Abstract
Most relatives of women with ovarian cancer are unaware of their increased risk for cancer and their eligibility for genetic counseling. State cancer registries offer a platform to communicate about inherited risk to this population.
We conducted a two-arm randomized trial to test a theory-based communication intervention—Your Family Connects (YFC)—compared to the standard Georgia Cancer Registry (GCR) contact. A total of 1,938 eligible ovarian cancer survivors were randomly assigned to either the YFC arm (n = 969) or the Standard Care arm (n = 969). We assessed the number of ovarian cancer survivors and their close relatives who logged on to the study website by arm.
Survivor reach was significantly higher in the Standard Care arm than YFC (20.8% vs. 15.2%, respectively; P < 0.001). However, reach to relatives was limited to listed relatives in the YFC arm (n = 20, 13.2%), with little participation from those in the Standard Care arm (n = 1, 0.4%). Pooling across arms, minority race, longer time since diagnosis, and older age were all significantly associated with a decreased likelihood that the survivor accessed the website.
The YFC intervention showed lower effectiveness for engaging survivors but was more effective than Standard Care in engaging at-risk relatives. Other factors (e.g., time since diagnosis) associated with lower reach must be considered in refining future outreach approaches.
Partnering with a state cancer registry to foster family communication about inherited cancer risk is feasible but the possibility for broad population reach warrants further testing.
Introduction
Ovarian cancer is the fifth leading cause of cancer-related deaths among women in the U.S. Researchers attribute 15% to 20% of epithelial ovarian cancers to inherited mutations in cancer predisposing genes that can influence risk among first- and second-degree relatives (1). Currently, the Health Insurance Portability and Accountability Act prohibits healthcare providers’ from reaching out directly to at-risk relatives about their potential hereditary risk. Thus, the duty to communicate hereditary risk to relatives has fallen to the patient, most often to patients identified to carry a pathogenic variant (i.e., proband). The current standard of care is to advise the proband to inform at-risk relatives and encourage them to seek genetic counseling.
Unfortunately, this approach leaves more than half of at-risk relatives unaware of their risk and the importance of seeking genetic counseling (2–6). A recent systematic review of 14 studies of families with inherited cancer syndromes, identified only eight that assessed family communication outcomes (e.g., the number of relatives contacted and/or informed of a family member carrying a pathogenic variant; ref. 7). These eight studies showed negligible improvements in family communication (g = 0.085, P = 0.344; ref. 7). In response to the low uptake of cancer genetic services among families at risk for hereditary ovarian cancer, the NCI initiated the Traceback Program PAR-18-616 to support pilot projects testing approaches to motivate women with ovarian cancer to refer at-risk relatives for genetic services. National guidelines recommend that all first- or second-degree relatives of those diagnosed with epithelial ovarian cancer seek genetic counseling regardless of whether the family member with cancer has undergone genetic testing and is found to carry a pathogenic variant (8–10). This low uptake by relatives suggests a need to consider additional avenues for contacting relatives—avenues that do not rely solely on the patient with ovarian cancer to initiate communication.
The challenge going forward is to develop and test contact strategies that preserve the autonomy and privacy of the patient with cancer while increasing risk awareness and access to genetic counseling among their close relatives. Thought leaders in bioethics stress the “relational autonomy,” the notion that individuals exist within a web of relationships with other people who share in decision-making (11–13). The quality and nature of these relationships can differ greatly suggesting a “one size fits all” communication strategy may not be optimal. Relational autonomy is supported when there are multiple contact options and no attempt is made to persuade or manipulate preferences.
Additionally, asking patients with cancer to provide information about close relatives can be viewed as a form of “helping behavior.” Social marketers frequently have used the foot-in-the-door technique (FITD; ref. 14) to encourage helping behavior. FITD entails first presenting a small request that is easy to accomplish and relevant to the individual’s identity (e.g., ovarian cancer survivors visit website to answer questions that could help close relatives), followed by a large request (e.g., list their relatives and choose an option for contacting them). Complying with the first, easy request increases compliance with the larger request, particularly when the small request is aligned with positive identity, prosocial norms, and made by an organization seen as benefiting the community (14, 15).
To expand the reach of communication interventions to whole populations of individuals with cancer and their families, contact approaches must be scalable, affordable, and sustainable (16, 17). Statewide cancer registries with the mandate to conduct surveillance activities offer a platform that can capture all relevant cancers in the population (18, 19). Although researchers have applied these registries to facilitate research recruitment (19, 20) and promote uptake of preventive screening, their potential for expanding family communication about genetic services is underexplored (20).
Technology monitoring groups in the U.S. show high rates of penetration for internet access (95%) and cell phones (97%; ref. 21). Websites and short message services (SMS) have the advantage of providing affordable and sustainable 24-hour access to health information to geographically dispersed households (22, 23) in locations where genomic services may be limited (24).
