Racial differences in family health history knowledge of type 2 diabetes: exploring the role of interpersonal mechanisms

Racial differences in family health history knowledge of type 2 diabetes: exploring the role of... Abstract Collecting complete and accurate family health history is critical to preventing type 2 diabetes. Whether there are any racial difference in family health history knowledge of type 2 diabetes and whether such differences are related to interpersonal mechanisms remain unclear. We seek to identify the interpersonal mechanisms that give rise to discrepancies in family health history knowledge of type 2 diabetes in families of different racial backgrounds. We analyze informant-dyad consensus with respect to shared family history of type 2 diabetes in 127 informants of 45 families in the greater Cincinnati area (white: 28 families, 78 informants; black/African-American: 17 families, 49 informants). We first document a difference in informant-dyad consensus by race and then test whether this difference can be explained by interpersonal ties, particularly health communication. Compared with their white counterparts, dyads in families of black/African-American background are more likely to have an uneven distribution of knowledge, with one informant knowing and the other not knowing his/her family health history. The racial difference is explained by dyads in families of black/African-American background having fewer reciprocal health communication ties. While associated with informant-dyad consensus, education, kinship ties, and closeness ties do not account for the observed racial difference. Activating health communication is a key to improving family health history knowledge, especially in families of black/African-American background. Researchers and clinicians should leverage communication ties in the family network for better collection and utilization of family health history in preventive services. Implications Practice: Standards of clinical training and practice in primary care should compel providers to encourage family health communication for more frequent and complete collection of family health history by individual patients. Policy: Public health campaigns and intervention programs aimed to improve family health history knowledge among racial minorities must be supported by targeted efforts to build effective health communication channels within families. Research: Future research should further leverage network data to explore the interpersonal mechanisms underlying the familial context of health-related knowledge and behaviors. INTRODUCTION Given that the onset and complications from many common, complex diseases can be prevented or delayed, accurate and timely risk assessment is critical to implementing surveillance and lifestyle management strategies in primary care. It is widely accepted that complex diseases are multifactorial, influenced by genetic, socioenvironmental, and lifestyle risk factors [1, 2]. Family history is a robust indicator that captures the joint effect of inherited genetic susceptibility and common environment/behaviors in one’s family [3]. These notions have led to the expanding use of family health history by clinicians to assess in individual patients their risk for developing complex diseases, such as heart disease [4–6] and type 2 diabetes [7, 8]. Even a simple self- reported family health history can be a very efficient and relatively accurate tool for tailoring preventive regimens in at-risk populations [9–11] and for aiding the interpretation of other genetic and genomic information [12]. Thus, there is continued interest in developing computerized platforms [13–16], educational tools [17, 18], and public health campaigns [15, 19] to encourage more proactive collection of family health history. It is anticipated that these approaches will facilitate the gathering of more accurate data for use in clinical practice and ultimately improve health care and population health outcomes [12, 20]. Efforts to improve the collection and use of family health history in the clinic tend to be patient-centric, largely seeking to ease the time and cognitive burden associated with reporting this information. Yet the influence of family systems should also be an important consideration [20, 21]. The process of collecting family health history entails direct and proactive communication among family members [22, 23]. Over time, such communication ties become structural characteristics of the family network. Indeed, individuals’ health-related knowledge, beliefs, and behaviors can be influenced by social interactions in the family network [24–27]. This paper focuses on two aspects of the family network that are salient to family health history knowledge: health communication through interpersonal ties and the family’s racial background. The latter is associated with various sociocultural factors that provide the context for interpersonal mechanisms to occur. Specifically, we focus on family health history knowledge of type 2 diabetes, due to its public health significance [28] and strong association with lineage [7–10]. We analyze this knowledge at the dyadic-level, investigating if health communication is an underlying mechanism that gives rise to discrepancies between family members with respect to common biological relatives’ diagnoses in families of different racial backgrounds. We believe that such an approach is important and timely for two reasons. First, although national efforts encourage individuals to obtain their family health history [14, 15], there is limited research that provides mechanistic insights, guiding more targeted interventions for more proactive collection and more frequent update of this information. Prior studies have conducted comprehensive assessments of family health history knowledge in nationally representative samples [29] and at-risk minority or underserved populations [18, 23, 30, 31]. Since they are primarily based on analysis of individuals rather than dyads comprising families, these studies do not tap into the distribution of knowledge in the family system. There is evidence suggesting that sharing health information is more likely to occur in close, trustworthy, and supportive relationships [23, 32], but little work has been done to systematically link health communication ties to dyadic-level discrepancies in family health history knowledge between family members. Such work can serve as the basis for interventions that leverage interpersonal relationships, letting information flow from more knowledgeable members to others. Second, there is an urgent need for effective interventions to reduce disease burden and promote health in at-risk minority populations. In particular, African-Americans are disproportionately affected by type 2 diabetes and its complications, posing a serious public health threat and health equity challenge [33]. Gathering accurate knowledge of family health history has important implications for quality health care delivery in this population, including more targeted lifestyle management and screening recommendations for prevention and early detection. Only a few studies have examined racial difference in family health history knowledge and the findings are mixed. Some studies have shown a racial disparity, with African-Americans less likely to document their family health history in writing [34] and having limited understanding of how this information is linked to disease [35]. In contrast, others have found no racial difference in knowledge level [29] or the patterning of sharing [32]. Prior studies suggest that less frequent and active collection of family history among African-Americans may be related to education [36], lack of support [32], and fear of interpersonal conflict [23], pointing to the need to consider the broader social context influencing family health history knowledge. Currently, we do not know whether families from varying racial backgrounds differ in dyad-level patterns of family health history knowledge and if so, whether this difference is related to any interpersonal process in the family system. Such an understanding is nevertheless essential to the design and implementation of public health interventions among racial minorities. This paper presents a dyadic-level analysis of family health history knowledge, focusing on consensus/discrepancies in informant dyads’ reports of common biological relatives’ type 2 diabetes diagnoses. We examine whether there is a racial difference in family health history knowledge and explore if interpersonal mechanisms account for any observed racial difference, by evaluating a mediation effect of health communication ties. We discuss the results in the context of improving disease prevention and potentially reducing racial health disparities with family network-based approaches. METHODS Study design Targeting families at risk for type 2 diabetes in the greater Cincinnati area, we recruited multi-informant families of different racial backgrounds from an existing pharmacogenetics study comprised of individuals with a known type 2 diabetes diagnosis and through community advertisement. Each participant was interviewed individually by interviewers specializing in genetic counseling (master’s level) and provided a $25 gift card as an incentive. The interviewer and the interviewee were not racially matched. The Institutional Review Boards at the National Human Genome Research Institute and the University of Cincinnati approved the study protocols. Written and verbal consent was obtained prior to participation. Enrollment Individuals were eligible to enroll if they were 18 or older and if any of the following criteria was met: (a) the participant was diagnosed with type 2 diabetes; (b) the participant had at least one first- or second-degree relative diagnosed with type 2 diabetes; (c) the participant’s spouse/partner was diagnosed with type 2 diabetes (Fig. 1). Enrolled primary participants (n = 70) were asked to refer biological relatives (i.e., secondary participants) to the study. Secondary participants were also asked to refer biological relatives (i.e., tertiary participants). Eighty-six secondary and tertiary participants were subsequently enrolled, generating an initial sample of 156 individuals from 70 families. Fig 1 View largeDownload slide Enrollment and analytic sample. Fig 1 View largeDownload slide Enrollment and analytic sample. Analytic sample We excluded from the initial sample five participants due to missing data and then 24 single-informant families. Of the remaining 127 participants from 45 multi-informant families (two to five informants per family), 39% self-identified as black or African-American. With the exception of one biracial informant (self-identified as white and black/African-American), there was no racial variation within families in our sample. Thus, our final analytic sample consisted of 78 white informants from 28 families and 49 informants of black/African-American background from 17 families. Pedigree construction Each informant independently enumerated their first- and second-degree biological relatives and then indicated, for each relative, whether he/she had been diagnosed with type 2 diabetes (Fig. 2). We linked diagnoses reported by informants