NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Wilson B, Qureshi N, Little J, et al. Clinical Utility of Cancer Family History Collection in Primary Care. Rockville (MD): Agency for Healthcare Research and Quality (US); 2009 Apr. (Evidence Reports/Technology Assessments, No. 179.)

Cover of Clinical Utility of Cancer Family History Collection in Primary Care

Clinical Utility of Cancer Family History Collection in Primary Care.

Show details



The Potential of Family History Information in Preventing Cancer

Family history has always been a core tool in medical practice. A person's family history reflects the combined influences of genetics, environmental exposures, and behaviors within families.1 A large number of reports demonstrate that a positive family history is a risk factor for many chronic diseases of clinical importance, including cardiovascular disease, diabetes mellitus, stroke, and several cancers.2, 3 The greater the number of relatives affected by a disorder, the younger their ages of onset, and the closer the relationship to the individual in question, the more likely it is that the family's disease experience has a genetic basis. However, given the low population prevalence of genetic forms of common, complex disorders, a screening approach based on identifying fairly extreme family histories offers very limited public health or clinical utility in practice.

Historically, the practice of clinical genetics has been largely predicated on detecting individuals marked out by membership in families with unusual patterns of disease, and who are at significantly elevated individual risk as a result of rare genetic disorders that have relatively high penetrance. The alternative approach, which is the focus of this review, is to recognize that family clustering of disease risk reflects combinations of lower penetrance, and moderate risk alleles which are reasonably common within a population. The latter example, a high prevalence of people with slightly or moderately elevated inherited disease risk, would lead to a much higher total disease burden within a population than that associated with a low prevalence of people at very high risk.52 Under the right circumstances, a ‘genomic test’ - such as the presence of particular family history characteristics - could have a significant positive predictive value.

Several years ago, Yoon and colleagues1 illustrated this point. They demonstrated how even fairly simple family history information could be used to clarify the risk of a number of common, complex disorders. Based on published data, they suggested that a healthy 23 year old man could have the following lifetime risks of:

  • 60 percent for cardiovascular disease (based on one male first degree relative (1DR) diagnosed after age 60)
  • 50 percent for colorectal cancer (based on two 1DRs diagnosed before age 50); and
  • 30 percent for type 2 diabetes (based on one 1DR diagnosed after age 60

Importantly, their risk prediction was based on family history information which would not be considered extreme. This application of family history information complements the vision of ‘personalized medicine’ which physician-geneticist Francis Collins M.D., PhD., predicted would be available in 2010, in the form of DNA-based genome profiling tests.53 While recent research indicates that such profiling is possible in principle,54 it is evident that this is not yet a technology ready for routine implementation in health care.52, 55 Until such time, family history represents a potential source of useful predictive information already available to any health care provider.

Discussing the role of family history in coronary heart disease risk, Kardia and colleagues describe the family history as “compressed information” which integrates risks arising from shared genomic components, and social and physical environments.4 They expand:

  • “For, example, we know that parents and children share exactly half their genes. This translates practically into sharing one copy of the 30,000 to 50,000 genes estimated to be in the human genome in the nucleus of each nucleated cell in the human body. ... it is quite possible that even with our ability to measure hundreds and thousands of genes and environments we may find that family history is the best, low-cost way to identify the at-risk subgroups in the population. This will be especially true if gene-gene and gene-environment interactions play a major role in determining risk of future disease.”

This perspective is supported by the work of Butterworth,3 who recently conducted a comprehensive meta-analysis of the risk conferred by family history for a number of common, complex disorders, including the cancers of interest in this review. Table 1 summarizes his findings for colorectal, breast, ovarian, and prostate cancer.

Table 1. Pooled relative risk estimates (95% confidence intervals) for cancers of interest.

Table 1

Pooled relative risk estimates (95% confidence intervals) for cancers of interest.

These meticulous analyses provide direct evidence to support the association between family history and risk of cancer even when family history is captured in a less specific or extensive way than is the case in a specialist genetics consultation.