To this end, we collaborated with the Georgia Cancer Registry (GCR) to develop an interactive website Your Family Connects (YFC) to reach women diagnosed with ovarian cancer in the state of Georgia and their first- or second-degree relatives. Using a two-arm randomized trial, we aimed to compare YFC to the standard cancer registry outreach and to assess its impact on reaching survivors of ovarian cancer and their close relatives and to offer free genetic counseling. We hypothesize that the YFC would achieve a greater reach among survivors and their relatives compared with the standard outreach.
Materials and Methods
Study population
The GCR collects patients with cancer’s demographic characteristics (e.g., age at diagnosis, race, gender) and cancer characteristics (e.g., cancer site, stage, histologic types). Women identified by the GCR were eligible for this trial if they: (i) were an adult (18+); (ii) were diagnosed with ovarian, fallopian tube, or peritoneal cancers between January 2005 and December 2017 in the state of Georgia; and (iii) were still living per the registry’s records. We defined relatives as eligible to participate if they were a first- or second-degree relative of the survivor, 25 years or older (age recommended to initiate preventive behaviors; ref. 25), able to access the internet, able to read English, and not incarcerated or institutionalized.
We deliberately chose to reach out to ovarian cancer survivors diagnosed as early as 2005 based on: (i) our communication outreach approach that targets ovarian cancer survivors with short- and long-term survivorship to include diagnoses that predated changes in testing guidelines (18); (ii) our recognition that survivors diagnosed more than a decade ago are underrepresented in the majority of case identification trials with self-selected samples in specialty clinical settings; (iii) our goal to maximize reach and learn about the intervention efficacy across a range of years since diagnosis, even though we expected rates of recruitment and enrollment to decline as we trace back; and (iv) the cost advantage of using the state-wide cancer registry to identify all of the locatable survivors in Georgia.
Study design
We designed the randomized trial to test the benefit of the additional components included in the YFC intervention compared with the GCR’s standard outreach approach (mailed materials and telephone follow-ups). Identified survivors were randomized in a 1:1 ratio to the YFC or Standard Care arms. We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline (Fig. 1). All study procedures were approved by the Institutional Review Board at Emory University.
Intervention
Figure 2 displays the elements of the intervention linked with the associated theoretical frameworks. A detailed description of the development of the YFC intervention appears elsewhere (26, 27). Our intervention, delivered through mailed packages and a web-based platform, aimed to promote participant engagement while strictly adhering to the privacy standards outlined in GCR’s outreach policy. This policy requires that the registry be the first to reach out to potential participants, ensuring their privacy is protected. The two arms of our intervention, Standard Care and YFC, varied in terms of the content, graphic design, and user navigation procedures of both the GCR’s mailed package and the study’s website (Fig. 3). Individuals in both arms were provided an access code for entering their respective website and were consented online prior to creating a website account if interested. Further details appear below.
Standard Care arm
Mailed package
Ovarian cancer survivors randomized to the Standard Care arm were mailed a package by the GCR that included a standard study invitation letter, an access code to the study website, a consent form for study website access, and $5 as a token of appreciation. These materials were mailed to survivors in a standard white catalog envelop used by the GCR.
Website sections
The Standard Care arm website sections included concise information about genetic risk for ovarian cancer, genetic counseling, availability of local patient navigators, and the importance of informing close relatives about their possible elevated risk for cancer. On entering the total number of relatives in the designated website portal, survivors received an equivalent number of unique relative access codes and editable letters generated by the system that they could download. The codes for the survivors and relatives were linked. Survivors could share the letter with their relatives either through e-mail or by postal mail.
Relatives who received the downloaded letters from the survivor were able to access the same website using their access code, consent to the study, and then create a website account login for future use.
YFC arm
Mailed package
We sent two mailings to survivors in the YFC arm via the GCR. The first warm-up contact mailing included an infographic and a three-question survey about ovarian cancer information seeking. The mailing also included a postage-paid envelope to return their responses. These short surveys were anonymous and represented an initial small request to survivors to visit the YCF website. We sent a second initial mailing packet after 7 to 10 days in a custom teal catalog envelop to match the ribbon color for ovarian cancer awareness. Individuals did not have to return the brief survey to receive the recruitment packet. The packet included a theory informed study invitation letter, a consent form for study website access, and $5 as a token of appreciation. We developed the information in the recruitment packet based on the input of a team of citizen scientists who collected quantitative and qualitative information from a broad group of ovarian cancer survivors and their relatives recruited through a variety of sources (206 surveys; 39 structured interviews; ref. 26).
