Family History for Public Health and Preventive Medicine: Developing a Research Agenda
A Report for the
Office of Genomics and Disease Prevention
National Center for Environmental Health
Centers for Disease Control and Prevention
Prepared by Susan Baker Toal, MPH
August 19, 2002
Table of Contents
Executive Summary........................................................... 3
I. Family History and Chronic Disease............................... 6
II. Analytic Validity............................................................ 7
III. Clinical Validity............................................................ 14
IV. Clinical Utility................................................................18
V. Ethical, Legal, and Social Implications...........................23
VI. Next Steps.................................................................. 26
On May 1-2, 2002, the Office of Genomics and Disease Prevention (OGDP) in the National Center for Environmental Health (NCEH), Centers for Disease Control and Prevention, convened a workshop in Atlanta, Georgia, to discuss the potential of family history as a public health tool for improved disease prevention, and begin to develop a research agenda for evaluating the feasibility, validity, and utility of this approach. The workshop, entitled “Family History for Public Health and Preventive Medicine: Developing a Research Agenda,” involved 36 invited participants from a variety of backgrounds and perspectives.
An evaluation framework was used to structure the workshop presentations and to help identify gaps in knowledge about the validity and utility of family history information for disease prevention. The four components of the framework were analytic validity, clinical validity, clinical utility, and the ethical, legal and social implications of family history screening. Presentations were also made on four diseases as potential candidates for inclusion in a family history tool – coronary heart disease, type 2 diabetes, asthma, and colorectal cancer. The workshop participants discussed a number of criteria that could be used to select the diseases to include in a public health oriented family history tool and outlined the issues that could form the basis of a research agenda for evaluating the validity and utility of the tool.
By the end of the workshop, the participants had concluded that a family history tool for public health and preventive medicine should be: 1) simple, easily applied, and inexpensive; 2) able to identify people at high and moderate risk for disease; 3) useful for targeting interventions and positively influencing healthy behaviors; and 4) without undue cost or harm. The participants also voiced overwhelming support for continuing both dialogue and active efforts to fill remaining gaps in knowledge about the validity and utility of this approach. Among the next steps outlined at the end of the workshop were: 1) publication of articles based on the workshop presentations; 2) formation of a working group to develop and test a family history tool; 3) identification of opportunities for using existing data to assess the validity and utility of family history for disease prevention; and 4) preparation of a “manifesto” that can begin to form a consensus on a research agenda for family history.
On May 1-2, 2002, the Office of Genomics and Disease Prevention (OGDP) in the National Center for Environmental Health (NCEH), Centers for Disease Control and Prevention, convened a national meeting in Atlanta, Georgia, to discuss the use of family history for improved disease prevention and management.
Because of the multifactorial nature of most common chronic diseases, DNA-based tests to predict the onset of these diseases may not be available for years. Therefore, OGDP began to think of ways to identify people who could be at increased susceptibility for diseases that are preventable. Family medical history seemed like a good candidate because it reflects the consequences of inherited genetic susceptibilities, shared environment, and common behaviors. Family history is not a new concept; it is known to be a risk factor for most chronic diseases of public health significance including coronary heart disease, diabetes, several cancers, osteoporosis, and asthma. However, the collection and interpretation of family history has rarely been applied in preventive medicine and public health to assess disease risk and influence early detection and prevention strategies.
A number of methods have been proposed for quantifying the risk associated with family history based on the number of family members affected, the closeness of the relatives affected, and the occurrence of disease at younger ages than would be expected. If this information could be used to stratify the population into risk groups (i.e., average, moderate, and high), would people who may be at above average risk benefit from targeted prevention and screening programs beyond what is recommended for the population at large?
The overall purpose of the meeting was to explore the potential of family history as a public health tool for improved disease prevention, and begin to develop a research agenda for evaluating the feasibility, validity, and utility of this approach. A major component of the research would be to evaluate simple tools for collecting family health history that can be used in public health and preventive medicine settings.
OGDP envisions a family history tool for public health and preventive medicine that
Is simple, easily applied, and inexpensive
Can identify people at high and moderate risk
Can be used in combination with other risk factors
Is useful for targeting interventions
Positively influences healthy behaviors
Is amenable to population-based use
This workshop, entitled “Family History for Public Health and Preventive Medicine: Developing a Research Agenda,” involved 36 invited participants from a variety of backgrounds and perspectives (see Appendix A for participant roster). The meeting was designed as a series of presentations followed by intense discussion (see Appendix B for meeting agenda). Participants were encouraged to share their thoughts and recommendations in writing on a structured worksheet (Appendix C). This report summarizes the meeting discussion and worksheet comments.
Specific goals of the workshop were to
Generate a list of diseases and conditions that could be included in a family history tool, and specify criteria for selection of these diseases;
Describe the specifications for a family history tool;
Identify gaps in knowledge about the analytic validity, clinical validity, and clinical utility of family history;
Describe the ethical, legal and social implications;
Specify the types of studies needed to fill in the knowledge;
Identify existing data and studies where analysis could be done;
Outline new studies and data collection that may be needed.
An evaluation framework was used to structure the presentations and discussion. The framework, as depicted in the wheel below, was developed by the Foundation for Blood Research as a model process for assembling, analyzing, and disseminating data on the safety and effectiveness of DNA-based genetic tests and testing algorithms. The workshop organizing committee determined that this framework could also be used to assess the validity and utility of using family history to stratify risk leading to improved disease prevention. The four components of the framework are analytic validity, clinical validity, clinical utility, and the ethical, legal, and social implications of using this approach.