In order to understand the specific nature of family history information which might be most useful in primary care settings, it is useful to consider how family history data may serve different purposes. Figure 1 maps out different contexts for family history taking in primary care against the general locus of clinical management (primary care versus specialist referral) if elevated risk is identified. It demonstrates how the goal of family history taking may include, but is most definitely not limited to, identifying rare, high risk disease patterns in families which warrant referral for formal genetic evaluation (domain A). Domain B represents systematic screening of general patient populations for a defined range of familial or genetic diseases. Domain C represents the assessment of disease-specific family history information with other risk factor data in the assessment of chronic disease risks in individual patients. In all three domains (A, B, C), a primary care practitioner may be the key provider who captures the family history data; however, the drivers for capturing such information, and the likely decisions that will follow, vary across the three domains and extend beyond the possible identification of classical genetic disorders.

Figure 1. Family history-based primary care activities.


Figure 1. Family history-based primary care activities.

Domain A represents the assessment of a patient in whom clinical suspicions of an inherited disorder or predisposition have arisen, for example in response to concerns about several relatives affected by cancer. In this situation, the main clinical goal would be to assess whether the individual met the criteria for referral for specialist genetics assessment and/or genetic testing. A family history ‘tool’ in this situation would be represented by guidelines for genetic referral or testing. Here, the focus would be on capturing specific family history information which might be rather extensive and specific, requiring attention to, for example, different combinations of diseases in the family, the lineage of affected relatives, and so forth, as illustrated in the Bethesda criteria for non-polyposis colorectal cancer.56

Domain B might typically represent a periodic health assessment in an otherwise healthy individual, where there would be no suspicion of underlying genetic disease and no particular indications of illness or disease susceptibility. Here, the goal would be to conduct a broad brush assessment across a range of common disorders, to identify issues which warrant further probing by the primary care practitioner. The most useful family history tool might incorporate a limited range of the most sensitive family history markers or “red flags” across a range of disorders, in the expectation that more detailed information would be collected for those conditions where an indicator item was positive.

In domain C, the primary care practitioner's goal would be to incorporate family history with other clinical and personal information to assess future risk of a specific chronic disease, with a view to ordering further investigations and advising on appropriate risk reduction strategies. In this situation, the focus would be on family history items which were either independently highly predictive of disease risk, or added useful incremental predictive value to other, established risk factors. A family history “tool” in this context might actually consist of a short set of questions within a more comprehensive disease-specific risk assessment guideline. An example of this approach would be the incorporation of information on parental history of myocardial infarction in the National Cholesterol Education Program III guidelines.57

These three domains are clearly not mutually exclusive. However, this approach illustrates that the demands of extensiveness and specificity of the type of information that is necessary is not uniform across all contexts. Until comprehensive and current family history information is available for patients through, for example, electronic records systems, a primary care practitioner's approach to family history-taking may be influenced by external constraints (e.g., time), their prior assessment of the patient's risk of disease, and the way in which the information will actually be used for decisionmaking.

Although this report borrows from the language and concepts of genetics and genomics (see below), its approach reflects the perspective on family history exemplified particularly by domain C described above. Its focus is on the utility of using family history information to stratify the risk of common, complex diseases in individuals who are not specifically selected for suspicion of high genetic risk.

Evaluating Family History for use in the Prevention of Cancers

The usefulness of family history stratification systems as predictive tools for common, chronic diseases can be approached using an evaluation framework originally developed by the Secretary's Advisory Committee on Genetic Testing.58 This framework has four elements (1) analytic validity; (2) clinical validity; (3) clinical utility; and (4) ethical legal and social implications of using a test. It has been further developed for application to the evaluation of family history information in disease prevention (see Table 2 for definitions).59 Put another way, the line of evidence between family history and individual and population benefit requires answers to the following questions:


Does a positive family history (however defined) predict future risk of cancer in an individual patient sufficiently accurately to be useful in a clinical setting?


If so, are there interventions available which reduce the risk of cancer? Do these interventions also carry risks?


If there are interventions which help prevent cancer, does information on family history-based disease risk mean a person is more likely to adhere to them (compared with advice which is not based on knowledge of family history)?


Are there any direct harms which arise from the process of taking a family history and feeding back personalized risk based on family history?

Table 2. Elements and key components of evaluation framework for family history as screening tool [reproduced from Yoon, Scheuner, and Khoury 2003].

Table 2

Elements and key components of evaluation framework for family history as screening tool [reproduced from Yoon, Scheuner, and Khoury 2003].