Website sections
Survivors in the YFC arm were instructed to enter the first names of all first-degree relatives (i.e., parents, children, siblings) and second-degree relatives (i.e., grandparents, grandchildren, aunts, uncles) and specify each relationship using a drop-down list (27). Survivors were presented with a “relative contact menu,” that offered three options: (i) self-contact (survivors download, edit the family letter, and contact their relatives directly), (ii) Study team contact (the study team sends the family letter to the selected relative via e-mail or postal mail, as per the survivor’s request), or (iii) delay the contact. Each contact option was accompanied by a brief description highlighting pros and cons of each approach and in which relationships the option might be optimal. For instance, a survivor contact might not be optimal if the participant had a personal history with the relative that complicates conversations. The YFC website also provided the rationale for reaching out to first- and second-degree relatives and a downloadable “tip” sheet for having sensitive conversations with relatives about inherited risk. Survivors also had the option to share their mobile numbers to receive SMS messages: (i) an initial message to thank them for using the website and (ii) a message urging them to revisit for further information and to reach out to relatives.
Relatives in the YFC arm received a unique website access code (linked to the survivor). The code enabled the relative to access YFC website content that was identical to that provided to the survivor. For survivors who opted to contact the relative themselves, the webpage provides a family letter with a unique log-in for the relative to access the YFC website. Survivors who selected the study team contact were asked to provide contact information (e.g., mailing address, e-mail, phone number) for the relative. These relatives received a letter signed by Drs. Guan and Bellcross regarding inherited cancer risk and an access code for the website. Relatives selected by survivors for delayed contact were not contacted. Relatives who provided cell phone numbers received the same two reminder SMS messages as the survivors.
Free telegenetic counseling service
The website offered survivors and relatives in both arms free telegenetic counseling service from supervised genetic counseling interns at Emory’s Genetic Counseling Training Program, supervised by Drs. Bellcross and Guan. Participants could choose a 1-hour individual session or a 1.5-hour family session. Study participants and genetic counselor interns were blinded to participant study arm allocation.
Randomization and recruitment
GCR identified a total of 1,938 eligible ovarian cancer survivors (Fig. 3). These women were randomly assigned to either the YFC arm (n = 969) or the Standard Care arm (n = 969). Randomization took place each month after removing individuals identified as deceased. We contacted survivors in batches of 172 dispatched each month. To test for balance in batches across arms, we generated a variable, “survival flag” that categorized the cohort into three groups based on the duration of survival post-diagnosis: (i) less than 5 years; (ii) 5 to <10 years; (iii) 10 years or more. Survival categories were distributed equally across the batches.
We aligned our recruitment strategies with the standard procedures used by the GCR (28–32). To maximize the likelihood the survivor would receive the mailed materials on the initial attempt, the GCR used Accurint, a LexisNexis database with coverage across the US, and state of Georgia voter registration files to obtain the most up-to-date address for each survivor. Throughout recruitment, we performed monthly checks against state vital records to avoid initiating contact with a recently deceased woman. We employed serialized contact strategies to prompt survivors in both arms to visit their assigned version of the website. Two weeks after the initial mailing, a GCR-trained research assistant called survivors who had not created a study website login using standard contact procedures that included calls made at different times of day, spaced 2 to 3 days apart, for up to nine attempts after which the survivor was categorized as “no contact”. The purpose of the phone call was to: (i) check on receipt of the recruitment packet, (ii) remind survivors to log in to the website, and (iii) answer questions the survivor had about the study. Callers were blinded to the study arm of the survivors.
Measures
Primary outcomes
We defined survivor reach as the proportion of eligible survivors contacted by the GCR who: (i) used their website access code to visit the study website, (ii) gave consent to participate in the study, or (iii) created a website login. We defined relative reach as the proportion of first- and second-degree relatives who: (i) used their access code to visit the assigned version of the website, (ii) gave consent to participate in the study, or (iii) created website login.
Secondary outcomes
Secondary outcomes included: (i) the contact option selected by survivors in the YFC arm and the number of relatives listed by survivors in both arms; and (ii) the number of survivors and relatives who scheduled and completed a telegenetic counseling appointment in both arms.
Participant characteristics
We obtained survivors’ sociodemographic and cancer data from the GCR. Available variables included survivors’ current age, race/ethnicity, age at diagnosis, time since diagnosis, cancer stage, marital status, health insurance type, and rurality.
Data analysis
Data collection spanned from July 2021 and February 2023. We used SAS 9.4 and SAS macros developed by the Biostatistics Shared Resource at the Winship Cancer Institute of Emory University to perform analyses and to test associations between variables of interest by study arms (33). We coded three primary outcomes of survivor reach outcomes (i.e., used the website access code, consented to participate, created a website login) as “0” or “1.” Summary statistics were applied to all covariates of interest. The univariate association between the two arms was described using ANOVA/Kruskal–Wallis test for numerical covariates and χ2/Fisher’s exact test for categorical covariates, where appropriate, based on the normality test of data distribution and the sample size. For the survivor access code use outcome, univariate and multivariable logistic regression was applied to model the probability of access code usage with the possible assessment of the interaction effect between the arms and covariates. We set alpha at <0.05.