within families and evaluated informant-dyad consensus with respect to each common biological relative. There were 75 informant dyads in white families who provided 623 informant-dyad comparisons, and 52 dyads in black/African-American families who provided 348 comparisons for final analysis. Fig 2 View largeDownload slide Example of three-informant family (three dyads) with 15 common biological relatives, providing up to 37 informant-dyad comparison (e.g., III3 and IV4 reporting on II5). Circle/square correspondents to female/male family member. Informants can be common biological relatives. Fig 2 View largeDownload slide Example of three-informant family (three dyads) with 15 common biological relatives, providing up to 37 informant-dyad comparison (e.g., III3 and IV4 reporting on II5). Circle/square correspondents to female/male family member. Informants can be common biological relatives. Measures We used a structured questionnaire to collect the following: (a) individual demographics (age, gender, race, and socioeconomic status), health history (height, weight, and disease diagnoses), use of heath care services, and health-related behaviors (physical activity, diet, and tobacco and alcohol use); (2) family networks (health communication, cohesion, and conflict); and (3) family health history of type 2 diabetes and comorbid conditions (heart disease, high cholesterol, and high blood pressure) in first- and second-degree relatives. Below, we describe measures of study variables. Dependent variable Response categories to each relative’s type 2 diabetes diagnosis were “Yes,” “No,” and “Don’t Know.” We constructed a measure of informant-dyad comparison with four categories: (a) agree—both responded “Yes” or “No”; (b) disagree—one responded “Yes” and the other responded “No”; (c) one do not know—one responded “Yes” or “No” and the other responded “Don’t Know”; and (d) both responded “Don’t Know”. Family-level predictors We used the participants’ self-reported race to code the family’s racial background. Since there was no racial variation within families in our sample, we coded the family’s racial background as black/African-American if all informants identified themselves as black or African-American. Similarly, the family’s racial background was coded as white if all informants self-identified as white. The family with the biracial informant was coded as black/African-American and included in the analysis because the other two informants self-identified as black/African-American. Excluding this biracial informant would result in a sizable reduction (5%) in the number of informant-dyad comparisons in the black/African-American sample. Family size was the total number of family members enumerated by all informants in family health history reports. We controlled for both family size and the number of informants per family in multivariate analysis. Individual-level predictors We measured informants’ education with three dichotomized variables—high school (GED included) or less (reference group), more than high school (associate degree, some college, technical/vocational school), and college or above (bachelor’s or post-graduate degree). If an informant had a body mass index of 30 or greater based on his/her self-reported weight and height, we coded his/her weight status as obese (= 1, else = 0). Each informant’s age was measured in years and gender was measured with a dichotomized variable, female (= 1, else = 0). Age and gender were not included in multivariate analysis because age was collinear with kinship and our sample lacked gender diversity. Dyadic-level predictors We measured kinship tie between informants with three dichotomized variables: parent–child (= 1, else = 0; reference group), sibling–sibling (= 1, else = 0), and other kinship (= 1, else = 0; including aunt/uncle–niece/nephew, grandparent–grandchild, and cousin–cousin ties). We did not have sufficient statistical power to further distinguish the effect of each kinship in the “other kinship” category. In family network assessment, informants enumerated names of their family members with whom they had an interpersonal tie. For each informant dyad, if both nominated each other and indicated that he/she felt close to the other informant, a closeness tie was coded as reciprocal (= 1, else = 0). If only one indicated a closeness tie and the other did not, it was coded as asymmetrical (= 1, else = 0). A closeness tie was coded null (= 1, else = 0; reference group) if neither informant indicated that they felt close to each other. Similarly, informants indicated with whom they discussed their health and general health discussion tie was measured with three dichotomized variables—reciprocal (= 1, else = 0), asymmetrical (= 1, else = 0), and null (= 1, else = 0; reference group). Analysis plan First, we described family, individual, and dyadic attributes. In bivariate analysis, we cross-tabulated informant-dyad consensus in family health history by family’s racial background. Second, we estimated multinomial logistic regression models to examine the associations between informant-dyad consensus in family health history and the predictors. Our main hypothesis was that interpersonal mechanisms give rise to the racial patterning in family health history knowledge. Accordingly, we estimated a series of nested models to evaluate if accounting for family, individual, and dyadic attributes could explain any observed racial difference, with a focus on the role of health communication ties. The Huber–White sandwich estimator was used to adjust for clustering of dyadic comparisons within families [37]. RESULTS Sample characteristics Table 1 presents family, individual, and dyadic attributes by race. The average family size was about 30 and there was no racial difference in family size or the number of informants per family. Informants in white families were older than those in families of black/African-American background. The vast majority of the informants were women (73% in white and 82% in black/African-American families). About a third of the informants in white families had a high school education or less, a third had more than a high school education, and the remaining had a bachelor’s or post-graduate degree. A greater proportion of informants in families of black/African-American background had more than a high school education (48%), but there was no significant difference in education by race. More than half of the informants were obese, possibly because our study targeted families at risk for type 2 diabetes. Most of the kinship ties between informants were parent–child or sibling–sibling. Less than a third of the informant dyads reported having reciprocal or asymmetrical closeness ties. Although there was no racial difference in kinship or closeness ties, dyads in black/African-American families had fewer reciprocal (12%) and asymmetrical (8%) health discussion ties than did those in white families (19% and 20%, respectively; p = .001). Table 1 Family-, individual-, and dyadic-level attributes by race White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 Means and standard deviations (in parentheses, wherever applicable) reported. Racial differences in attributes tested using chi-squared tests or t tests wherever appropriate. nf number of families; ni number of individuals; nd number of dyads. View Large Table 1 Family-, individual-, and dyadic-level attributes by race White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 Means and standard deviations (in parentheses, wherever applicable) reported. Racial differences in attributes tested using chi-squared tests or t tests wherever appropriate. nf number of families; ni number of individuals; nd number of dyads. View Large Informant-dyad consensus in family health history by race Figure 3 describes informant-dyad consensus in family health history of type 2 diabetes by race. There was little racial difference for two categories—“agree” and “disagree.” More than 60% of the informant-dyad comparisons were in agreement and 5% of them were in disagreement. In white families, one-fifth of the comparisons were “one ‘don’t know’,” meaning that one informant was more knowledgeable about family health history than the other informant. The percent of “one ‘don’t know’” was even higher, at 26%, for families of black/African-American background. White families had more “both ‘don’t know’” (13%) comparisons than did black/African-American families (8%). The racial difference in informant-dyad consensus in family health history was statistically significant in bivariate analysis (χ2(3) = 8.85, p = .031). We decomposed the overall χ2 statistic and the largest contribution to racial difference, 33%, was due to more “one ‘don’t know’” informant-dyad comparisons in families of black/African-American background. Fig 3 View largeDownload slide Informant-dyad comparisons of family health history (type 2 diabetes) reports by race. (+) Has disease; (−) no disease. Fig 3 View largeDownload slide Informant-dyad comparisons of family health history (type 2 diabetes) reports by race. (+) Has disease; (−) no disease. Predictors of informant-dyad consensus and the role of interpersonal ties Table 2 presents results from multinomial logistic regression models, where we examine predictors associated with informant-dyad consensus in family health history. The reference category of the dependent variable was “agree.” Controlling for family size and the number of informants (effects not statistically significant; results not shown), black/African-American families were more likely to have informant-dyad comparisons where one informant was more knowledgeable than the other (relative risk ratio [RRR thereafter] = 1.44, p < .05). There was no racial difference for the other two categories in the dependent variable—“disagree” and “both ‘don’t know’,” comparing with the reference category, “agree.” Table 2 Multinomial logistic regression models estimating informant-dyad comparisons of family health history (type 2 diabetes) reports Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** aAll models control for family size and number of informants per family (results omitted). bReference category for dependent variable is “agree”; intercepts omitted. cRelative risk ratios reported; clustering adjusted for using robust standard errors (omitted). dIncludes second-degree relatives and cousins. *p < .05; **p < .01; ***p < .001. View Large Table 2 Multinomial logistic regression models estimating informant-dyad comparisons of family health history (type 2 diabetes) reports Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** aAll models control for family size and number of informants per family (results omitted). bReference category for dependent variable is “agree”; intercepts omitted. cRelative risk ratios reported; clustering adjusted for using robust standard errors (omitted). dIncludes second-degree relatives and cousins. *p < .05; **p < .01; ***p < .001. View Large In subsequent models, we entered individual- and dyadic-level predictors to examine their associations with informant-dyad consensus and how including them affected the observed racial difference. Comparing across the nested models, we found that the racial difference remained significant in Model 2 (RRR = 