Analytic validity addresses how accurately and reliably the tool identifies disease among a person’s relatives. The key elements of analytic validity are sensitivity, a measure of how well the family history tool identifies relatives with disease, and specificity, a measure of how well the tool identifies the relatives who do not have disease.
Clinical validity is how well family history of disease can be used to stratify disease risk and predict future disease in a person. The specific elements of clinical validity include sensitivity, specificity, and negative and positive predictive value.
Clinical utility is an assessment of the impact and usefulness of the family history tool for individuals, families, and society. Of particular interest is whether the classification of individuals into risk groups would improve the effectiveness of early diseases detection methods and interventions.
Ethical, legal and social implications are also important because knowledge of family history may bring unexpected stigma, psychological impact, discrimination, informed consent requirements, and risks to privacy and confidentiality.
The meeting began with presentations on existing knowledge about family history as a risk factor for four diseases – coronary heart disease, colorectal cancer, type 2 diabetes, and asthma. Presentations followed that addressed each of the four components of the evaluation framework – analytic validity, clinical validity, clinical utility, and ethnical, legal and social implications. The first section of this report summarizes the four disease-specific presentations. The next four sections address the components of the evaluation framework and the questions posed during meeting discussion and in the structured worksheet (Appendix C). The final section summarizes next steps proposed by meeting participants. The issues and suggestions raised here should serve as the foundation for a research and action agenda that furthers the use of family history for public health and disease prevention.
The first four presentations summarized the literature on family history as a risk factor for four chronic diseases.
Coronary heart disease
Sharon Kardia described several key studies of the family history of coronary heart disease. Family history information that can be used to predict disease includes: first-degree relations (father/mother), paternal history, maternal history, paternal myocardial infarction (MI)/coronary artery bypass grafting (CABG) before age 55 years, and maternal MI/CABG before age 65 years. Although the results of the studies vary, they seem to indicate that family history is a nonthreatening, noninvasive measure of “genomic risk” or “gene-environment combinations” for coronary heart disease and should be integrated into public health practice. Family history is something genomic we can do now, identifies high-risk families for primary prevention, can be used to educate the public about genetics, can be used to assess people’s attitudes toward the collection of genetic information, and personalizes a family’s risk.
Bob Millikan discussed epidemiologic studies of family history and cancer and their causal explanations, non-causal explanations, and future directions. Studies consistently report a two-fold increased risk for colorectal cancer in persons with one, or more relatives with the disease, a finding consistent across many studies. Causal explanations include major and minor genes, shared environmental risk factors, or a combination of the two. Major genes syndromes, such as HNPCC and FAP have high penetrance, but common, low penetrance alleles such as NAT1 and NAT2 are also possible. Environmental risk factors that family members could share include: meat intake and cooking methods; folate, methionine, alcohol; medication use; physical activity; and insulin resistance. Their effects may be mediated by low penetrance alleles (gene-environment interaction). Noncausal explanations include increased screening or access to care in persons with affected family members, as well as recall bias. Although what family history of colorectal cancer represents it is not yet clear, it may be a useful proxy for a higher prevalence of at-risk genotypes and environmental factors, especially genes and environmental factors that interact.
Type 2 diabetes
Karen Edwards noted that family history for type 2 diabetes has a consistent positive association with disease and aids in risk stratification. Relative risks ranging from 1.5 to 6, with the greater risk is associated with earlier age of onset in relatives, number of affected relatives, first-degree vs. second-degree relatives, and maternal effects. Issues in assessing risk include misclassification of type 1 vs. type 2, undiagnosed diabetes, and reduced penetrance; recall bias; confounders; statistical issues such as differences in family size and lack of independence; and analytic validity. Evidence is insufficient to determine whether family history information improves early detection and prevention of diabetes and whether it influences health promoting behaviors.
Wylie Burke commented on asthma, a complex disease caused by genetic and environmental factors. A PubMed search on “asthma & family history” and additional articles identified through citations revealed nine relevant studies with populations greater than 750 published after 1990 addressing family history as risk for childhood asthma. The limitations of these studies were determination of family history by parental reporting; use of different definitions of asthma; determination mostly by parental reporting of wheezing, current wheeze or cough, physician-diagnosed asthma, and recent use of asthma medications; and variation in populations by age, method of recruitment and location. Even though they indicated that family history is useful for identifying increased risk for asthma, the degree of risk is uncertain. More outcome data are needed on the usefulness of family history for primary prevention and identification of risk for severe disease.
Analytic validity addresses how accurately and reliably the tool identifies disease among a person’s relatives. The key elements of analytic validity are sensitivity, a measure of how well the family history tool identifies relatives with disease, and specificity, a measure of how well the tool identifies the relatives who do not have disease.
Kristen Peterson provided an overview of family history tools. Family histories are collected by providers, insurers, employers, individuals, genealogists, courts, social services workers researchers. Pedigrees, tables or charts, questionnaires, and narratives are employed to gather medical and health information in families, social and cultural traditions, behaviors and habits that impact health, and environmental exposures. Information is used primarily to provide care, assess risks, offer guidance, prevent disease, aid adoptions and social services, support genealogy, and conduct research.