Data availability
The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request. Data are located in controlled access data storage at Emory University.
Results
Survivor characteristics
Table 1 presents the sample characteristics. We observed no statistically significant differences between the YFC and the Standard Care arms on any patient characteristics, suggesting the randomization worked as intended.
. | Arm . | |||
---|---|---|---|---|
Covariate . | Total N = 1,938 (100%) . | Standard care n = 969 (50%) . | YFC n = 969 (50%) . | P-valuea . |
Current age | 0.233b | |||
Mean (SD) | 65.0 (13.4) | 64.7 (13.3) | 65.3 (13.5) | |
Median (Q1–Q3) | 66 (57–75) | 65 (57–74) | 66 (57–75) | |
Min–Max | 23–99 | 24–98 | 23–99 | |
Time since diagnosis | 0.621b | |||
Mean (SD) | 9.8 (3.8) | 9.7 (3.8) | 9.8 (3.7) | |
Median (Q1–Q3) | 9 (6–13) | 9 (6–13) | 9 (7–13) | |
Min–Max | 4–17 | 4–17 | 4–17 | |
Age at diagnosis | 0.328b | |||
Mean (SD) | 54.7 (13.5) | 54.5 (13.5) | 55.0 (13.4) | |
Median (Q1–Q3) | 56 (47–65) | 55 (46–64) | 56 (47–65) | |
Min–Max | 18–91 | 18–89 | 18–91 | |
Race | N (Col %) | 0.33 | |||
Other | 26 (1.3) | 15 (1.5) | 11 (1.1) | |
White | 1,477 (76.2) | 733 (75.6) | 744 (76.8) | |
Black | 381 (19.7) | 188 (19.4) | 193 (19.9) | |
Asian | 54 (2.8) | 33 (3.4) | 21 (2.2) | |
Hispanic | N (Col %) | 0.648 | |||
Non-Hispanic | 1,858 (95.9) | 931 (96.1) | 927 (95.7) | |
Hispanic | 80 (4.1) | 38 (3.9) | 42 (4.3) | |
Cancer stage | N (Col %) | 0.743 | |||
In situ | 41 (2.1) | 22 (2.3) | 19 (2.0) | |
Localized | 628 (32.4) | 301 (31.1) | 327 (33.7) | |
Regional | 614 (31.7) | 317 (32.7) | 297 (30.7) | |
Distant | 579 (29.9) | 291 (30.0) | 288 (29.7) | |
Unstaged | 76 (3.9) | 38 (3.9) | 38 (3.9) | |
Marital status | N (Col %) | 0.446 | |||
Single | 400 (21.5) | 204 (22) | 196 (21) | |
Married/domestic partner | 1,065 (57.2) | 536 (57.9) | 529 (56.6) | |
Separated/Divorced/Widowed | 396 (21.3) | 186 (20.1) | 210 (22.5) | |
Payer/Insurance | N (Col %) | 0.622 | |||
Not insured | 144 (8.1) | 67 (7.6) | 77 (8.7) | |
Private insurance | 1,065 (60) | 545 (61.4) | 520 (58.6) | |
Medicaid/Medicare | 532 (30) | 259 (29.2) | 273 (30.7) | |
Military/VA/Public health services | 34 (1.9) | 16 (1.8) | 18 (2.0) | |
Rural/Urban commuting area (RUCA 2010) | N (Col %) | 0.23 | |||
Urban commuting area | 1,577 (85.9) | 797 (86.9) | 780 (85.0) | |
Not an urban commuting area | 258 (14.1) | 120 (13.1) | 138 (15.0) | |
Urban/Rural indicator (URIC 2010) | N (Col %) | 0.763 | |||
All urban | 893 (48.7) | 446 (48.6) | 447 (48.7) | |
Mostly urban | 512 (27.9) | 256 (27.9) | 256 (27.9) | |
Mostly rural | 251 (13.7) | 131 (14.3) | 120 (13.1) | |
All rural | 179 (9.8) | 84 (9.2) | 95 (10.3) |
. | Arm . | |||
---|---|---|---|---|
Covariate . | Total N = 1,938 (100%) . | Standard care n = 969 (50%) . | YFC n = 969 (50%) . | P-valuea . |