1.43, p < .05) where we controlled for individual attributes, and in Model 3 (RRR = 1.52, p <.05) where we controlled for individual attributes, kinship ties, and closeness ties. The racial difference was eliminated only after accounting for general health discussion ties in Model 4 (RRR = 1.28, p = 0.19). A reciprocal health discussion tie had a direct effect on informant-dyad consensus in family health history, lowering the chance of one informant reporting “don’t know” (RRR = 0.34, p < .01), and fully mediated the association between the family’s racial background and informant-dyad comparisons. Although they did not explain the observed racial difference, education, weight status, kinship ties, and closeness ties were all significant predictors of informant-dyad comparisons of family health history. As shown in Model 4, if the younger informant had college or more education, it was associated with a greater chance for the informant-dyad comparison to be “one ‘don’t know’” (RRR = 2.87, p < .001). The younger informant being obese increased the chance of the comparison being “one ‘don’t know’” (RRR = 1.81, p < .01). The older informant’s education and weight status, however, had no effect. Compared with parent-child dyads, sibling–sibling dyads were more likely to have a pattern of “one ‘don’t know’ (RRR = 1.63, p < .05).” The effect of other kinship was not distinguishable from that of parent–child dyads (RRR = 1.04, p = .90). A reciprocal closeness tie was associated with a lower chance of the comparison being “one ‘don’t know’” in Model 3 (RRR = 0.48, p < .001), but this effect disappeared (RRR = 0.96, p = .91) after we accounted for health communication ties in the final model. Finally, although there was no racial difference in the pattern of “both ‘don’t know’,” the models suggested that higher levels of education were associated with a greater chance of both informants in the dyads not knowing their shared family health history. Compared with parent–child dyads, sibling–sibling dyads were significantly more likely to report “both ‘don’t know’.” A reciprocal closeness tie was associated with lower chance of both informants responding “don’t know,” but a general health discussion tie was not. DISCUSSION In this study, we have sought to address gaps in family health history research by exploring the interpersonal mechanisms associated with dyadic-level discrepancies in knowledge in families of different racial backgrounds. It is well known that knowledge of one’s family health history can guide risk-specific prevention, potentially reducing the burden caused by common, complex diseases in at-risk populations [9–11], but how to develop and implement tools and services most effectively is a difficult task [18, 23, 33, 38]. Most notably, our findings highlight the promise of leveraging health communication ties as a more effective intervention approach in diverse familial contexts. We observe a racial difference in family health history knowledge, with families of black/African-American background being less knowledgeable compared with white families, a finding consistent with the work of Thompson et al. [34], Ashida et al. [35], and Hughes Halbert et al. [36]. Our results provide a more nuanced understanding of this difference: the inequality is a dyadic-level phenomenon indicated by the uneven distribution of knowledge in families of black/African-American background, rather than an individual-level phenomenon where familial risk information is simply lacking in racial minorities regardless of family dynamics. This means that the standards of training and practice in primary care should compel providers to recognize that family health history knowledge is relational and to encourage communication between family members for more frequent, complete, and up-to-date collection of this information. Previous studies typically measure the lack of family health history knowledge with rates of “don’t know” responses in groups of independently sampled individuals [29, 30]. We have demonstrated that these “don’t know” responses can be categorized into two dyad-level patterns that arise from different mechanisms and likely require different interventions. The uneven distribution of family health history knowledge in dyads is a function of lack of reciprocal health communication. Consistent with analysis of Thompson et al.[23], our results suggest that families of black/African-American background have fewer reciprocal health discussion ties than do white families, which in turn leads to lack of consensus in shared familial risk. Some members in these families already have the knowledge that is relevant and salient to others and they may withhold information due to a lack of supportive ties [32] or fear of interpersonal conflict [23]. For policymakers, this means that public health campaigns aiming to improve family health history knowledge among racial minorities must be supported by targeted efforts to build effective communication channels which will allow the information to flow. In both racial groups, there are instances where both informants in the dyads are unaware of their family health history. Such a pattern is related to kinship and closeness ties, but not health communication ties. Sibling dyads are more likely to have “both ‘don’t know’” comparisons than parent–child dyads, likely because parents are more knowledgeable about disease diagnosis in older generations than children [30, 39]. A reciprocal closeness tie also reduces the likelihood of a “both ‘don’t know’” dyadic response. The findings suggest that encouraging parents to take on the role as family historians and leveraging close interpersonal relationships can improve family health history knowledge. However, interventions capitalizing solely on kinship and closeness ties may be less effective in reducing racial disparities which, as we have found, result from a different interpersonal mechanism—health communication. Our results can aid the interpretation of the mixed findings from prior research about racial differences in family health history knowledge [34–36]. Consistent with others [23, 32], we found that family health history knowledge was closely tied to health communication in families regardless of their racial background. Therefore, this knowledge is likely to be family-specific, depending on characteristics of the relational mechanisms in each family network. Between-family heterogeneity can make it difficult to detect the effect of race—or other social determinants—when the analysis is conducted in independently sampled individuals. Future research should focus on the network level, analyzing factors that involve micro-interactions in a social system, in order to fully capture social influences on family health history knowledge. Often, education and health literacy are thought to play a role in determining health-related knowledge [36, 40, 41]. Our analysis yields some novel findings regarding education, suggesting the need to reconsider the effects of individual attributes in the context of dyad-level dynamics. We found that higher levels of education of the younger informant was associated with increased chance of uneven knowledge distribution between informant dyads and more missing data in family health history, when the older informant’s education was controlled for. This somewhat contradicts the notion that education is positively correlated with health-related knowledge and behaviors [42]. One possible explanation is that individuals with more education are also more likely to live far away from their family of origin. Distance can put limits on contact, which makes it more difficult for these individuals to obtain and update their family health history in a timely manner, even if there are communication channels between parents and children and between siblings. It is important for researchers to interpret the effect of education in the context of contact, social interaction, and the family network. Future research should collect information about the frequency of contact to test this hypothesis. Such work will help elucidate how individual characteristics operate through kinship ties to influence health-related knowledge and behaviors. This study has the advantage of permitting a dyadic examination of how interpersonal mechanisms affect family health history knowledge in families of difference races. At the same time, the study has several limitations, primarily a result of its limited, nonrepresentative sample. Like most family-based recruitment efforts, our sample is relatively small, consists of predominantly women, and lacks diversity in kinship ties, even though we did not limit the criteria of inclusion to parent–child and full sibling dyads. Men are not as actively involved in risk information gathering and dissemination as are women [38, 43, 44]. Half-siblings confer risk and may have a different communication and exchange dynamics within the family. With current data, however, we are unable to disentangle the role of gender or fully explore the nuances of kinship ties. Also, this is a cross-sectional study focusing on dyadic exchanges and using the family’s racial background as a proxy for a broader cultural context. Larger-scale, prospective research employing more representative samples to examine other interpersonal processes and familial cultural influences, especially in multiracial families, is urgently needed for moving this line of inquiry forward. Finally, our study focuses on family health history knowledge of type 2 diabetes. Although the interpersonal mechanisms we explored are broadly conceived and are potentially applicable across disease contexts, future research should validate the findings in other disease contexts with varied genetic etiologies. These limitations notwithstanding, our study represents an important step forward in family health history research. Current tools and programs have yet to fully leverage communication ties in families to facilitate better family health history collection, despite intrafamily communication being a dyadic process. We believe that interventions focusing on health communication provide opportunities to engage family members in collecting, documenting, and consolidating their shared family health history which can lead to improved risk assessment, disease prevention, and health promotion efforts and potentially reduce racial health disparities. Acknowledgements This research was supported by the National Human Genome Research Institute’s Intramural Research Program (ZIAHG200335 to L.M.K.) and a National Institute of Diabetes and Digestive and Kidney Diseases grant (K18DK095473 to M.F.M.). Author contributions: J.L., C.S.M., M.F.M. and L.M.K. conceived the study. J.L. designed and performed the analysis, interpreted the results and wrote the manuscript. M.F.M. supervised data collection and provided critical revisions of the manuscript. C.S.M. helped in data interpretation. L.M.K. oversaw the project and revised drafts of the manuscript. Compliance with Ethical Standards Conflict of interest: The authors have no conflict of interest. Primary Data: The authors have full control of all primary data and agree to allow the journal to review data if requested. The findings reported in the current manuscript have not been previously presented or published and the manuscript is not under consideration for publication elsewhere. There has, however, been a publication generated from Ref. 21. PMID: 28062275. Ethical Approval: The IRBs at the National Human Genome Research Institute and the University of Cincinnati approved the study. The study was conducted per the Ethical Standards for the Protection of Human Participants in Research. This research did not involve any animals and as such the welfare of animals was not compromised during the implementation of this project. Informed Consent: Written and verbal consent were obtained prior to study participation. References 1. Rees J . Complex disease and the new clinical sciences . Science . 