The multigeneration pedigree is the gold standard because it captures large amounts of information in a compact standardized format that is amenable to analysis and is a dynamic document that can be continually updated. However, the pedigree requires training and skill for maximal effectiveness, is time consuming to collect, and is of questionable usefulness as a screening tool.
Many other tools are available in various formats and from different sources. They all have a common purpose: diagnosis of a present medical condition or risk assessment to prevent a medical condition. Using a standard set of descriptive evaluation criteria, these tools were characterized in terms of tool format, degree of relation, length of tool, source of tool, type of questions, information collected, and intended use. Several tools were analyzed, excluding the Family Health Tree Tool described in a later presentation. A listing of several tools can be found in Appendix D.
Hoda Anton-Culver then discussed issues related to validating family history tools. Family history information about cancer is collected in clinical and research settings and is used to infer risk of the disease in population-based, case-control, cohort, and family-based studies. However, little information is available about the accuracy of proband reporting. The question remains: Can we use positive family history of cancer as a surrogate to estimate and characterize high-risk populations? Using cancer registries, Anton-Culver validated the reporting of family history of cancer by cancer-affected probands in population-based and clinic-based family registries of breast, ovarian, and colorectal cancer. The analysis found a high reliability of reporting family history for most cancer sites among first-degree relatives and moderate for second- and third-degree relatives. Other conclusions were that over-reporting of cancer was rare (2.4%); race or ethnicity and sex of the proband did not influence the accuracy of reporting; and degree of relationship to the proband, age at diagnosis of the proband’s cancer, and source of ascertainment of probands were statistically significant predictors of reporting.
What factors will affect analytic validity?
Setting: What settings are likely to yield the most valid information?
Different settings should be evaluated, including medical offices and schools.
Take-home questionnaires issued by physicians allow patients to learn what is important and to query family members on their own time. Discussion can be based on completed questionnaires and the questionnaires can be updated with physicians during annual visits.
Telephone interviews are not effective because they rely on immediate recall. In addition, information that can be captured by telephone is limited (because of time, detail and mode limitations); thus telephone interviews work best for simple information. Ideally, patients should have time before the interview to think about the questions, look up information, and talk to relatives.
The appropriate type of setting depends on the disease or disorder in question. For instance, if the condition is Mendelian, then a telephone or one-time interview may be sufficient. However, if the disease is multifactorial, the questionnaire should be a take- home instrument, continuous, and validated against relatives’ information.
The setting should encourage cooperative completion by various family members, followed with edits by a medically knowledgeable interviewer.
Format: What formats are likely to yield the most valid information?
A combination of formats and media is needed to reach everyone. Cultural, educational, and other personal factors must be considered.
The appropriate format may depend on the disease or disorder.
The format used needs to be able to be updated and revisited periodically because family histories change from year to year.
Web-based methods allow easy storage, retrieval, and updating of information. Computer access is not yet universal, however, so studying the validity of this delivery method in lower socioeconomic and minority populations may be difficult. An alternative is written self-administered or interview-administered questionnaires that can be analyzed by a health professional.
A workbook or a computerized questionnaire can help construct a graphical family history, as demonstrated in the SAGE-PAGE trial.
Pedigrees may be more educational than family histories (e.g., to explain pattern of inheritance and demonstrate disease variability).
Disease: What diseases are likely to yield the most valid information? What criteria should be used for including specific diseases?
Validity tends to be higher for breast, colorectal, ovarian, and brain cancer in first-degree relatives than for cancers in other sites. Reliability for coronary heart disease also seems to be good. On the other hand, because of the high prevalence of undiagnosed diabetes and asthma, family histories for these diseases may be less valid.
Reporting family history information for diabetes may be influenced by definition of family (biological vs. adopted); perception of condition (e.g., if taking drugs and under control, then diabetes is not present); cultural norms (diabetes is not considered a disease state but rather a natural part of aging); and concerns about confidentiality (from proband and relatives).
Criteria for inclusion need to be independent from the likely validity of family history information.
Criteria to consider are diseases that
Are more serious and have a higher death rate
Have a defined set of diagnostic (direct or differential) criteria
Are preventable, with available interventions or simple treatments
Are known genetic disorders
Have an increased relative risk associated with a positive family history and are most commonly encountered in practice or occur in the populations
Have a family history that is likely to be known, rather than undiagnosed or concealed
Have a good “operational definition” of the phenotype of interest (specificity)
Criteria should reflect public health objectives and priorities. Taking a family history of cancer, for example, may instigate earlier screening. regardless of the availability of genetic testing. In addition, “screening” for some diseases may be more acceptable than “screening” for others (e.g., family history of suicide or depression).
Research is needed to look at combining three or four polymorphisms or environmental factors to produce populations of high risk.
Risk Factors: Should information about risk factors (e.g., diet, exercise, smoking) be included in the tool?
Risk factors are helpful for complex disease in which several genes and environmental or lifestyle factors play a role in modifying risk. Such information allows targeted interventions to both individuals and high-risk populations. However, it may add imprecision, unrealistically increase questionnaire length, and therefore be less feasible in a public health setting.
Gathering risk factor information is especially important if it
Is diagnostically relevant
Elicits information about tobacco and alcohol or substance use
Modifies the relation with family history
Relates to the research goal
Whether information should be collected only on risk factors of the proband or also on family members is not clear. If the latter, the information could be linked to reflect family habits, cultural norms and environmental factors.
Information about risk factors should be part of the general advice offered to all patients and should be customized for each person on the basis of the results of family histories and exams.