In terms of efficient health care resource use, it is also legitimate to examine the incremental benefits and costs associated with capturing and using family history information, whether it substitutes for, or adds informational value to, other risk factor information, and whether it is easier and/or cheaper to obtain.

In a previous effectiveness report,5 evidence regarding the accuracy of reporting by individuals of their family history of breast, ovarian, colorectal, and prostate cancer was synthesized, and a large number of family history tools and family history-based risk assessment tools identified and reviewed. The current review is designed to inform issues (a) to (d) above, which address the issues of clinical validity and, in part, clinical utility.

Risk Stratification

A number of risk classification systems exist, generally in the form of guidelines or algorithms developed to assist in decisionmaking around referral to genetic services or genetic testing (e.g., Rodriguez-Bigas56). Few risk stratification systems have been developed specifically for direct application in primary care, with a focus on recommending behavior changes and/or preventive clinical interventions in which referral for genetic counselling would be relevant for only a small sub-group of patients. One such system is that proposed by Scheuner and colleagues60 who, in a paper predating Butterworth's analysis,3 used available epidemiological data6170 to define three risk strata for a number of complex disorders (Table 3). An approach like this has the merit of appearing practical for immediate and easy application in primary care settings, but further evaluation is necessary to determine its predictive ability for the disorders of interest (clinical validity).

Table 3. Family history-based risk stratification guidelines for breast, ovarian, colorectal, prostate cancers.

Table 3

Family history-based risk stratification guidelines for breast, ovarian, colorectal, prostate cancers.

Effectiveness of Cancer Risk Reduction Interventions

Even the most highly predictive and practical risk stratification system cannot change health outcomes unless effective cancer prevention interventions are available for patients at risk. While there is extensive epidemiological literature on a wide range of cancer risk factors, much of the evidence regarding cancer prevention is based on observational studies. Contradictory evidence has emerged when some apparently protective factors, identified through observational studies7173 were associated with increased risk of cancer in experimental studies, as was the case for alpha-tocopherol.74 It is crucial that any assessment of the clinical utility of family history-based risk stratification take into account the highest quality evidence on the benefits and risks of recommended preventive interventions.

Effect of family history taking on uptake of interventions. Assuming that proven preventive interventions are available, the question remains whether individualized, family history-based advice actually promotes uptake of such interventions by patients. Simplistically, it might be assumed that individuals would be more motivated to act on a health care provider's advice if they know they are at higher than average risk. However, there is a body of literature in the area of predictive and predisposition genetic testing which cautions that knowledge of genetic risk might lead to fatalism in those at higher risk (failing to take up available interventions because of the ‘inevitability’ of disease) or complacency in those at average risk (failing to take up available interventions because of lower perceived personal risk - the ‘certificate of health’ effect).75 Like any healthcare intervention, family history taking may incur incremental costs in time, energy, and money over and above standard care, therefore an examination of the incremental benefits is important.

Risks inherent in family history-based risk assessment. No healthcare intervention should be assumed to be safe without formal assessment of harms as well as benefits. While family history taking is seen as a standard activity in all areas of healthcare, the systematic capture of more extensive information, and its purposeful use in individual risk assessment, merits objective review. The ACCE framework58 emphasizes the assessment of both harms and benefits, and there is substantial literature that examines the impacts of predictive genetic tests beyond simple accuracy.76 While some guidelines77 suggest a cautionary approach, with cancer family history collection undertaken only in response to patients' enquiries, many other observers argue that family history taking is an integral part of good clinical practice.78, 79 Objective evidence on the prevalence of specific harms of family history taking, and of its use in advising patients about disease risk reduction, is necessary in order to clarify the appropriate level of caution to be exercised, and the types of situations in which patients might be most vulnerable to potential harms.

Scope and Purpose of the Systematic Review

This report is intended to build on a published evidence report focused on collection and use of family history for breast, ovarian, colorectal, and prostate cancers.5 The key questions for that project are available at: The current systematic review addresses four key research questions relating to the clinical validity and utility of routinely using family history information in risk assessment and prevention of breast, ovarian, colorectal, and prostate cancer in primary care, as follows:


Which risk stratification algorithms or guidelines delineate risk accurately, and in a clinically meaningful way?