2002 ; 296 ( 5568 ): 698 – 700 . Google Scholar CrossRef Search ADS PubMed 2. Sing CF , Stengård JH , Kardia SL . Genes, environment, and cardiovascular disease . Arterioscler Thromb Vasc Biol . 2003 ; 23 ( 7 ): 1190 – 1196 . Google Scholar CrossRef Search ADS PubMed 3. Scheuner MT , Wang SJ , Raffel LJ , Larabell SK , Rotter JI . Family history: a comprehensive genetic risk assessment method for the chronic conditions of adulthood . Am j Med Genet . 1997 ; 71 ( 3 ): 315 – 324 . Google Scholar CrossRef Search ADS PubMed 4. Qureshi N , Armstrong S , Dhiman P , et al. ; ADDFAM (Added Value of Family History in CVD Risk Assessment) Study Group . Effect of adding systematic family history enquiry to cardiovascular disease risk assessment in primary care: a matched-pair, cluster randomized trial . Ann Intern Med . 2012 ; 156 ( 4 ): 253 – 262 . Google Scholar CrossRef Search ADS PubMed 5. Williams RR , Hunt SC , Heiss G , et al. Usefulness of cardiovascular family history data for population-based preventive medicine and medical research (the Health Family Tree Study and the NHLBI Family Heart Study) . Am j Cardiol . 2001 ; 87 ( 2 ): 129 – 135 . Google Scholar CrossRef Search ADS PubMed 6. Woodward M , Brindle P , Tunstall-Pedoe H ; SIGN group on risk estimation . Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC) . Heart . 2007 ; 93 ( 2 ): 172 – 176 . Google Scholar CrossRef Search ADS PubMed 7. Annis AM , Caulder MS , Cook ML , Duquette D . Family history, diabetes, and other demographic and risk factors among participants of the National Health and Nutrition Examination Survey 1999-2002 . Prev Chronic Dis . 2005 ; 2 ( 2 ): A19 . Google Scholar PubMed 8. Heikes KE , Eddy DM , Arondekar B , Schlessinger L . Diabetes risk calculator: a simple tool for detecting undiagnosed diabetes and pre-diabetes . Diabetes Care . 2008 ; 31 ( 5 ): 1040 – 1045 . Google Scholar CrossRef Search ADS PubMed 9. Hariri S , Yoon PW , Qureshi N , Valdez R , Scheuner MT , Khoury MJ . Family history of type 2 diabetes: a population-based screening tool for prevention ? Genet Med . 2006 ; 8 ( 2 ): 102 – 108 . Google Scholar CrossRef Search ADS PubMed 10. Harrison TA , Hindorff LA , Kim H , et al. Family history of diabetes as a potential public health tool . Am j Prev Med . 2003 ; 24 ( 2 ): 152 – 159 . Google Scholar CrossRef Search ADS PubMed 11. Yoon PW , Scheuner MT , Peterson-Oehlke KL , Gwinn M , Faucett A , Khoury MJ . Can family history be used as a tool for public health and preventive medicine ? Genet Med . 2002 ; 4 ( 4 ): 304 – 310 . Google Scholar CrossRef Search ADS PubMed 12. Guttmacher AE , Collins FS , Carmona RH . The family history–more important than ever . n Engl j Med . 2004 ; 351 ( 22 ): 2333 – 2336 . Google Scholar CrossRef Search ADS PubMed 13. Cohn WF , Ropka ME , Pelletier SL , et al. Health Heritage© a web-based tool for the collection and assessment of family health history: initial user experience and analytic validity . Public Health Genomics . 2010 ; 13 ( 7-8 ): 477 – 491 . Google Scholar CrossRef Search ADS PubMed 14. O’Neill SM , Rubinstein WS , Wang C , et al. ; Family Healthware Impact Trial group . Familial risk for common diseases in primary care: the Family Healthware Impact Trial . Am j Prev Med . 2009 ; 36 ( 6 ): 506 – 514 . Google Scholar CrossRef Search ADS PubMed 15. Owens KM , Marvin ML , Gelehrter TD , Ruffin MT 4th , Uhlmann WR . Clinical use of the Surgeon General’s “My Family Health Portrait” (MFHP) tool: opinions of future health care providers . j Genet Couns . 2011 ; 20 ( 5 ): 510 – 525 . Google Scholar CrossRef Search ADS PubMed 16. Sweet K , Sturm AC , Rettig A , McElroy J , Agnese D . Clinically relevant lessons from Family HealthLink: a cancer and coronary heart disease familial risk assessment tool . Genet Med . 2015 ; 17 ( 6 ): 493 – 500 . Google Scholar CrossRef Search ADS PubMed 17. Koehly LM , Morris BA , Skapinsky K , Goergen A , Ludden A . Evaluation of the Families SHARE workbook: an educational tool outlining disease risk and healthy guidelines to reduce risk of heart disease, diabetes, breast cancer and colorectal cancer . bmc Public Health . 2015 ; 15 : 1120 . Google Scholar CrossRef Search ADS PubMed 18. Petruccio C , Mills Shaw KR , Boughman J , et al. Healthy choices through family history: a community approach to family history awareness . Community Genet . 2008 ; 11 ( 6 ): 343 – 351 . Google Scholar PubMed 19. Johnson J , Giles RT , Larsen L , Ware J , Adams T , Hunt SC . Utah’s Family High Risk Program: bridging the gap between genomics and public health . Prev Chronic Dis . 2005 ; 2 ( 2 ): A24 . Google Scholar PubMed 20. Welch BM , Dere W , Schiffman JD . Family health history: the case for better tools . Jama . 2015 ; 313 ( 17 ): 1711 – 1712 . Google Scholar CrossRef Search ADS PubMed 21. Lin J , Marcum CS , Myers MF , Koehly LM . Put the family back in family health history: a multiple-informant approach . Am j Prev Med . 2017 ; 52 ( 5 ): 640 – 644 . Google Scholar CrossRef Search ADS PubMed 22. Black L , McClellan KA , Avard D , Knoppers BM . Intrafamilial disclosure of risk for hereditary breast and ovarian cancer: points to consider . j Community Genet . 2013 ; 4 ( 2 ): 203 – 214 . Google Scholar CrossRef Search ADS PubMed 23. Thompson T , Seo J , Griffith J , Baxter M , James A , Kaphingst KA . The context of collecting family health history: examining definitions of family and family communication about health among African American women . j Health Commun . 2015 ; 20 ( 4 ): 416 – 423 . Google Scholar CrossRef Search ADS PubMed 24. Broderick CB. Understanding Family Process: Basics of Family Systems Theory . Thousand Oaks, CA : Sage ; 1993 . 25. Rusbult CE , Van Lange PA . Interdependence, interaction, and relationships . Annu Rev Psychol . 2003 ; 54 : 351 – 375 . Google Scholar CrossRef Search ADS PubMed 26. Berkman LF , Glass T . Social integration, social networks, social support and health . In: Berkman LF , Kawachi I , eds. Social Epidemiology . New York, NY : Oxford University Press ; 2000 : 137 – 173 . 27. Umberson D , Crosnoe R , Reczek C . Social relationships and health behavior across life course . Annu Rev Sociol . 2010 ; 36 : 139 – 157 . Google Scholar CrossRef Search ADS PubMed 28. Center for Disease Control and Prevention . National diabetes statistics report , 2017 . Available at https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf. Accessed 1 July 2017. 29. Centers for Disease Control and Prevention (CDC) . Awareness of family health history as a risk factor for disease--United States, 2004 . MMWR Morb Mort Wkly Rep . 2004 ; 53 ( 44 ): 1044 – 1047 . 30. Goergen AF , Ashida S , Skapinsky K , de Heer HD , Wilkinson AV , Koehly LM . What you don’t know: improving family health history knowledge among multigenerational families of Mexican origin . Public Health Genomics . 2016 ; 19 ( 2 ): 93 – 101 . Google Scholar CrossRef Search ADS PubMed 31. Wallace JP , Baugh C , Cornett S , et al. A family history demonstration project among women in an urban Appalachian community . Prog Community Health Partnersh . 2009 ; 3 ( 2 ): 155 – 163 . Google Scholar CrossRef Search ADS PubMed 32. Ashida S , Kaphingst KA , Goodman M , Schafer EJ . Family health history communication networks of older adults: importance of social relationships and disease perceptions . Health Educ Behav . 2013 ; 40 ( 5 ): 612 – 619 . Google Scholar CrossRef Search ADS PubMed 33. Spanakis EK , Golden SH . Race/ethnic difference in diabetes and diabetic complications . Curr Diab Rep . 2013 ; 13 ( 6 ): 814 – 823 . Google Scholar CrossRef Search ADS PubMed 34. Thompson T , Seo J , Griffith J , Baxter M , James A , Kaphingst KA . “You don’t have to keep everything on paper”: African American women’s use of family health history tools . j Community Genet . 2013 ; 4 ( 2 ): 251 – 261 . Google Scholar CrossRef Search ADS PubMed 35. Ashida S , Goodman MS , Stafford J , Lachance C , Kaphingst KA . Perceived familiarity with and importance of family health history among a medically underserved population . j Community Genet . 2012 ; 3 ( 4 ): 285 – 295 . Google Scholar CrossRef Search ADS PubMed 36. Hughes Halbert C , Welch B , Lynch C , et al. Social determinants of family health history collection . j Community Genet . 2016 ; 7 ( 1 ): 57 – 64 . Google Scholar CrossRef Search ADS PubMed 37. Williams RL . A note on robust variance estimation for cluster-correlated data . Biometrics . 2000 ; 56 ( 2 ): 645 – 646 . Google Scholar CrossRef Search ADS PubMed 38. Welch BM , O’Connell N , Schiffman JD . 10 years later: assessing the impact of public health efforts on the collection of family health history . Am j Med Genet a . 2015 ; 167A ( 9 ): 2026 – 2033 . Google Scholar CrossRef Search ADS PubMed 39. Case DO . Collection of family health histories: the link between genealogy and public health . J Assoc Inf Sci Technol . 2008 ; 59 ( 4 ): 2313 – 2319 . 40. Kickbusch IS . Health literacy: addressing the health and education divide . Health Promot Int . 2001 ; 16 ( 3 ): 289 – 297 . Google Scholar CrossRef Search ADS PubMed 41. Nutbeam D . The evolving concept of health literacy . Soc Sci Med . 2008 ; 67 ( 12 ): 2072 – 2078 . Google Scholar CrossRef Search ADS PubMed 42. Lantz PM , House JS , Lepkowski JM , Williams DR , Mero RP , Chen J . Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults . Jama . 1998 ; 279 ( 21 ): 1703 – 1708 . Google Scholar CrossRef Search ADS PubMed 43. Koehly LM , Peters JA , Kenen R , et al. Characteristics of health information gatherers, disseminators, and blockers within families at risk of hereditary cancer: implications for family health communication interventions . Am j Public Health . 2009 ; 99 ( 12 ): 2203 – 2209 . Google Scholar CrossRef Search ADS PubMed 44. Koehly LM , Peters JA , Kuhn N , et al. Sisters in hereditary breast and ovarian cancer families: communal coping, social integration, and psychological well-being . Psychooncology . 2008 ; 17 ( 8 ): 812 – 821 . Google Scholar CrossRef Search ADS PubMed Published by Oxford University Press on behalf of the Society of Behavioral Medicine 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Translational Behavioral Medicine Oxford University Press

Racial differences in family health history knowledge of type 2 diabetes: exploring the role of interpersonal mechanisms

Loading next page...
 
/lp/ou_press/racial-differences-in-family-health-history-knowledge-of-type-2-iNsobF3rmm
Copyright
Published by Oxford University Press on behalf of the Society of Behavioral Medicine 2018.
ISSN
1869-6716
eISSN
1613-9860
D.O.I.