Assessing diet with brief screening tools may be difficult except for infrequent behaviors like vegetarianism.
Other factors that may affect validity include multivitamin use and residency location of family members.
The ultimate criteria for a family history tool is how well it can stimulate participation in proven interventions that prevent disease and improve health and the impact it has on disease incidence and prevention.
The few cancer validity studies that compare validity of data about first- and second- degree relatives suggest lower validity for second-degree relatives. However, inheritance patterns may be difficult to identify in small families or for diseases affecting one sex.
A common language that uses standard definitions and pedigree symbols is important. The symbols and definitions developed and published in 1995 are an international standard (see Bennett et al, “Recommendations for Standardized Pedigree Nomenclature,” Am J Hum Genet 1995;56:745-52; and Bennett, “Practical Guide to the Genetic Family History, Chapter 4 and appendix with minimal components of family history).
If feasible, family structure and composition should be collected once–and for all family members–so they can be shared throughout the family without compromising confidentiality. Then the pedigree, with information about diseases of interest, can be superimposed on this family structure. Details for each person could be kept private, but familial risk stratification (“bottom line”) could be shared with all at-risk family members. In this way, a “skeleton” pedigree would be available for multiple uses and users.
How can the sensitivity and specificity of a tool be assessed?
Ideally, a gold standard is needed, to compare information from the tool with more detailed and valid information. This is difficult, but one reasonable approach is to use an instrument, then follow up with a detailed diary, in-depth interview, or record check. The key is to determine whether the detailed interview picks up more or different information than the form being tested.
Possible gold standards are relatives’ self-reports, death certificates, pathology reports, and medical records.
Sensitivity and specificity depend not only on the disease or disorder but also on the diagnostic criteria. Each disease should be approached uniquely.
Not much comparative research exists (at least for cancer) on the validity of specific tools in assessing family history accurately, especially in a population-based setting.
What studies have already been done to assess analytic validity?
A small but substantial body of literature, both in population-based and clinical settings, deals with the assessment of sensitivity of cancer family history (less about specificity). Validation studies have been conducted using medical records, cancer registries, and interviews of relatives to verify data collected from a specific tool. However, assessment of variables affecting reporting of cancer family history varies greatly from study to study, and few population-based comprehensive validation studies exist.
In addition, several researchers have studied cardiovascular disease, substance abuse, family violence and certain perinatal or congenital conditions (see studies by Benson et al, and those cited by Kardia and Edwards in their presentations).
What studies need to be done to assess analytic validity?
Validation studies are labor intensive and expensive when well designed, which may be why the few existing validation studies use only one ascertainment method and do not compare various methods.
Important studies to pursue include
Assessment of negative report validity
Validity of general populations reports
Comprehensive, population-based validation studies of cancer family history using pathology or medical records as the gold standard and incorporating all variables affecting family history report (including family structure and tools used)
Analysis of the pros and cons of approaching other family members for information
Determination of the best age for taking family histories
Assessments of what people know, the accuracy of their information and the means by which the information is transmitted.
Validation studies of the family history of asthma, mental illness, and diabetes
Ways to improve recall, especially if the history of different diseases is assessed at one time
Differences between various methods of data collection for different populations
Study design should ensure sufficient “power” to address the relevant questions
Do data sources exist that could address analytic validity?
Cancer registries are good validation sources if they have high levels of completeness and pathology-confirmed data. Death certificates are another good source, but their accuracy varies by disease and is limited to confirming disease in deceased relatives.
Confirmation sources are limited for many chronic diseases not covered in registries.
Other sources are hospital discharge datasets; managed-care office visit data; and data from health maintenance organizations.
What additional studies are needed?
Studies are needed to answer the following questions
What is the best tool for collecting family history compatible with use in large populations (particularly those at moderate risk)?
Can the “free market” develop an effective, user-friendly, and inexpensive software package for documenting family history?
Is having an official “family history day” worthwhile?
What is the validity of reporting age at diagnosis? What are the appropriate cut-off ages to deem diagnosis or death from various diseases premature?
Should family history include all or only site-specific cancers? Should it include such conditions as benign tumors or birth defects?
Which tools best encourage recall?
What is the appropriate balance between accuracy and burden of collecting family history information?
What is the basis of family history information?
What is the validity of family history data collected at multiple intervals on a population basis through an interviewer’s administered questionnaire when the collected data are verified by registries or another gold standard?
What are the core elements of family history?
What is the validity of a report of family history for primary relatives compared with that of grandparents or other secondary relatives?
If only first-degree relatives were included in a family history, how many individuals at high risk for disease would be missed?
What strategies should be adopted to motivate probands to share and discuss information about family histories with their relatives?
Clinical validity addresses how well family history of disease can be used to stratify disease risk and predict future disease in a person. The specific elements of clinical validity include sensitivity, specificity, and negative and positive predictive value.
Maren Scheuner discussed family history collection and pedigree analysis for chronic disease. Occasionally single-gene disorders exist, but inheritance is more often multifactorial with combinations of genes and environmental factors. Characteristics associated with genetic susceptibility to chronic disease are early onset of disease; risk for multifocal or recurrent disease; aggressive course; resistance to conventional risk factor modification and therapies; the same or related conditions in affected family members; and occurrence of conditions consistent with a known genetic syndrome. Genetic risk can be measured in a clinic using a lab test, physical exam or procedures, or pedigree analysis. A review was conducted of pedigrees for coronary artery disease, stroke, diabetes, and cancer (colon, breast, endometrial, ovarian, and prostate). Results were stratified by average risk (sporadic), moderate risk (familial), and high risk (hereditary). Researchers concluded that family history for genetic risk identification and stratification is comprehensive, prevalent, quantitative, qualitative, and accurate.