Current age | 0.233b | |||
Mean (SD) | 65.0 (13.4) | 64.7 (13.3) | 65.3 (13.5) | |
Median (Q1–Q3) | 66 (57–75) | 65 (57–74) | 66 (57–75) | |
Min–Max | 23–99 | 24–98 | 23–99 | |
Time since diagnosis | 0.621b | |||
Mean (SD) | 9.8 (3.8) | 9.7 (3.8) | 9.8 (3.7) | |
Median (Q1–Q3) | 9 (6–13) | 9 (6–13) | 9 (7–13) | |
Min–Max | 4–17 | 4–17 | 4–17 | |
Age at diagnosis | 0.328b | |||
Mean (SD) | 54.7 (13.5) | 54.5 (13.5) | 55.0 (13.4) | |
Median (Q1–Q3) | 56 (47–65) | 55 (46–64) | 56 (47–65) | |
Min–Max | 18–91 | 18–89 | 18–91 | |
Race | N (Col %) | 0.33 | |||
Other | 26 (1.3) | 15 (1.5) | 11 (1.1) | |
White | 1,477 (76.2) | 733 (75.6) | 744 (76.8) | |
Black | 381 (19.7) | 188 (19.4) | 193 (19.9) | |
Asian | 54 (2.8) | 33 (3.4) | 21 (2.2) | |
Hispanic | N (Col %) | 0.648 | |||
Non-Hispanic | 1,858 (95.9) | 931 (96.1) | 927 (95.7) | |
Hispanic | 80 (4.1) | 38 (3.9) | 42 (4.3) | |
Cancer stage | N (Col %) | 0.743 | |||
In situ | 41 (2.1) | 22 (2.3) | 19 (2.0) | |
Localized | 628 (32.4) | 301 (31.1) | 327 (33.7) | |
Regional | 614 (31.7) | 317 (32.7) | 297 (30.7) | |
Distant | 579 (29.9) | 291 (30.0) | 288 (29.7) | |
Unstaged | 76 (3.9) | 38 (3.9) | 38 (3.9) | |
Marital status | N (Col %) | 0.446 | |||
Single | 400 (21.5) | 204 (22) | 196 (21) | |
Married/domestic partner | 1,065 (57.2) | 536 (57.9) | 529 (56.6) | |
Separated/Divorced/Widowed | 396 (21.3) | 186 (20.1) | 210 (22.5) | |
Payer/Insurance | N (Col %) | 0.622 | |||
Not insured | 144 (8.1) | 67 (7.6) | 77 (8.7) | |
Private insurance | 1,065 (60) | 545 (61.4) | 520 (58.6) | |
Medicaid/Medicare | 532 (30) | 259 (29.2) | 273 (30.7) | |
Military/VA/Public health services | 34 (1.9) | 16 (1.8) | 18 (2.0) | |
Rural/Urban commuting area (RUCA 2010) | N (Col %) | 0.23 | |||
Urban commuting area | 1,577 (85.9) | 797 (86.9) | 780 (85.0) | |
Not an urban commuting area | 258 (14.1) | 120 (13.1) | 138 (15.0) | |
Urban/Rural indicator (URIC 2010) | N (Col %) | 0.763 | |||
All urban | 893 (48.7) | 446 (48.6) | 447 (48.7) | |
Mostly urban | 512 (27.9) | 256 (27.9) | 256 (27.9) | |
Mostly rural | 251 (13.7) | 131 (14.3) | 120 (13.1) | |
All rural | 179 (9.8) | 84 (9.2) | 95 (10.3) |
aThe P-value is calculated by either parametric (ANOVA, χ2) or nonparametric (Kruskal–Wallis, Fisher’s exact) test, where appropriate, based on the normality test of data distribution and the sample size.
bA nonparametric test (Kruskal–Wallis or Fisher’s exact test) is applied.
Primary outcomes: survivor and relative reach
Compared to the Standard Care arm (Table 2), survivors in the YFC arm were significantly less likely to use the assigned access code (20.8% vs. 15.2%; P = 0.001), to consent to participate in the study (19% vs. 13.4%; P < 0.001), and to create a website login (18% vs. 13%; P = 0.001).