10.1093/tbm/ibx062
Publisher site
See Article on Publisher Site

Abstract

Abstract Collecting complete and accurate family health history is critical to preventing type 2 diabetes. Whether there are any racial difference in family health history knowledge of type 2 diabetes and whether such differences are related to interpersonal mechanisms remain unclear. We seek to identify the interpersonal mechanisms that give rise to discrepancies in family health history knowledge of type 2 diabetes in families of different racial backgrounds. We analyze informant-dyad consensus with respect to shared family history of type 2 diabetes in 127 informants of 45 families in the greater Cincinnati area (white: 28 families, 78 informants; black/African-American: 17 families, 49 informants). We first document a difference in informant-dyad consensus by race and then test whether this difference can be explained by interpersonal ties, particularly health communication. Compared with their white counterparts, dyads in families of black/African-American background are more likely to have an uneven distribution of knowledge, with one informant knowing and the other not knowing his/her family health history. The racial difference is explained by dyads in families of black/African-American background having fewer reciprocal health communication ties. While associated with informant-dyad consensus, education, kinship ties, and closeness ties do not account for the observed racial difference. Activating health communication is a key to improving family health history knowledge, especially in families of black/African-American background. Researchers and clinicians should leverage communication ties in the family network for better collection and utilization of family health history in preventive services. Implications Practice: Standards of clinical training and practice in primary care should compel providers to encourage family health communication for more frequent and complete collection of family health history by individual patients. Policy: Public health campaigns and intervention programs aimed to improve family health history knowledge among racial minorities must be supported by targeted efforts to build effective health communication channels within families. Research: Future research should further leverage network data to explore the interpersonal mechanisms underlying the familial context of health-related knowledge and behaviors. INTRODUCTION Given that the onset and complications from many common, complex diseases can be prevented or delayed, accurate and timely risk assessment is critical to implementing surveillance and lifestyle management strategies in primary care. It is widely accepted that complex diseases are multifactorial, influenced by genetic, socioenvironmental, and lifestyle risk factors [1, 2]. Family history is a robust indicator that captures the joint effect of inherited genetic susceptibility and common environment/behaviors in one’s family [3]. These notions have led to the expanding use of family health history by clinicians to assess in individual patients their risk for developing complex diseases, such as heart disease [4–6] and type 2 diabetes [7, 8]. Even a simple self- reported family health history can be a very efficient and relatively accurate tool for tailoring preventive regimens in at-risk populations [9–11] and for aiding the interpretation of other genetic and genomic information [12]. Thus, there is continued interest in developing computerized platforms [13–16], educational tools [17, 18], and public health campaigns [15, 19] to encourage more proactive collection of family health history. It is anticipated that these approaches will facilitate the gathering of more accurate data for use in clinical practice and ultimately improve health care and population health outcomes [12, 20]. Efforts to improve the collection and use of family health history in the clinic tend to be patient-centric, largely seeking to ease the time and cognitive burden associated with reporting this information. Yet the influence of family systems should also be an important consideration [20, 21]. The process of collecting family health history entails direct and proactive communication among family members [22, 23]. Over time, such communication ties become structural characteristics of the family network. Indeed, individuals’ health-related knowledge, beliefs, and behaviors can be influenced by social interactions in the family network [24–27]. This paper focuses on two aspects of the family network that are salient to family health history knowledge: health communication through interpersonal ties and the family’s racial background. The latter is associated with various sociocultural factors that provide the context for interpersonal mechanisms to occur. Specifically, we focus on family health history knowledge of type 2 diabetes, due to its public health significance [28] and strong association with lineage [7–10]. We analyze this knowledge at the dyadic-level, investigating if health communication is an underlying mechanism that gives rise to discrepancies between family members with respect to common biological relatives’ diagnoses in families of different racial backgrounds. We believe that such an approach is important and timely for two reasons. First, although national efforts encourage individuals to obtain their family health history [14, 15], there is limited research that provides mechanistic insights, guiding more targeted interventions for more proactive collection and more frequent update of this information. Prior studies have conducted comprehensive assessments of family health history knowledge in nationally representative samples [29] and at-risk minority or underserved populations [18, 23, 30, 31]. Since they are primarily based on analysis of individuals rather than dyads comprising families, these studies do not tap into the distribution of knowledge in the family system. There is evidence suggesting that sharing health information is more likely to occur in close, trustworthy, and supportive relationships [23, 32], but little work has been done to systematically link health communication ties to dyadic-level discrepancies in family health history knowledge between family members. Such work can serve as the basis for interventions that leverage interpersonal relationships, letting information flow from more knowledgeable members to others. Second, there is an urgent need for effective interventions to reduce disease burden and promote health in at-risk minority populations. In particular, African-Americans are disproportionately affected by type 2 diabetes and its complications, posing a serious public health threat and health equity challenge [33]. Gathering accurate knowledge of family health history has important implications for quality health care delivery in this population, including more targeted lifestyle management and screening recommendations for prevention and early detection. Only a few studies have examined racial difference in family health history knowledge and the findings are mixed. Some studies have shown a racial disparity, with African-Americans less likely to document their family health history in writing [34] and having limited understanding of how this information is linked to disease [35]. In contrast, others have found no racial difference in knowledge level [29] or the patterning of sharing [32]. Prior studies suggest that less frequent and active collection of family history among African-Americans may be related to education [36], lack of support [32], and fear of interpersonal conflict [23], pointing to the need to consider the broader social context influencing family health history knowledge. Currently, we do not know whether families from varying racial backgrounds differ in dyad-level patterns of family health history knowledge and if so, whether this difference is related to any interpersonal process in the family system. Such an understanding is nevertheless essential to the design and implementation of public health interventions among racial minorities. This paper presents a dyadic-level analysis of family health history knowledge, focusing on consensus/discrepancies in informant dyads’ reports of common biological relatives’ type 2 diabetes diagnoses. We examine whether there is a racial difference in family health history knowledge and explore if interpersonal mechanisms account for any observed racial difference, by evaluating a mediation effect of health communication ties. We discuss the results in the context of improving disease prevention and potentially reducing racial health disparities with family network-based approaches. METHODS Study design Targeting families at risk for type 2 diabetes in the greater Cincinnati area, we recruited multi-informant families of different racial backgrounds from an existing pharmacogenetics study comprised of individuals with a known type 2 diabetes diagnosis and through community advertisement. Each participant was interviewed individually by interviewers specializing in genetic counseling (master’s level) and provided a $25 gift card as an incentive. The interviewer and the interviewee were not racially matched. The Institutional Review Boards at the National Human Genome Research Institute and the University of Cincinnati approved the study protocols. Written and verbal consent was obtained prior to participation. Enrollment Individuals were eligible to enroll if they were 18 or older and if any of the following criteria was met: (a) the participant was diagnosed with type 2 diabetes; (b) the participant had at least one first- or second-degree relative diagnosed with type 2 diabetes; (c) the participant’s spouse/partner was diagnosed with type 2 diabetes (Fig. 1). Enrolled primary participants (n = 70) were asked to refer biological relatives (i.e., secondary participants) to the study. Secondary participants were also asked to refer biological relatives (i.e., tertiary participants). Eighty-six secondary and tertiary participants were subsequently enrolled, generating an initial sample of 156 individuals from 70 families. Fig 1 View largeDownload slide Enrollment and analytic sample. Fig 1 View largeDownload slide Enrollment and analytic sample. Analytic sample We excluded from the initial sample five participants due to missing data and then 24 single-informant families. Of the remaining 127 participants from 45 multi-informant families (two to five informants per family), 39% self-identified as black or African-American. With the exception of one biracial informant (self-identified as white and black/African-American), there was no racial variation within families in our sample. Thus, our final analytic sample consisted of 78 white informants from 28 families and 49 informants of black/African-American background from 17 families. Pedigree construction Each informant independently enumerated their first- and second-degree biological relatives and then indicated, for each relative, whether he/she had been diagnosed with type 2 diabetes (Fig. 2). We linked diagnoses reported by informants within families and evaluated informant-dyad consensus with respect to each common biological relative. There were 75 informant dyads in white families who provided 623 informant-dyad comparisons, and 52 dyads in black/African-American families who provided 348 comparisons for final analysis. Fig 2 View largeDownload slide Example of three-informant family (three dyads) with 15 common biological relatives, providing up to 37 informant-dyad comparison (e.g., III3 and IV4 reporting on II5). Circle/square