Joelyn Tonkin reported on the extent to which Americans know their family histories of asthma and heart disease. Using a nationwide, population-based survey called Healthstyles that is conducted each year by CDC, data were analyzed to determine the proportion of people who can report a complete family history; review the association between incomplete family history knowledge and demographic and health related variables; and determine whether useful information about disease risk can be obtained from people reporting incomplete family histories. Study findings indicated that most people can report complete family history; some characteristics may be related to reporting an incomplete family history; disease status is associated with knowledge of family history; and some family history information is better than none.
Ingrid Hall explored family history of cancer and cancer screening in the general population. Her research used the Cancer Control Topical Module of the 2000 National Health Interview Survey to determine the prevalence of cancer screening behaviors among persons with and without a first-degree family history of breast, prostate, and colorectal cancers. She concluded that 1) persons with a family history of cancer are more likely to be screened, regardless of test (and are also more likely to receive screening according to recommended guidelines for mammography and colonoscopy or flexible sigmoidoscopy); 2) although most respondents received screening during a routine physical, a small proportion of those with a family history sought out screening; and 3) the lack of a screening recommendation from a physician was a common reason for not being screened among the never screened. Although no guidelines exist for PSA testing, persons with a family history of prostate cancer were more likely to be screened within the past year.
What factors will affect the clinical validity of the tool?
Relatives: Should more than first-degree relatives be included in the tool?
Studies of gene characterization will lead ultimately to the assessment of clinical validity. For some of these, full pedigree information may be necessary.
The more family history knowledge, the better. Information should be gathered on parents, children, siblings, aunts and uncles, first cousins, and grandparents from both paternal and maternal sides.
The inclusion of extended family members depends on the value of added information relative to the effort needed to gather it. It also depends on the age of the proband and on what is best for a valid prediction.
Disease: What diseases are likely to yield the most valid information?
Diseases likely to yield more valid information are
Diseases people are concerned about–cancer (especially breast and colon) and heart disease
Serious diseases and those that shorten lifespan
- Diseases that produce physically evident symptoms
Diseases involving surgery
Common diseases–diabetes, coronary artery disease, cardiovascular disease, cancers, Alzheimers disease, and asthma
Diseases associated with higher mortality e.g., ovarian cancer, cardiovascular disease associated with a salient event (e.g., myocardial infarction or stroke)
Diseases with better case definitions and well-established diagnostic criteria
Diseases for which interventions exist but are not yet applied population-wide
Whether the same criteria are valid for a variety of adult diseases on the basis of number of affected relatives, degree of relationship, side of family affected, and age at onset is not clear.
Algorithms developed to classify families for risk assessment purposes should be reviewed to determine which are most useful.
Risk stratification: What types of classification systems or family history scores are useful for stratifying risk?
Number of relatives affected (first and second degree, respectively)
Laterality or mulitfocality of the disease
Age of relatives at diagnosis
First-degree relatives plus grandparents may be the optimum combination for simplified data collection and prediction.
Systems need to be disease specific and adhere to the KISS (keep it simple, stupid) principle. One option is to set an update interval (e.g., 2 years or 5 years), supplemented with regular or enhanced screening.
Definitions for average, moderate, and high risk may differ for different diseases.
Continuous variables are needed for research, incorporating degree of genetic relationship and age of onset.
According to Silberberg, et al., different schemas for producing family risk scores are fairly equivalent.
People at average risk and moderate risk may have difficulty comprehending and maintaining accurate risk estimates. Clinicians should carefully explain these estimates to their patients.
How can the sensitivity, specificity, and predictive value be assessed?
To advance the field of systematic family history data collection, standardized data for identifying family relationships should be adopted. Security measures and policies for sharing family data within families and among clinicians would also be helpful. Uniform “tags” for type of relative (e.g., brother, half-brother with same mother) should be used by all professionals. Ways of labeling epidemiologic data by family clusters should be explored. Also, ways of linking existing geneologic data with health data should be developed and archival pedigree showing family composition could then have an “overlay” of disease information.
Comparative research (at least for cancer) is scant on the validity of specific tools in assessing accurate family histories, especially in a population-based setting.
Assessment requires prospective studies, ideally through a national health system with integrated medical records. Thorough family history data should include number of relatives affected, age at diagnosis, and laterality. Data should be verified by other means.
Pilot studies are needed of draft tools developed jointly with the professionals who will use them.
How can the attributable risk due to family history be determined?
Calculating attributable risk requires
Gathering population data on prevalence, both of different categories of familial risk and of the disease or condition of interest
Measuring the incidence of each disease in families with positive family history for each disease
Subtracting risk for disease among unexposed group (with no family history) from the risk for disease among exposed group (with family history), given that the two groups are comparable with regard to other risk factors
What studies have already assessed clinical validity?
Relevant studies include
The Gail model, which predicts risk for breast cancer on the basis of several known risk factors including family history. The model does not account for family structure or size, however, and does not include information about second-degree relatives;
The Claus models for breast cancer;
Colon cancer studies among health professionals and nurses;
Studies correlating family history information with actual genetic testing data.