Primary outcomes . | Arm . | P-valuea . | |||
---|---|---|---|---|---|
Survivor reach . | Effect sizeb . | Total number of eligible survivors N = 1,938 (100%) . | Standard Care n = 969 (50%) . | YFC n = 969 (50%) . | . |
Access coded used | N (Col %) | 0.074 | 0.001 | |||
No | 1,589 (82.0) | 767 (79.2) | 822 (84.8) | ||
Yes | 349 (18.0) | 202 (20.8) | 147 (15.2) | ||
Study consented | N (Col %) | 0.076 | <0.001 | |||
No | 1,624 (83.8) | 785 (81.0) | 839 (86.6) | ||
Yes | 314 (16.2) | 184 (19.0) | 130 (13.4) | ||
Login created | N (Col %) | 0.068 | 0.003 | |||
No | 1,638 (84.5) | 795 (82.0) | 843 (87.0) | ||
Yes | 300 (15.5) | 174 (18.0) | 126 (13.0) |
Primary outcomes . | Arm . | P-valuea . | |||
---|---|---|---|---|---|
Survivor reach . | Effect sizeb . | Total number of eligible survivors N = 1,938 (100%) . | Standard Care n = 969 (50%) . | YFC n = 969 (50%) . | . |
Access coded used | N (Col %) | 0.074 | 0.001 | |||
No | 1,589 (82.0) | 767 (79.2) | 822 (84.8) | ||
Yes | 349 (18.0) | 202 (20.8) | 147 (15.2) | ||
Study consented | N (Col %) | 0.076 | <0.001 | |||
No | 1,624 (83.8) | 785 (81.0) | 839 (86.6) | ||
Yes | 314 (16.2) | 184 (19.0) | 130 (13.4) | ||
Login created | N (Col %) | 0.068 | 0.003 | |||
No | 1,638 (84.5) | 795 (82.0) | 843 (87.0) | ||
Yes | 300 (15.5) | 174 (18.0) | 126 (13.0) |
Primary outcomes . | Arm . | P-valuea . | |||
---|---|---|---|---|---|
Relative reach . | Effect sizeb . | Total number of relatives listed by survivors N = 393 (100%) . | Standard Care n = 241 (61.3%) . | YFC n = 152 (38.7%) . | . |
Access coded used |N (Col %) | 0.276 | <0.001c | |||
No | 372 (94.7) | 240 (99.6) | 132 (86.8) | ||
Yes | 21 (5.3) | 1 (0.4) | 20 (13.2) | ||
Study consented | N (Col %) | 0.268 | <0.001c | |||
No | 373 (94.9) | 240 (99.6) | 133 (87.5) | ||
Yes | 20 (5.1) | 1 (0.4) | 19 (12.5) | ||
Login created | N (Col %) | 0.284 | <0.001c | |||
No | 374 (95.2) | 241 (100) | 133 (87.5) | ||
Yes | 19 (4.8) | 0 (0.0) | 19 (12.5) |
Primary outcomes . | Arm . | P-valuea . | |||
---|---|---|---|---|---|
Relative reach . | Effect sizeb . | Total number of relatives listed by survivors N = 393 (100%) . | Standard Care n = 241 (61.3%) . | YFC n = 152 (38.7%) . | . |
Access coded used |N (Col %) | 0.276 | <0.001c | |||
No | 372 (94.7) | 240 (99.6) | 132 (86.8) | ||
Yes | 21 (5.3) | 1 (0.4) | 20 (13.2) | ||
Study consented | N (Col %) | 0.268 | <0.001c | |||
No | 373 (94.9) | 240 (99.6) | 133 (87.5) | ||
Yes | 20 (5.1) | 1 (0.4) | 19 (12.5) | ||
Login created | N (Col %) | 0.284 | <0.001c | |||
No | 374 (95.2) | 241 (100) | 133 (87.5) | ||
Yes | 19 (4.8) | 0 (0.0) | 19 (12.5) |
aThe P-value is calculated by either parametric (χ2) or nonparametric (Fisher’s exact) test, where appropriate, based on the sample size.
Effect size: a value of 0.1 is considered a small effect, 0.3 a medium effect, and 0.5 a large effect.
cA nonparametric test (Fisher’s exact test) is applied.
Relatives in the YFC arm were significantly more likely than those in the Standard Care arm to use the assigned access code (13.2% vs. 0.4%), to give consent to participate (12.5% vs. 0.4%), and to create a website login (12.5% vs. 0%). All P-values were <0.001.
Secondary outcomes: relative contact option Selection by survivors
Survivors assigned to the Standard Care arm only had the option to personally contact their at-risk relatives. Survivors who created a website login (n = 174) generated 241 unique access codes for their relatives (i.e., relatives listed).
In the YFC arm, survivors who established logins (n = 126) listed 152 relatives. Among these, survivors had the opportunity to select different relative contact options and showed varied preferences: 78 relatives (51.3%) were designated for “Self-contact,” 34 relatives (22.4%) for “Study team contact,” 24 relatives (15.8%) for “Delayed” contact, and survivors did not choose any contact option for 16 relatives (10.5%).
Among relatives listed by survivors, the YFC intervention yielded a 61.8% download rate for family letters (94 downloads for 152 relatives listed), which was significantly greater than the 31.5% rate (76 downloads for 241 relatives listed) observed in the Standard Care arm (P < 0.001).