correspondents to female/male family member. Informants can be common biological relatives. Fig 2 View largeDownload slide Example of three-informant family (three dyads) with 15 common biological relatives, providing up to 37 informant-dyad comparison (e.g., III3 and IV4 reporting on II5). Circle/square correspondents to female/male family member. Informants can be common biological relatives. Measures We used a structured questionnaire to collect the following: (a) individual demographics (age, gender, race, and socioeconomic status), health history (height, weight, and disease diagnoses), use of heath care services, and health-related behaviors (physical activity, diet, and tobacco and alcohol use); (2) family networks (health communication, cohesion, and conflict); and (3) family health history of type 2 diabetes and comorbid conditions (heart disease, high cholesterol, and high blood pressure) in first- and second-degree relatives. Below, we describe measures of study variables. Dependent variable Response categories to each relative’s type 2 diabetes diagnosis were “Yes,” “No,” and “Don’t Know.” We constructed a measure of informant-dyad comparison with four categories: (a) agree—both responded “Yes” or “No”; (b) disagree—one responded “Yes” and the other responded “No”; (c) one do not know—one responded “Yes” or “No” and the other responded “Don’t Know”; and (d) both responded “Don’t Know”. Family-level predictors We used the participants’ self-reported race to code the family’s racial background. Since there was no racial variation within families in our sample, we coded the family’s racial background as black/African-American if all informants identified themselves as black or African-American. Similarly, the family’s racial background was coded as white if all informants self-identified as white. The family with the biracial informant was coded as black/African-American and included in the analysis because the other two informants self-identified as black/African-American. Excluding this biracial informant would result in a sizable reduction (5%) in the number of informant-dyad comparisons in the black/African-American sample. Family size was the total number of family members enumerated by all informants in family health history reports. We controlled for both family size and the number of informants per family in multivariate analysis. Individual-level predictors We measured informants’ education with three dichotomized variables—high school (GED included) or less (reference group), more than high school (associate degree, some college, technical/vocational school), and college or above (bachelor’s or post-graduate degree). If an informant had a body mass index of 30 or greater based on his/her self-reported weight and height, we coded his/her weight status as obese (= 1, else = 0). Each informant’s age was measured in years and gender was measured with a dichotomized variable, female (= 1, else = 0). Age and gender were not included in multivariate analysis because age was collinear with kinship and our sample lacked gender diversity. Dyadic-level predictors We measured kinship tie between informants with three dichotomized variables: parent–child (= 1, else = 0; reference group), sibling–sibling (= 1, else = 0), and other kinship (= 1, else = 0; including aunt/uncle–niece/nephew, grandparent–grandchild, and cousin–cousin ties). We did not have sufficient statistical power to further distinguish the effect of each kinship in the “other kinship” category. In family network assessment, informants enumerated names of their family members with whom they had an interpersonal tie. For each informant dyad, if both nominated each other and indicated that he/she felt close to the other informant, a closeness tie was coded as reciprocal (= 1, else = 0). If only one indicated a closeness tie and the other did not, it was coded as asymmetrical (= 1, else = 0). A closeness tie was coded null (= 1, else = 0; reference group) if neither informant indicated that they felt close to each other. Similarly, informants indicated with whom they discussed their health and general health discussion tie was measured with three dichotomized variables—reciprocal (= 1, else = 0), asymmetrical (= 1, else = 0), and null (= 1, else = 0; reference group). Analysis plan First, we described family, individual, and dyadic attributes. In bivariate analysis, we cross-tabulated informant-dyad consensus in family health history by family’s racial background. Second, we estimated multinomial logistic regression models to examine the associations between informant-dyad consensus in family health history and the predictors. Our main hypothesis was that interpersonal mechanisms give rise to the racial patterning in family health history knowledge. Accordingly, we estimated a series of nested models to evaluate if accounting for family, individual, and dyadic attributes could explain any observed racial difference, with a focus on the role of health communication ties. The Huber–White sandwich estimator was used to adjust for clustering of dyadic comparisons within families [37]. RESULTS Sample characteristics Table 1 presents family, individual, and dyadic attributes by race. The average family size was about 30 and there was no racial difference in family size or the number of informants per family. Informants in white families were older than those in families of black/African-American background. The vast majority of the informants were women (73% in white and 82% in black/African-American families). About a third of the informants in white families had a high school education or less, a third had more than a high school education, and the remaining had a bachelor’s or post-graduate degree. A greater proportion of informants in families of black/African-American background had more than a high school education (48%), but there was no significant difference in education by race. More than half of the informants were obese, possibly because our study targeted families at risk for type 2 diabetes. Most of the kinship ties between informants were parent–child or sibling–sibling. Less than a third of the informant dyads reported having reciprocal or asymmetrical closeness ties. Although there was no racial difference in kinship or closeness ties, dyads in black/African-American families had fewer reciprocal (12%) and asymmetrical (8%) health discussion ties than did those in white families (19% and 20%, respectively; p = .001). Table 1 Family-, individual-, and dyadic-level attributes by race White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 Means and standard deviations (in parentheses, wherever applicable) reported. Racial differences in attributes tested using chi-squared tests or t tests wherever appropriate. nf number of families; ni number of individuals; nd number of dyads. View Large Table 1 Family-, individual-, and dyadic-level attributes by race White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 White Black/African-American p-values Family attributes (nf = 28) (nf = 17)  Family size 27.36 (8.24) 29.88 (9.67) .356  Number of informants per family 2.79 (0.69) 2.88 (0.86) .679 Individual attributes (ni = 78) (ni = 49)  Age 50.24 (16.37) 42.67 (15.10) .011  Female 73% 82% .269  Education .238   High school or less 0.31 0.27   Associate 0.33 0.48   College and above 0.36 0.25  Obese 0.54 0.63 .296 Dyadic attributes (nd = 75) (nd = 52)  Kinship ties .155   Parent–child 0.53 0.46   Sibling–sibling 0.33 0.27   Other 0.14 0.27  Closeness ties .600   Null 0.69 0.73   Asymmetrical 0.11 0.13   Reciprocal 0.20 0.13  General health discussion ties .001   Null 0.61 0.81   Asymmetrical 0.20 0.08   Reciprocal 0.19 0.12 Means and standard deviations (in parentheses, wherever applicable) reported. Racial differences in attributes tested using chi-squared tests or t tests wherever appropriate. nf number of families; ni number of individuals; nd number of dyads. View Large Informant-dyad consensus in family health history by race Figure 3 describes informant-dyad consensus in family health history of type 2 diabetes by race. There was little racial difference for two categories—“agree” and “disagree.” More than 60% of the informant-dyad comparisons were in agreement and 5% of them were in disagreement. In white families, one-fifth of the comparisons were “one ‘don’t know’,” meaning that one informant was more knowledgeable about family health history than the other informant. The percent of “one ‘don’t know’” was even higher, at 26%, for families of black/African-American background. White families had more “both ‘don’t know’” (13%) comparisons than did black/African-American families (8%). The racial difference in informant-dyad consensus in family health history was statistically significant in bivariate analysis (χ2(3) = 8.85, p = .031). We decomposed the overall χ2 statistic and the largest contribution to racial difference, 33%, was due to more “one ‘don’t know’” informant-dyad comparisons in families of black/African-American background. Fig 3 View largeDownload slide Informant-dyad comparisons of family health history (type 2 diabetes) reports by race. (+) Has disease; (−) no disease. Fig 3 View largeDownload slide Informant-dyad comparisons of family health history (type 2 diabetes) reports by race. (+) Has disease; (−) no disease. Predictors of informant-dyad consensus and the role of interpersonal ties Table 2 presents results from multinomial logistic regression models, where we examine predictors associated with informant-dyad consensus in family health history. The reference category of the dependent variable was “agree.” Controlling for family size and the number of informants (effects not statistically significant; results not shown), black/African-American families were more likely to have informant-dyad comparisons where one informant was more knowledgeable than the other (relative risk ratio [RRR thereafter] = 1.44, p < .05). There was no racial difference for the other two categories in the dependent variable—“disagree” and “both ‘don’t know’,” comparing with the reference category, “agree.” Table 2 Multinomial logistic regression models estimating informant-dyad comparisons of family health history (type 2 diabetes) reports Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** aAll models control for family size and number of informants per family (results omitted). bReference category for dependent variable is “agree”; intercepts omitted. cRelative risk ratios reported; clustering adjusted for using robust standard errors (omitted). dIncludes second-degree relatives and cousins. *p < .05; **p < .01; ***p < .001. View Large Table 2 Multinomial logistic regression models estimating informant-dyad comparisons of family health history (type 2 diabetes) reports Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** Model 1a Model 2 Model 3 Model 4 Disagreeb One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Disagree One “Don’t Know” Both “Don’t Know” Black/African- American 1.17c 1.44* 0.72 1.20 1.43* 0.77 1.43 1.52* 0.71 1.35 1.28 0.78 Younger informant  Education   ≤ High school (reference)   > High school 1.51 1.31 1.82 2.36* 1.80* 2.98** 2.37* 1.90* 2.92**   ≥ College 2.04 2.41*** 2.79*** 3.03* 2.92*** 3.95*** 2.99* 2.87*** 4.13**  Obese 1.11 2.02*** 1.43 1.32 1.80** 1.03 1.06 1.81** 1.04 Older informant  Education   ≤ High school (reference)   > High school 0.52 1.23 2.55*** 0.37* 1.03 1.86* 0.37 1.06 1.81*   ≥ College 0.49 0.92 1.15 0.30* 0.63 0.61 0.31* 0.68 0.58  Obese 1.15 0.67* 0.62* 1.32 0.83 0.96 0.78 0.99 Kinship ties  Parent–child (reference)  Sibling–sibling 2.12 1.91*** 4.28*** 2.03 