What studies need to be done to assess clinical validity?
Studies are needed to
The Characterize penetrance (population) and identify effect modifiers for known susceptibility genes
Validate risk assessment models using existing case-control or cohort studies that collected information about the appropriate risk factors. (Accuracy of prediction varies because of misclassification of risk factors, or failure to include all possible risk factors in the model.)
Determine the best way to stratify
Correlate family history with genetics
Learn more about genotype-phenotype correlation, controlling for shared environmental factors
Determine the predictive value of family history using prospective, population-based studies
Evaluate the feasibility of systematic family history collection in conjunction with epidemiologic studies and random clinical trials of prevention and treatment (to identify the differential effect on those at higher familial risk)
Determine the importance of clustering of risk factors by family history and the age at diagnosis
Recommend the interval at which family histories should be updated
Review methods for examining the best threshold of information collection in the context of clinical validity
Determine the maximum number of diseases and environmental and behavioral factors that can be included in a general family history tool before responses and interest wane and the minimum number without sacrificing important conditions with a strong genetic component
Assess the value of having one risk figure that integrates family histories of various relatives and weights their contributions appropriately
Determine whether family histories outlived their usefulness because family size is decreasing and diseases are being diagnosed earlier
Examine the sensitivity and specificity affected by contributions from multiple family members; size of family; number of affected individuals; format (graphic vs. text or tabular); distance of relative (first or second generation); age structure of family; and age of informant
Explore how far back (in generations) should family histories go to be accurate and useful
Do data sources exist that could address clinical validity?
Some sources are
Cancer family registries
Cancer genetics network
Data collected in the clinical settings
Health maintenance organizations that have piloted computer-generated reminders that could be adapted to prompt physicians to conduct a family history
Retrospective cohort studies where family history and health status have been verified for relatives and household members
Clinical utility is an assessment of the impact and usefulness of the family history tool for individuals, families, and society. Given a tool that has reasonable analytic and clinical validity, would the classification of individuals into risk groups improve the effectiveness of available early detection methods and interventions?
Steven Hunt introduced the session on clinical utility of family history with a presentation titled, “Population Versus High-Risk Prevention Strategies: Do We Need to Choose?”. According to Hunt, family history is important to assess because it significantly and independently predicts disease incidence and often represents the expression of multiple risk factors. In addition, most families will have a family history of at least one disease. Those at highest risk need greater help but also respond in the best manner. Lastly, family history achieves the goals of both a population and high-risk approach.
Family history is an independent predictor of disease. On a practical level, family history can be the only basis of diagnosis, although it is difficult to know which contributing factors are accounting for disease. To determine family diagnosis, one can look for shared genes and risk factors in affected relatives, study unshared risk factors in unaffected siblings, and examine whether risk factors are penetrant for disease endpoints.
Hunt discussed the Health Family Tree Study, particularly in relation to relative risk for stroke and coronary artery disease. The cost of identifying high-risk families is often seen as a big drawback of the high-risk approach because it may cost many times the intervention. Estimated at $27 per high-risk family ($3-$5 per family) using paper versions of the Health Family Tree Study, the cost becomes very low ($1-$3) using an internet version.
The benefits of using a high-risk strategy for family history are profound. Family histories of an entire population are collected and analyzed, families become aware of diseases in relatives as they collect the information, a family history report will put the family history into perspective, and medical personnel can encourage lifestyle modification and further assessment.
National family history screening can easily be expanded from a school-based approach for use by public health agencies, medical personnel (e.g., clinics, physicians) and the general public. This was done in Utah with a school-based Internet family history protocol accompanied by a school curriculum, student and teacher manuals, training on DVD and the Internet, and collection and reporting tools. Efforts are underway to make this a statewide program in Utah. At the same time, more research is needed to assess the impact on high- and average-risk families of collecting the family history and receiving a family history report with health recommendations based on that report. This impact can be assessed by change in diet, exercise, smoking; use of medical resources; and long-term health benefits.
Janet Audrain also addressed the topic of clinical utility, specifically addressing the question of whether awareness of family history affects behavior change. Using family history of breast cancer as an example, she examined a variety of studies in terms of the impact of family history on risk counseling, risk comprehension, and screening as well as behavioral and psychological factors that may interact with awareness to impact screening and health behavior. She concluded that targeting interventions to individuals who have a family history of disease may be an effective strategy for improving health-promoting behaviors. However, interventions that focus only on increasing awareness of risk or accuracy of perceived risk may not be sufficient for large or long-term behavior change.
The psychological and behavioral variables that may affect the processing of risk information or participation in health-promoting behavior need to be considered.
Does knowing about a family history of a particular chronic disease make people engage in behaviors that improve health? Are public health interventions more effective if they are targeted to high-risk groups?
The answer to this question may vary by disease. For example, if screening for a particular disease is already high, the utility of family history may be low (e.g., breast cancer) unless it can be shown that knowledge of family history will motivate those who are not getting screened to do so.
Intervention may be more cost-effective if targeted to those at highest risk. Intervention also may be more effective if the attributable risk is appreciable, the cost of stratifying risk is low, and proven interventions exist.
The school-based approach to family history can be expanded to any type of national assessment tool for use by public health agencies, medical personnel, and the general public.
Although the ratio of cases prevented to population targeted will probably increase, the total number of cases prevented in the general population may not necessarily increase.