Secondary outcomes: uptake of genetic counseling by survivors and relatives
Among survivors, we observed no significant arm differences in genetic counseling uptake: 15 (1.5%) survivors in the YFC arm scheduled and completed a telegenetic counseling session compared with 9 (0.9%) survivors in the Standard Care arm (P = 0.218). For relatives, uptake of genetic counseling was significantly higher in the YFC arm with 9 (5.9%) relatives scheduling and 8 (5.3%) completing a telegenetic counseling session, versus no relatives scheduling a counseling in the Standard Care arm (P < 0.001).
Factors associated with survivor website access code use
We tested whether our demographic factors were associated with survivor website access code use (Table 3). Univariate logistic regression analysis for each demographic variable (race, time since diagnosis, living a rural versus urban commuting area, and current age) revealed that survivors were significantly less likely to access the study website if they identified with a minority group (Black, Asian, or other) than if they identified as White (OR = 0.74; 95% CI, 0.55–0.98; P = 0.038), if their time of diagnosis was more distal (OR = 0.96; 95% CI, 0.93–0.99; P = 0.007), and if they were older (OR = 0.99; 95% CI, 0.98–1.00; P = 0.004). In multivariable logistic regression analysis, we found that minority race, longer time since diagnosis, and older age were all significantly associated with a decreased likelihood that the survivor had used the website access code, after controlling for arm assignment. Finally, neither survivors’ race, urban/rural residence, time since diagnosis, or age significantly interacted with study arm to predict survivor logging on to the website (all Ps > 0.01).
. | Univariate regressionAccess code used = Yes . | Multivariable regressionAccess code used = Yes . | ||||
---|---|---|---|---|---|---|
Covariate . | Level . | n . | Odds ratio (95% CI) . | ORP-value . | Odds ratio (95% CI) . | ORP-value . |
Race | Other | 443 | 0.74 (0.55–0.98) | 0.038 | 0.68 (0.50–0.91) | 0.010 |
White | 1,392 | — | — | — | — | |
Rural/Urban commuting area (RUCA 2010) | Urban commuting area | 1,577 | 1.38 (0.95–2.00) | 0.088 | 1.36 (0.94–1.98) | 0.107 |
Not an urban commuting area | 258 | — | — | — | — | |
Time since diagnosis | 1,835 | 0.96 (0.93–0.99) | 0.007 | 0.96 (0.92–0.99) | 0.015 | |
Current age | 1,835 | 0.99 (0.98–1.00) | 0.004 | 0.99 (0.98–1.00) | 0.018 | |
Arm | Usual care | 917 | 1.47 (1.17–1.86) | 0.001 | 1.47 (1.16–1.88) | 0.002 |
YFC | 918 | — | — | — | — |
. | Univariate regressionAccess code used = Yes . | Multivariable regressionAccess code used = Yes . | ||||
---|---|---|---|---|---|---|
Covariate . | Level . | n . | Odds ratio (95% CI) . | ORP-value . | Odds ratio (95% CI) . | ORP-value . |
Race | Other | 443 | 0.74 (0.55–0.98) | 0.038 | 0.68 (0.50–0.91) | 0.010 |
White | 1,392 | — | — | — | — | |
Rural/Urban commuting area (RUCA 2010) | Urban commuting area | 1,577 | 1.38 (0.95–2.00) | 0.088 | 1.36 (0.94–1.98) | 0.107 |
Not an urban commuting area | 258 | — | — | — | — | |
Time since diagnosis | 1,835 | 0.96 (0.93–0.99) | 0.007 | 0.96 (0.92–0.99) | 0.015 | |
Current age | 1,835 | 0.99 (0.98–1.00) | 0.004 | 0.99 (0.98–1.00) | 0.018 | |
Arm | Usual care | 917 | 1.47 (1.17–1.86) | 0.001 | 1.47 (1.16–1.88) | 0.002 |
YFC | 918 | — | — | — | — |
*Number of observations in the original data set = 1,938. Number of observations used = 1,835.
Discussion
Our study did not support our hypotheses and showed that the YFC intervention was not associated with increased reach of survivors. The Standard Care arm significantly outperformed the YFC arm in reaching more survivors based on all measures of reach. Survivors in the Standard Care arm listed more relatives (n = 241) than survivors in the YFC arm (n = 152). This trend, however, reversed when examining our primary objective to increase relative reach using a population-based approach. The download rate of family letters was higher in the YFC arm yielded (61.8%) than in the in the Standard Care arm (31.5%). Furthermore, website engagement was exclusively limited to relatives in the YFC arm, with no participation from survivors in the Standard Care arm. Thus, while the YFC intervention was less effective in engaging survivors, it demonstrated greater efficacy than Standard Care in engaging at-risk relatives.