1.63* 4.67***  Other kinshipd 1.61 1.15 1.38 1.57 1.04 1.46 Closeness  Null (reference)  Asymmetrical 0.20 0.73 1.00 0.26 1.25 0.72  Reciprocal 0.78 0.48*** 0.52* 1.01 0.96 0.38* Discuss health  Null (reference)  Asymmetrical 0.75 0.53 1.38  Reciprocal 0.74 0.34** 1.61 Wald χ2(df)Sig. 16.25(9) 96.83(27)*** 134.24(39)*** 143.64(45)*** aAll models control for family size and number of informants per family (results omitted). bReference category for dependent variable is “agree”; intercepts omitted. cRelative risk ratios reported; clustering adjusted for using robust standard errors (omitted). dIncludes second-degree relatives and cousins. *p < .05; **p < .01; ***p < .001. View Large In subsequent models, we entered individual- and dyadic-level predictors to examine their associations with informant-dyad consensus and how including them affected the observed racial difference. Comparing across the nested models, we found that the racial difference remained significant in Model 2 (RRR = 1.43, p < .05) where we controlled for individual attributes, and in Model 3 (RRR = 1.52, p <.05) where we controlled for individual attributes, kinship ties, and closeness ties. The racial difference was eliminated only after accounting for general health discussion ties in Model 4 (RRR = 1.28, p = 0.19). A reciprocal health discussion tie had a direct effect on informant-dyad consensus in family health history, lowering the chance of one informant reporting “don’t know” (RRR = 0.34, p < .01), and fully mediated the association between the family’s racial background and informant-dyad comparisons. Although they did not explain the observed racial difference, education, weight status, kinship ties, and closeness ties were all significant predictors of informant-dyad comparisons of family health history. As shown in Model 4, if the younger informant had college or more education, it was associated with a greater chance for the informant-dyad comparison to be “one ‘don’t know’” (RRR = 2.87, p < .001). The younger informant being obese increased the chance of the comparison being “one ‘don’t know’” (RRR = 1.81, p < .01). The older informant’s education and weight status, however, had no effect. Compared with parent-child dyads, sibling–sibling dyads were more likely to have a pattern of “one ‘don’t know’ (RRR = 1.63, p < .05).” The effect of other kinship was not distinguishable from that of parent–child dyads (RRR = 1.04, p = .90). A reciprocal closeness tie was associated with a lower chance of the comparison being “one ‘don’t know’” in Model 3 (RRR = 0.48, p < .001), but this effect disappeared (RRR = 0.96, p = .91) after we accounted for health communication ties in the final model. Finally, although there was no racial difference in the pattern of “both ‘don’t know’,” the models suggested that higher levels of education were associated with a greater chance of both informants in the dyads not knowing their shared family health history. Compared with parent–child dyads, sibling–sibling dyads were significantly more likely to report “both ‘don’t know’.” A reciprocal closeness tie was associated with lower chance of both informants responding “don’t know,” but a general health discussion tie was not. DISCUSSION In this study, we have sought to address gaps in family health history research by exploring the interpersonal mechanisms associated with dyadic-level discrepancies in knowledge in families of different racial backgrounds. It is well known that knowledge of one’s family health history can guide risk-specific prevention, potentially reducing the burden caused by common, complex diseases in at-risk populations [9–11], but how to develop and implement tools and services most effectively is a difficult task [18, 23, 33, 38]. Most notably, our findings highlight the promise of leveraging health communication ties as a more effective intervention approach in diverse familial contexts. We observe a racial difference in family health history knowledge, with families of black/African-American background being less knowledgeable compared with white families, a finding consistent with the work of Thompson et al. [34], Ashida et al. [35], and Hughes Halbert et al. [36]. Our results provide a more nuanced understanding of this difference: the inequality is a dyadic-level phenomenon indicated by the uneven distribution of knowledge in families of black/African-American background, rather than an individual-level phenomenon where familial risk information is simply lacking in racial minorities regardless of family dynamics. This means that the standards of training and practice in primary care should compel providers to recognize that family health history knowledge is relational and to encourage communication between family members for more frequent, complete, and up-to-date collection of this information. Previous studies typically measure the lack of family health history knowledge with rates of “don’t know” responses in groups of independently sampled individuals [29, 30]. We have demonstrated that these “don’t know” responses can be categorized into two dyad-level patterns that arise from different mechanisms and likely require different interventions. The uneven distribution of family health history knowledge in dyads is a function of lack of reciprocal health communication. Consistent with analysis of Thompson et al.[23], our results suggest that families of black/African-American background have fewer reciprocal health discussion ties than do white families, which in turn leads to lack of consensus in shared familial risk. Some members in these families already have the knowledge that is relevant and salient to others and they may withhold information due to a lack of supportive ties [32] or fear of interpersonal conflict [23]. For policymakers, this means that public health campaigns aiming to improve family health history knowledge among racial minorities must be supported by targeted efforts to build effective communication channels which will allow the information to flow. In both racial groups, there are instances where both informants in the dyads are unaware of their family health history. Such a pattern is related to kinship and closeness ties, but not health communication ties. Sibling dyads are more likely to have “both ‘don’t know’” comparisons than parent–child dyads, likely because parents are more knowledgeable about disease diagnosis in older generations than children [30, 39]. A reciprocal closeness tie also reduces the likelihood of a “both ‘don’t know’” dyadic response. The findings suggest that encouraging parents to take on the role as family historians and leveraging close interpersonal relationships can improve family health history knowledge. However, interventions capitalizing solely on kinship and closeness ties may be less effective in reducing racial disparities which, as we have found, result from a different interpersonal mechanism—health communication. Our results can aid the interpretation of the mixed findings from prior research about racial differences in family health history knowledge [34–36]. Consistent with others [23, 32], we found that family health history knowledge was closely tied to health communication in families regardless of their racial background. Therefore, this knowledge is likely to be family-specific, depending on characteristics of the relational mechanisms in each family network. Between-family heterogeneity can make it difficult to detect the effect of race—or other social determinants—when the analysis is conducted in independently sampled individuals. Future research should focus on the network level, analyzing factors that involve micro-interactions in a social system, in order to fully capture social influences on family health history knowledge. Often, education and health literacy are thought to play a role in determining health-related knowledge [36, 40, 41]. Our analysis yields some novel findings regarding education, suggesting the need to reconsider the effects of individual attributes in the context of dyad-level dynamics. We found that higher levels of education of the younger informant was associated with increased chance of uneven knowledge distribution between informant dyads and more missing data in family health history, when the older informant’s education was controlled for. This somewhat contradicts the notion that education is positively correlated with health-related knowledge and behaviors [42]. One possible explanation is that individuals with more education are also more likely to live far away from their family of origin. Distance can put limits on contact, which makes it more difficult for these individuals to obtain and update their family health history in a timely manner, even if there are communication channels between parents and children and between siblings. It is important for researchers to interpret the effect of education in the context of contact, social interaction, and the family network. Future research should collect information about the frequency of contact to test this hypothesis. Such work will help elucidate how individual characteristics operate through kinship ties to influence health-related knowledge and behaviors. This study has the advantage of permitting a dyadic examination of how interpersonal mechanisms affect family health history knowledge in families of difference races. At the same time, the study has several limitations, primarily a result of its limited, nonrepresentative sample. Like most family-based recruitment efforts, our sample is relatively small, consists of predominantly women, and lacks diversity in kinship ties, even though we did not limit the criteria of inclusion to parent–child and full sibling dyads. Men are not as actively involved in risk information gathering and dissemination as are women [38, 43, 44]. Half-siblings confer risk and may have a different communication and exchange dynamics within the family. With current data, however, we are unable to disentangle the role of gender or fully explore the nuances of kinship ties. Also, this is a cross-sectional study focusing on dyadic exchanges and using the family’s racial background as a proxy for a broader cultural context. Larger-scale, prospective research employing more representative samples to examine other interpersonal processes and familial cultural influences, especially in multiracial families, is urgently needed for moving this line of inquiry forward. Finally, our study focuses on family health history knowledge of type 2 diabetes. Although the interpersonal mechanisms we explored are broadly conceived and are potentially applicable across disease contexts, future research should validate the findings in other disease contexts with varied genetic etiologies. These limitations notwithstanding, our study represents an important step forward in family health history research. Current tools and programs have yet to fully leverage communication ties in families to facilitate better family health history collection, despite intrafamily communication being a dyadic process. We believe that interventions focusing on health communication provide opportunities to engage family members in collecting, documenting, and consolidating their shared family health history which can lead to improved risk assessment, disease prevention, and health promotion efforts and potentially reduce racial health disparities. Acknowledgements This research was supported by the National Human Genome Research Institute’s Intramural Research Program (ZIAHG200335 to L.M.K.) and a National Institute of Diabetes and Digestive and Kidney Diseases grant (K18DK095473 to M.F.M.). Author contributions: J.L., C.S.M., M.F.M. and L.M.K. conceived the study. J.L. designed and performed the analysis, interpreted the results and wrote the manuscript. M.F.M. supervised data collection and provided critical revisions of the manuscript. C.S.M. helped in data interpretation. L.M.K. oversaw the project and revised drafts of the manuscript. Compliance with Ethical Standards Conflict of interest: The authors have no conflict of interest. Primary Data: The authors have full control of all primary data and agree to allow the journal to review data if requested. The findings reported in the current manuscript have not been previously presented or published and the manuscript is not under consideration for publication elsewhere. There has, however, been a publication generated from Ref. 21. PMID: 28062275. Ethical Approval: The IRBs at the National Human Genome Research Institute and the University of Cincinnati approved the study. The study was conducted per the Ethical Standards for the Protection of Human Participants in Research. This research did not involve any animals and as such the welfare of animals was not compromised during the implementation of this project. Informed Consent: Written and verbal consent were obtained prior to study participation. References 1. Rees J . Complex disease and the new clinical sciences . Science . 