A standard definition of “public health intervention” is needed.
Thorough literature reviews are needed to assess the state of the knowledge and design a research agenda.
Are individuals more motivated to improve their health if they know they may be at higher risk than the average population? Are they more likely to adhere to screening recommendations?
Such knowledge clearly has some impact. A higher level of awareness and most likely an increase in screening exist, but changes in longer-term behavior and lifestyle are far less definitive. People seem more willing to make low-effort changes than larger sacrifices. Theoretically, such changes could be maximized if people were told their absolute risks based on their family history, and how much behavior change could reduce that risk.
Physicians should be trained and motivated to use family history as an effective tool. Medical school curricula should be reviewed to determine what is currently taught and what changes are needed.
A good literature review could help to make sense of mixed evidence. Also valuable would be data on communication and support for healthy behaviors in families as a result of identification of familial risk.
More research is needed for different populations and ages.
Would individuals in the average-risk groups become complacent and less likely to engage in healthy behaviors? Does the public’s perception of genetic determinism influence behavior?
Although complacency is possible, there is no hard evidence that this will occur. Depending on the health behavior, the general public may already be complacent because most people have a known family history of some chronic disease.
Many of the preventive behaviors are similar for a number of these conditions; thus, family history might be used to encourage adoption of healthful behaviors for everyone. Family-based approaches can then reinforce and complement population-based approaches.
Genetic determinism may promote fatalism, and it affects people’s ability to accept uncertainty of incomplete penetrance, unknown significance of mutations, and unproven management. Physicians have a definite role to play in improving knowledge about risk and its prevention and management.
General public health messages about healthy behaviors should still occur, but people at risk due to family history could receive more intensive efforts.
Is the use of family history to stratify risk and target interventions a cost-effective approach?
Research is needed on appropriate follow-up care for people who screen positive for family history and are at high risk for disease.
Interventions need to be defined from the public health perspective.
This approach appears to be useful, particularly when “population” screening is not accepted or effective. A less expensive way to obtain relevant information is difficult to imagine. However, this may depend on the disease and the cost of collecting information.
Perhaps interventions or recommendations may not change, but intensity of promotion or support in complying with these recommendations might increase for individuals at elevated (high or moderate) risk for disease.
Modeling would help to determine whether risk assessment and targeted screening, which may be expensive on an individual basis, prove less costly if all close relatives who are at risk (by virtue of the index person’s risk status) can also be targeted for intervention without the high cost of initial risk assessment.
What studies have already assessed clinical utility?
Past studies have examined
Change resulting from various types of information interventions
The value of family history information in increasing screening or improving healthy lifestyle behaviors
Attitudes toward genetic testing and behavioral consequences (by Eleanor Singer, a sociologist and survey researcher at the University of Michigan)
What studies need to be done?
Research questions include the following:
What is the added value of family history? Does identification of risk increase primary prevention efforts? Is it cost-effective or ethical to allocate resources for secondary prevention based on family history?
Do different medical specialties need different family history tools?
How do perceptions of “high risk” differ by such variables as culture, race, ethnicity, religion, and age?
What effect might family history have on participation in healthy behaviors of people without a positive family history?
Do people need to completely understand the concept of risk to take appropriate preventive behaviors, such as screening?
Can family histories affect the health of people at moderate risk for disease?
How can family histories best be embedded into physicians’ already busy routines?
When should physicians refer their patients to a genetic counselor? What criteria and family history results would lead to such referrals?
How are family history tools being used in the clinical setting?
Which theories have the most relevance to motivating people to take positive action on the basis of their family histories?
What are the outcomes and behaviors of patients told they are at average or above average risk in terms of their adherence to screening recommendations and adoption of a fatalistic approach?
To what extent does the intensity of promoting interventions or recommendations for people at elevated (high or moderate) risk for disease affect compliance with those recommendations?
What is the difference in lifetime risk of disease between a group with family history information and counseling and another group without that information?
What are the health outcomes and costs of systematic and prospective family history as part of routine clinical care?
Is perceived risk causally related to screening and health promoting behaviors?
How is information about elevated risk best communicated?
Does focusing on specific prevention messages detract from others (e.g., give people “permission” to engage in fewer other preventive health behaviors)?
Do data sources exist that could address clinical utility?
Potential data sources are
CDC study on colon cancer
Utah data from Family Health Tree Experience of the University of Utah and Utah Department of Health
States with active breast cancer or colon cancer programs
Province of Ontario and Britain, which have instituted rule-based criteria on referral for BRCA genetic assessment (and may be collecting relevant data)
Harvard cancer risk tool, which provides an individualized message about cancer prevention and may be the subject of effectiveness evaluation
Ethical, legal and social implications are also important because knowledge of family history may bring unexpected stigma, psychological impact, discrimination, informed consent requirements, and risks to privacy and confidentiality.
Wylie Burke delivered the final presentation of the meeting on the ethical, legal and social implications relevant to family history. Issues addressed included avoiding harm, family history versus predictive genetic testing, duty to warn, potential for harm, anxiety associated with knowledge of affected relatives, human subjects in research, and confidentiality. She concluded that family history is most useful as an independent predictor when preventive interventions are available and when those interventions are imperfect, expensive, inappropriate for most people, and difficult to prioritize.
Is stigma associated with being at above average risk for disease?
Considerable “inappropriate” labeling exists because of public misconceptions. Appropriate messages that will protect and motivate people need to be developed.