Consistent with the current standard of care approach, “self-contact” was preferred most by survivors in the YFC arm (51.3%). Nevertheless, our study revealed the need to support survivors’ relational autonomy in contacting relatives as the remaining half of participants expressed a preference for contact options other than “self-contact.” Notably, survivors who selected “self-contact” for one relative tended to select the same option for other listed relatives and the same was true for the “study contact” option.
Although the number of relatives in the YFC arm who used the website access code was limited (n = 20), we explored if there was any trend to support our relational autonomy approach of offering a menu of contact options. Subgroup analysis revealed that relatives were more likely to use the access code if contact was initiated by the study team (32.4%, 11 out of 34) than by the survivor themselves (11.5%, 9 out of 78; P = 0.008). Our population-based study was not able to granularly explore how survivors make judgments about conveying inherited risk and this warrants further exploration.
In interpreting these findings, it is important to acknowledge that overall reach of both study arms was quite low (20.8% in Standard Care arm vs. 15.2% in the YFC arm) and significantly below our projected target reach of 45%. Due to low recruitment, we did not have sufficient statistical power to examine the uptake of genetic counseling as a primary outcome. Despite the high levels of computer (93.2%) and internet (86.2%) coverage in the state of Georgia based on census data, the GCR follow-up with participating and nonparticipating survivors indicated that the older members of our target audience may have lacked skills in using the internet. This finding is critical for future efforts aimed at assessing the digital literacy of older cancer survivors and devising strategies to enhance the accessibility of web-based tools.
Our study deliberately focused on survivors diagnosed as far back as 2005, aiming to include all identifiable ovarian cancer survivors in the state and to specifically include survivors diagnosed over a decade ago, an underrepresented group in recent genomic research. Research suggests that a cancer diagnosis can act as a “teachable moment” for health behavior change (34–37). Previous studies with cancer survivors suggest that duration since a cancer diagnosis is negatively associated with engagement in preventive behavior (e.g., consuming fruits and vegetables, quitting smoking; ref. 34). We too may have missed the teachable moment. The optimal timing for family risk communication between the emotionally demanding time of diagnosis/treatment and the point at which survivors and their families prioritize putting the experience behind them remains unclear (35). Offering a flexible portal for survivors to use when they are ready to do communicate with family members is worth considering.
The GCR collects limited medical history information. Thus, we do not have information about the extent to which survivors and relatives already had genetic counseling and testing. If they had previously received genetic services, they may have perceived website access as less valuable. Alternatively, other health professionals or through other online resources may have advised survivors and relatives to seek genetic counseling or information. If so, our findings may underestimate the actual level of family communication that occurred in the context of this intervention. Additionally, a discrepancy between the listed and actual numbers of at-risk relatives on the YFC website may exist, as our qualitative interviews indicated variations in adherence to website instructions. Finally, the generalizability of our results is potentially limited because only a small proportion of the survivors we contacted chose to participate in the study.
Although our approach has limitations, it lays the groundwork for developing population-based approaches that have the potential for broadening family communication about inherited cancer risk. We showed the feasibility of stakeholder engagement in developing the intervention (26). We successfully partnered with a state-wide cancer registry and a genetic counseling training program, demonstrating that it is feasible to implement a low-cost intervention with potential for reaching the entire target population. Future work is needed to further refine these approaches to optimize their effectiveness and reach for genetic risk screening.
Authors’ Disclosures
No disclosures were reported.
Authors’ Contributions
Y. Guan: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. C.M. McBride: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–eview and editing. J. Zhao: Data curation, formal analysis, investigation, visualization, methodology, project administration, writing–review and editing. R.D. Pentz: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. C. Escoffery: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. Y. Liu: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, project administration, writing–review and editing. Y. Cao: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. W. An: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. J.A. Shepperd: Conceptualization, formal analysis, supervision, validation, investigation, visualization, methodology, writing–review and editing. K.C. Ward: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, project administration, writing–review and editing.
Acknowledgments
We gratefully thank our citizen science collaborators for their important contributions in their community data collection outreach efforts: Jess BeCraft, Susan Bossert, Jan Byrne, Kelly Cannova, Jade Gibson, Phyllis Gilbert, Nancy Hicks, Aryn Kinney, Cindy McKinnon Deurloo, Regina Parker, Stacy Saravo, Marilyn Slodki, Summer Southern, and Kamilah Staggers. We acknowledge Peachtree Solutions for programming the YFC website (www.yourfamilyconnects.org). We express our gratitude to Drs. Cecelia Bellcross and Jane L. Meisel for her invaluable contributions to the study’s conceptualization, investigation, and methodology. The collection of cancer incidence data in Georgia was supported by contract HHSN261201800003I, Task Order HHSN26100001 from the NCI and cooperative agreement 6NU58DP006352-05-01 from the CDC. Research reported in this publication was supported by the NCI of the NIH under award number U01CA24058. C.M. McBride had been awarded this grant. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.