2002 ; 296 ( 5568 ): 698 – 700 . Google Scholar CrossRef Search ADS PubMed 2. Sing CF , Stengård JH , Kardia SL . Genes, environment, and cardiovascular disease . Arterioscler Thromb Vasc Biol . 2003 ; 23 ( 7 ): 1190 – 1196 . Google Scholar CrossRef Search ADS PubMed 3. Scheuner MT , Wang SJ , Raffel LJ , Larabell SK , Rotter JI . Family history: a comprehensive genetic risk assessment method for the chronic conditions of adulthood . Am j Med Genet . 1997 ; 71 ( 3 ): 315 – 324 . Google Scholar CrossRef Search ADS PubMed 4. Qureshi N , Armstrong S , Dhiman P , et al. ; ADDFAM (Added Value of Family History in CVD Risk Assessment) Study Group . Effect of adding systematic family history enquiry to cardiovascular disease risk assessment in primary care: a matched-pair, cluster randomized trial . Ann Intern Med . 2012 ; 156 ( 4 ): 253 – 262 . Google Scholar CrossRef Search ADS PubMed 5. Williams RR , Hunt SC , Heiss G , et al. Usefulness of cardiovascular family history data for population-based preventive medicine and medical research (the Health Family Tree Study and the NHLBI Family Heart Study) . Am j Cardiol . 2001 ; 87 ( 2 ): 129 – 135 . Google Scholar CrossRef Search ADS PubMed 6. Woodward M , Brindle P , Tunstall-Pedoe H ; SIGN group on risk estimation . Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC) . Heart . 2007 ; 93 ( 2 ): 172 – 176 . Google Scholar CrossRef Search ADS PubMed 7. Annis AM , Caulder MS , Cook ML , Duquette D . Family history, diabetes, and other demographic and risk factors among participants of the National Health and Nutrition Examination Survey 1999-2002 . Prev Chronic Dis . 2005 ; 2 ( 2 ): A19 . Google Scholar PubMed 8. Heikes KE , Eddy DM , Arondekar B , Schlessinger L . Diabetes risk calculator: a simple tool for detecting undiagnosed diabetes and pre-diabetes . Diabetes Care . 2008 ; 31 ( 5 ): 1040 – 1045 . Google Scholar CrossRef Search ADS PubMed 9. Hariri S , Yoon PW , Qureshi N , Valdez R , Scheuner MT , Khoury MJ . Family history of type 2 diabetes: a population-based screening tool for prevention ? Genet Med . 2006 ; 8 ( 2 ): 102 – 108 . Google Scholar CrossRef Search ADS PubMed 10. Harrison TA , Hindorff LA , Kim H , et al. Family history of diabetes as a potential public health tool . Am j Prev Med . 2003 ; 24 ( 2 ): 152 – 159 . Google Scholar CrossRef Search ADS PubMed 11. Yoon PW , Scheuner MT , Peterson-Oehlke KL , Gwinn M , Faucett A , Khoury MJ . Can family history be used as a tool for public health and preventive medicine ? Genet Med . 2002 ; 4 ( 4 ): 304 – 310 . Google Scholar CrossRef Search ADS PubMed 12. Guttmacher AE , Collins FS , Carmona RH . The family history–more important than ever . n Engl j Med . 2004 ; 351 ( 22 ): 2333 – 2336 . Google Scholar CrossRef Search ADS PubMed 13. Cohn WF , Ropka ME , Pelletier SL , et al. Health Heritage© a web-based tool for the collection and assessment of family health history: initial user experience and analytic validity . Public Health Genomics . 2010 ; 13 ( 7-8 ): 477 – 491 . Google Scholar CrossRef Search ADS PubMed 14. O’Neill SM , Rubinstein WS , Wang C , et al. ; Family Healthware Impact Trial group . Familial risk for common diseases in primary care: the Family Healthware Impact Trial . Am j Prev Med . 2009 ; 36 ( 6 ): 506 – 514 . Google Scholar CrossRef Search ADS PubMed 15. Owens KM , Marvin ML , Gelehrter TD , Ruffin MT 4th , Uhlmann WR . Clinical use of the Surgeon General’s “My Family Health Portrait” (MFHP) tool: opinions of future health care providers . j Genet Couns . 2011 ; 20 ( 5 ): 510 – 525 . Google Scholar CrossRef Search ADS PubMed 16. Sweet K , Sturm AC , Rettig A , McElroy J , Agnese D . Clinically relevant lessons from Family HealthLink: a cancer and coronary heart disease familial risk assessment tool . Genet Med . 2015 ; 17 ( 6 ): 493 – 500 . Google Scholar CrossRef Search ADS PubMed 17. Koehly LM , Morris BA , Skapinsky K , Goergen A , Ludden A . Evaluation of the Families SHARE workbook: an educational tool outlining disease risk and healthy guidelines to reduce risk of heart disease, diabetes, breast cancer and colorectal cancer . bmc Public Health . 2015 ; 15 : 1120 . Google Scholar CrossRef Search ADS PubMed 18. Petruccio C , Mills Shaw KR , Boughman J , et al. Healthy choices through family history: a community approach to family history awareness . Community Genet . 2008 ; 11 ( 6 ): 343 – 351 . Google Scholar PubMed 19. Johnson J , Giles RT , Larsen L , Ware J , Adams T , Hunt SC . Utah’s Family High Risk Program: bridging the gap between genomics and public health . Prev Chronic Dis . 2005 ; 2 ( 2 ): A24 . Google Scholar PubMed 20. Welch BM , Dere W , Schiffman JD . Family health history: the case for better tools . Jama . 2015 ; 313 ( 17 ): 1711 – 1712 . Google Scholar CrossRef Search ADS PubMed 21. Lin J , Marcum CS , Myers MF , Koehly LM . Put the family back in family health history: a multiple-informant approach . Am j Prev Med . 2017 ; 52 ( 5 ): 640 – 644 . Google Scholar CrossRef Search ADS PubMed 22. Black L , McClellan KA , Avard D , Knoppers BM . Intrafamilial disclosure of risk for hereditary breast and ovarian cancer: points to consider . j Community Genet . 2013 ; 4 ( 2 ): 203 – 214 . Google Scholar CrossRef Search ADS PubMed 23. Thompson T , Seo J , Griffith J , Baxter M , James A , Kaphingst KA . The context of collecting family health history: examining definitions of family and family communication about health among African American women . j Health Commun . 2015 ; 20 ( 4 ): 416 – 423 . Google Scholar CrossRef Search ADS PubMed 24. Broderick CB. Understanding Family Process: Basics of Family Systems Theory . Thousand Oaks, CA : Sage ; 1993 . 25. Rusbult CE , Van Lange PA . Interdependence, interaction, and relationships . Annu Rev Psychol . 2003 ; 54 : 351 – 375 . Google Scholar CrossRef Search ADS PubMed 26. Berkman LF , Glass T . Social integration, social networks, social support and health . In: Berkman LF , Kawachi I , eds. Social Epidemiology . New York, NY : Oxford University Press ; 2000 : 137 – 173 . 27. Umberson D , Crosnoe R , Reczek C . Social relationships and health behavior across life course . Annu Rev Sociol . 2010 ; 36 : 139 – 157 . Google Scholar CrossRef Search ADS PubMed 28. Center for Disease Control and Prevention . National diabetes statistics report , 2017 . Available at https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf. Accessed 1 July 2017. 29. Centers for Disease Control and Prevention (CDC) . Awareness of family health history as a risk factor for disease--United States, 2004 . MMWR Morb Mort Wkly Rep . 2004 ; 53 ( 44 ): 1044 – 1047 . 30. Goergen AF , Ashida S , Skapinsky K , de Heer HD , Wilkinson AV , Koehly LM . What you don’t know: improving family health history knowledge among multigenerational families of Mexican origin . Public Health Genomics . 2016 ; 19 ( 2 ): 93 – 101 . Google Scholar CrossRef Search ADS PubMed 31. Wallace JP , Baugh C , Cornett S , et al. A family history demonstration project among women in an urban Appalachian community . Prog Community Health Partnersh . 2009 ; 3 ( 2 ): 155 – 163 . Google Scholar CrossRef Search ADS PubMed 32. Ashida S , Kaphingst KA , Goodman M , Schafer EJ . Family health history communication networks of older adults: importance of social relationships and disease perceptions . Health Educ Behav . 2013 ; 40 ( 5 ): 612 – 619 . Google Scholar CrossRef Search ADS PubMed 33. Spanakis EK , Golden SH . Race/ethnic difference in diabetes and diabetic complications . Curr Diab Rep . 2013 ; 13 ( 6 ): 814 – 823 . Google Scholar CrossRef Search ADS PubMed 34. Thompson T , Seo J , Griffith J , Baxter M , James A , Kaphingst KA . “You don’t have to keep everything on paper”: African American women’s use of family health history tools . j Community Genet . 2013 ; 4 ( 2 ): 251 – 261 . Google Scholar CrossRef Search ADS PubMed 35. Ashida S , Goodman MS , Stafford J , Lachance C , Kaphingst KA . Perceived familiarity with and importance of family health history among a medically underserved population . j Community Genet . 2012 ; 3 ( 4 ): 285 – 295 . Google Scholar CrossRef Search ADS PubMed 36. Hughes Halbert C , Welch B , Lynch C , et al. Social determinants of family health history collection . j Community Genet . 2016 ; 7 ( 1 ): 57 – 64 . Google Scholar CrossRef Search ADS PubMed 37. Williams RL . A note on robust variance estimation for cluster-correlated data . Biometrics . 2000 ; 56 ( 2 ): 645 – 646 . Google Scholar CrossRef Search ADS PubMed 38. Welch BM , O’Connell N , Schiffman JD . 10 years later: assessing the impact of public health efforts on the collection of family health history . Am j Med Genet a . 2015 ; 167A ( 9 ): 2026 – 2033 . Google Scholar CrossRef Search ADS PubMed 39. Case DO . Collection of family health histories: the link between genealogy and public health . J Assoc Inf Sci Technol . 2008 ; 59 ( 4 ): 2313 – 2319 . 40. Kickbusch IS . Health literacy: addressing the health and education divide . Health Promot Int . 2001 ; 16 ( 3 ): 289 – 297 . Google Scholar CrossRef Search ADS PubMed 41. Nutbeam D . The evolving concept of health literacy . Soc Sci Med . 2008 ; 67 ( 12 ): 2072 – 2078 . Google Scholar CrossRef Search ADS PubMed 42. Lantz PM , House JS , Lepkowski JM , Williams DR , Mero RP , Chen J . Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults . Jama . 1998 ; 279 ( 21 ): 1703 – 1708 . Google Scholar CrossRef Search ADS PubMed 43. Koehly LM , Peters JA , Kenen R , et al. Characteristics of health information gatherers, disseminators, and blockers within families at risk of hereditary cancer: implications for family health communication interventions . Am j Public Health . 2009 ; 99 ( 12 ): 2203 – 2209 . Google Scholar CrossRef Search ADS PubMed 44. Koehly LM , Peters JA , Kuhn N , et al. Sisters in hereditary breast and ovarian cancer families: communal coping, social integration, and psychological well-being . Psychooncology . 2008 ; 17 ( 8 ): 812 – 821 . Google Scholar CrossRef Search ADS PubMed Published by Oxford University Press on behalf of the Society of Behavioral Medicine 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US.

Journal

Translational Behavioral MedicineOxford University Press

Published: Jul 17, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off