For which behaviors and clinical preventive services (‘interventions’) is there evidence of benefits in terms of actual reduction in disease risk, and what harms, if any, have been identified?


For those interventions identified as being based on reasonable evidence, what is the evidence that providing information on risk status results in behavior change or increased uptake of services on the part of individual patients?


What are the harms or risks to individual patients that may result from the collection of family history information in itself, and/or the provision of family history-based risk information?

Addressing these four key questions requires a focus on different types of evidence and different sets of literatures. The focus of key question (Q1) is not only on identifying family history-based risk prediction systems (which may be presented as guidelines, algorithms or other tools) suited to use in primary care, but also on assessing their actual predictive ability when applied to individual patients. This requires review of primary studies addressing discriminatory accuracy in cohorts representative of relevant general populations. Key question (Q2) asks whether effective cancer prevention interventions are available. While answering this question is an essential step in addressing the overall question of clinical utility of cancer family history taking, we note that the primary goal of this report is on family history taking, rather than evidence for cancer control per se. Given the considerable practical implications in attempting to synthesize the very extensive literature on cancer risk factors and prevention (see, for example, the reports by the World Cancer Research Fund26, 45, 46), the reasonable expectation that evidence reviews might be available for cancer prevention interventions which are considered standard of care, and direction from the sponsors of the report, the focus is therefore on reviewing systematic reviews for primary and secondary (screening) cancer prevention. Key questions (Q3) and (Q4) are evaluative questions. Question 3 asks whether taking and using family history information is more likely to lead to desired behavior changes in patients than other approaches which do not use family history information. In addressing this question it is important to distinguish between studies which examine the behavior of people who have pre-existing perceptions of elevated disease risk because of living with a “family disease” from the clinical strategy of systematically capturing family history information and personalizing risk assessment for all patients, in order to promote adherence to recommended preventive behaviors. The possibility for confounding of the latter by the former cannot fully be addressed in observational studies, and steers the review towards examining evidence from well-designed intervention studies. The same issue applies in question (Q4), where adverse outcomes from the clinical strategy of taking and using family history information needs to be separated out from the psychological and social impacts of living with the implications of pre-existing perceived familial disease risk.

As discussed above, the focus of this review is firmly on using family history information in a primary care context. This has driven the eligibility criteria for studies towards

  • study populations that resemble those in primary care - with an inherent range of disease risks but not selected because of suspicion of genetic disease
  • study settings where primary care providers such as family physicians, internists, nurse practitioners, and obstetricians are taking family histories and assessing risk
  • family history taking as an intervention carried out by primary care practitioners and directed primarily towards chronic disease risk assessment and prevention as an end in itself
  • cancer prevention interventions evaluated in primary care or general populations with an inherent range of disease risks, but not selected because of special high risk (genetic or otherwise)

Even within primary care, we recognize that there is an inevitable gradation (rather than a clear cut separation) between family history taking as a means to promote effective primary care-based disease risk assessment and management (domain C, Figure 1), and family history taking as a way of identifying individuals who may be at high genetic risk, where referral to specialist genetics services is warranted (domain A). We therefore sought to include studies examining family history taking as a generalist activity, and to exclude studies which focused on family history as part of a clinical genetics assessment. Thus, even though we recognize that family history taking is a core activity in specialist genetics, this review excludes literature where the focus is primarily on the assessment of genetically high risk patients in specialist settings.

It is also important to emphasize that the focus on study populations “unselected” for high risk implies groups of participants which represent a full range of risks, potentially from very low to very high, with clustering around an “average” value (by definition). This criterion was adopted in an effort to reflect professional and patient decisionmaking in “typical” primary care contexts where patients with a wide range of risks (but mostly “average”) are encountered. Thus, it would be expected that a predictably small proportion of patients from an unselected population would be classified by a risk stratification system into a high risk category. This situation is distinctly different from those where patients and their providers already know or suspect that they are at high disease risk, by virtue of, for example, an uncommon pattern of familial disease, a positive genetic test result in a close relative, a previous diagnosis of the condition, or diagnosis of a pre-disease state. In short, populations unselected for high risk may include some high risk individuals whereas populations selected for high risk definitively exclude average and low risk individuals. We suggest that it cannot be assumed that findings from “selected for high risk” studies are directly applicable to the general primary care context.

PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...