would facilitate pooling data across studies for future research.

This study had several limitations. First, the data gathered about cohort studies that did not respond to the survey was based on available data, which may not have been as comprehensive or coded with the same detail as data provided by survey respondents. Because of the low rates of response to the surveys, study authors utilized internet searches and online databases to gather information regarding cohorts that were nonrespondents. These websites were best source of study information for currently ongoing cohorts. Second, results were limited to the search terms and phrases used to identify cohorts in the PubMed and MEDLINE literature review as well as the NIH RePORTER query; additional terms may have identified additional studies, including clinical trials, which may have provided psychosocial, symptom, and biospecimen data from participants. New studies that have yet to publish would likely have been missed. This study also only included U.S.-based cohorts, which limits the generalizability of study findings. Finally, the internet-based questionnaire was mostly close-ended to facilitate data collection and reporting; however, domains not included in the questionnaire were excluded from this report limiting the ability to fully characterize the breadth of information collected across all cancer survivor cohorts.

In summary, cancer survivorship research continues to grow as has the population of survivors (2, 3). Researchers have many opportunities to explore the myriad of issues survivors of all cancers experience. To maximize the opportunity to address the needs among cancer survivors, future research should include survivors of other types of cancers in addition to breast cancer and collect information about understudied areas including biospecimens/biomarkers and social and environmental factors.

E.D. Paskett reports receiving other commercial research support from Merck Foundation and has ownership interest (including patents) in Pfizer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.L. Krok-Schoen, J.W. Elena, E.D. Paskett

Development of methodology: J.L. Krok-Schoen, J.W. Elena, E.D. Paskett

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.L. Krok-Schoen, B.M. Bernardo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Peng, E.D. Paskett

Writing, review, and/or revision of the manuscript: J.L. Krok-Schoen, B.M. Bernardo, J.W. Elena, P.A. Green, E. Hoover, J. Peng, G.L. Anderson, B. Caan, L. Johnson, E.D. Paskett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.L. Krok-Schoen, J.W. Elena, E.D. Paskett

Study supervision: E.D. Paskett

This work was supported by the Supplement to the Women's Health Initiative Life and Longevity After Cancer (LILAC) grant (UM1CA173642).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.