Stigma is influenced by individual, societal, and cultural factors. It also varies by disease severity and availability of intervention.
To advance the field of systematic family history data collection, standardized data for identifying family relationships should be adopted. Security measures and policies for sharing family data within families and among clinicians would also be helpful. All professionals should use uniform “tags” for type of relative (e.g., brother, half-brother with same mother) and explore ways of labeling epidemiologic data by family clusters. Ways of linking existing geneologic data with health data should be developed, and archival pedigrees showing family composition could then overlay disease information.
What is the psychological impact to the individual on being at above average risk?
On the basis of their own knowledge of family history, individuals may already believe they are at high risk.
The psychological impact of being at high risk is an area ripe for research. The range of impacts could include fatalism (unwillingness to make changes), unnecessary anxiety, impairment of self-image, depression, shame, and blame.
Impact varies with the individual and family coping style and the way in which information is presented and used.
Does potential exist for discrimination or adverse effects on personal and family life?
Without more quantitative and qualitative data, the potential for discrimination or adverse effects on personal and family life is hard to judge. Family history may create concerns about future decisions (e.g., children and marriage partners). Affected relatives may be blamed for “causing” disease in younger family members.
What are the informed consent requirements for collecting medical information on individuals and family members?
Requirements should include a description of
Rationale for why the consent is important
Benefits and risks associated with giving consent
Obligations of the consent givers and collectors
Specific choices for exclusion and inclusion of specific information
Affected individuals should be approached first to seek permission to contact relatives.
Members of the public generally believe that family members have a duty to inform each other of medical and genetic risk factors if that knowledge could affect the relative’s health or health care. However, this may vary by culture and ethnicity and should be studied further.
A distinction should remain between research and clinical or public health practice.
Do effective safeguards exist that should be in place to protect privacy and confidentiality?
Safeguards should be comparable at least to safeguards that apply to other medical records. They may need to be more stringent because they pertain to data collected on other people.
Few states have laws that protect private medical information and discrimination; thus, comprehensive legislation is needed at the federal level. Such laws would clearly demonstrate the public’s concern to insurers and employers.
To develop effective safeguards, we need more knowledge about the real (not just the theoretical) threats.
Research is needed to develop a “reasonable person standard”–the point at which a family member becomes a research subject and has a right to confidentiality protections.
The NIH Certificate of Confidentiality may be a useful model.
Encrypted electronic transmission should be considered.
What studies have already assessed the ethical. legal, and social implications of the use of family history?
Studies on genetic testing should be examined because the issues related to predictive genetic testing are similar to family history issues. Other recent studies include:
BRCA detection follow-up (Toader, Rowley, Genetic Testing in 1999)
The Hereditary Hemochromatosis and Iron Overload Study (NIH/NHLBI), which is using an integrated ELSI component in phenotypic and genotypic screening, medical exams and family studies
Studies of risk perception and psychological responses to cancer risk assessment
Studies that inform people of at-risk genotype (e.g., alcohol use, smoking) and observe preventive behaviors.
What studies need to be done to assess the ELSI associated with the use of family history?
Some important research questions follow are
What are the real risks? How can they be countered or minimized?
What impact can be expected from an educational message that “everyone is at increased risk for something?”
How does family history affect insurance and employment?
To what extent and how does family history affect social relationships? How is this complicated by the various types of family structures that exist today?
Do data sources exist that could address these issues?The legal and insurance industries may have useful data.
Even thought much progress has been made in using family histories to further public health aims, more work is needed. Meeting participants overwhelmingly supported continuing both dialogue and active efforts to fill remaining gaps in knowledge through research and practice. Among the immediate next steps mentioned were
Posting the workshop presentations on the CDC Web site.
Publishing articles based on the presentations.
Preparing a “manifesto” that can begin to form a consensus on a research agenda for family history
Considering a resolution to make Thanksgiving Day a Family History Day
Participants also agreed to form a small subgroup to interact with relevant groups on family history issues, remain engaged with CDC, digest the meeting discussion, and plot next steps. CDC offered to lead the coordination of such a group, and several participants volunteered to serve as members. Potential tasks include
Designing and publishing a family history tool using materials already developed by many groups. The tool must reflect consensus on the core data elements and may indeed consist of more than one instrument–a set of tools that can be employed in various ways in different settings by different preventive medicine providers and public health practitioners;
Validating the tool using a chosen gold standard;
Field testing the tool on a population basis. The tool should be viewed as “Version 1.0,” with all relevant disclaimers, and its development an iterative process. To the extent possible, the field test should involve users–both consumers and providers–and include social and clinical measures;
Establishing baseline information about current use of family history tools and information for prevention;
Preparing a research agenda in each of the four areas of the evaluation framework. The agenda should be multidisciplinary and focus on filling knowledge gaps that will help validate family history tools and their use for prevention;
Conducting a literature review on whether perceived risk for common chronic diseases can affect behavior change;
Exploring the concept and content of clinical guidelines to advise providers on how best to administer and use family history tools, what to recommend to their patients, and how to motivate and support the adoption of healthier behaviors;
Incorporating family history questions into the CDC Behavioral Risk Factor Surveillance System (BRFSS) so that common risk factors (e.g., diet, exercise, smoking) can be stratified by family history risk (i.e., average, moderate, and high).
All of this work, and more, will help lay the foundation for clinical trials and ultimate widespread use of family history for prevention.