3
Biological Material in Household Surveys: The Interface Between Epidemiology and Genetics

Kaare Christensen

Many traits—from health outcomes to behavior and wealth— have a tendency to run in families. Families share not only environment (including sociological and socialization factors), but also genetic factors. Epidemiologists have traditionally looked for environmental causes for variations in health outcome and behavior, while geneticists have focused on tracing genetic factors of importance. The interaction between these two research traditions has been surprisingly slow in emerging (e.g., the International Genetic Epidemiology Society was founded in 1991), although many researchers in both areas agree that the determinants of most health outcomes and behaviors are to be found in the interaction between genes and environment. Not only are both genes and environment etiological factors, but the effect of an environmental factor depends on the genetic background upon which it acts and vice versa. One of the major obstacles to conducting gene-environment interaction studies is that it usually requires large sample sizes to get reasonable statistical power to detect an interaction. The collection of biological material in sizeable household surveys could provide a sound basis for future gene-environment interaction studies.

The inclusion of biological material in household surveys can also improve information about environmental factors usually studied by epidemiologists, such as diet and exposure to heavy metal compounds and pesticides. Biological material in household surveys can be collected in many ways (e.g., blood, cheek brushes, saliva, urine, sperm, hair, and nails) for many purposes. For the interface between epidemiology and



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? 3 Biological Material in Household Surveys: The Interface Between Epidemiology and Genetics Kaare Christensen Many traits—from health outcomes to behavior and wealth— have a tendency to run in families. Families share not only environment (including sociological and socialization factors), but also genetic factors. Epidemiologists have traditionally looked for environmental causes for variations in health outcome and behavior, while geneticists have focused on tracing genetic factors of importance. The interaction between these two research traditions has been surprisingly slow in emerging (e.g., the International Genetic Epidemiology Society was founded in 1991), although many researchers in both areas agree that the determinants of most health outcomes and behaviors are to be found in the interaction between genes and environment. Not only are both genes and environment etiological factors, but the effect of an environmental factor depends on the genetic background upon which it acts and vice versa. One of the major obstacles to conducting gene-environment interaction studies is that it usually requires large sample sizes to get reasonable statistical power to detect an interaction. The collection of biological material in sizeable household surveys could provide a sound basis for future gene-environment interaction studies. The inclusion of biological material in household surveys can also improve information about environmental factors usually studied by epidemiologists, such as diet and exposure to heavy metal compounds and pesticides. Biological material in household surveys can be collected in many ways (e.g., blood, cheek brushes, saliva, urine, sperm, hair, and nails) for many purposes. For the interface between epidemiology and

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? genetics, however, the genetic material—the DNA—is of the greatest interest, and due to technological developments DNA can now be obtained from very small samples of body fluids or tissues. Genetic factors obviously play a central role in a broad range of monogenic diseases (i.e., diseases caused by mutation in one gene), from cystic fibrosis to Huntington disease to early-onset dementia. Most common diseases, however, are not monogenic, and are more likely to be influenced by a large number of environmental and genetic factors and their interactions. The first part of this chapter will consider the evidence for the impact of genetic factors on variations in survival, reproduction, health, and behavior—traits which are central to demographers and social scientists. The second part will describe examples of gene-environment interaction, point towards design options suitable for studying geneenvironment interaction within the framework of a household survey, and provide estimates of required sample sizes. Finally, the feasibility of collecting biological material in household surveys will be evaluated based on a series of recent Danish surveys among middle-aged and elderly persons and the oldest-old, including such aspects as the methods for and the costs of, both monetary and in terms of potential response rate decrease, sampling biological material. DO GENETIC FACTORS PLAY A SIGNIFICANT ROLE IN THE VARIATION IN SURVIVAL, REPRODUCTION, HEALTH, AND BEHAVIOR? This is a central question when considering whether demographers and social scientists should include genetic material in household surveys. If there is no strong evidence that genetic factors play a role in the variation in a number of important traits, there seems to be little reason to undertake the logistic and ethical challenge of including genetic material in household surveys. Twin and adoption studies can estimate the overall genetic influence on a given trait, while other kinds of family studies are not very well suited to disentangle the effects of genes and common environment. Adoption Studies For logistical reasons adoption studies are much fewer and smaller than other kinds of family studies. Nevertheless, adoption studies have had a major impact on the nature-nurture debate concerning a number of traits, because these studies have produced remarkable results and their designs are easily understood and interpreted. Adoption studies use the fact that adoptees share genes but not environment with their biological

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? families, and environment but not genes with their adoptive families. Among the most notable findings from adoption studies is one from Heston (1966), which showed that among 47 children who had schizophrenic mothers and who were adopted away, 5 developed schizophrenia, while none of the 50 control adoptees developed schizophrenia. Although the sample size is small, the study convincingly demonstrated that schizophrenia has a strong genetic component. Another intriguing finding that surprised many nongenetically oriented researchers comes from a Danish adoption study of body-mass index (weight in kilos divided by height in cm squared), which is a measure of body composition (obesity). This study showed that the body-mass index of adoptees correlated more with the body-mass index of their biological relatives than their adoptive relatives, indicating a strong genetic component to variation in body composition in settings with no shortage of food supply (Stunkard et al., 1986). An investigation of early-adult life mortality in the same study population similarly showed a genetic component to premature death (Sørensen et al., 1988). The adoption studies, however, are not without weaknesses. In particular, a bias can be introduced by selective placement of adoptees (i.e., the adoptees are preferably placed with adoptive parents who resemble the birth parents in some ways). This bias tends to overestimate the effect of both genetic and shared family environment. Twin Studies In humans two types of twinning occur: monozygotic (identical) twins, who share all their genetic material, and dizygotic (fraternal) twins, who on average share 50 percent of their genes, like ordinary siblings. A twin study of a condition/disease in its simplest form is based on a comparison of monozygotic and dizygotic concordance rates (i.e., the probability that a twin has the condition under study given that the co-twin has it). A significantly higher concordance rate in monozygotic than in dizygotic twins indicates that genetic factors play a role in the etiology. For continuous traits, intraclass correlations are used instead of concordance rates. The twin study does not identify specific genes that affect a given trait, but rather, assesses the overall effect of genetic factors: the degree to which differences in the phenotype are attributable to genetic differences between people. To estimate the heritability of a trait (i.e., the proportion of the population variance attributable to genetic variation), twin data can be analyzed using standard biometric models (Neale and Cardon, 1992). A number of recent developments in twin methodology have taken place based on the incorporation of measured genotype information. This enables twin models to estimate how much of the genetic variation is due to variation in a specific gene. Gene-environment interaction studies, link-

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? age analyses, and association studies can also be performed within a twin population. As with adoption studies, twin studies are not without weaknesses. Of particular concern has been the “equal environment assumption,” that is, the assumption that the degree of intrapair environmental similarity is the same in monozygotic and dizygotic pairs. If, in fact, the degree of environmental similarity is greater in monozygotic twins, then the heritability is overestimated. Plomin et al. (1994) have made a comprehensive review of twin studies of a number of important medical and behavior disorders as well as personality traits (Figures 3–1, 3–2, and 3–3). Below is a description of recent twin studies of particular interest to demographers and social scientists. FIGURE 3–1 Identical twin [monozygotic (MZ)] and fraternal twin [dizygotic (DZ)] probandwise concordances for behavioral disorders. Average weighted concordances were derived from a series of studies. SOURCE: Reprinted with permission from R.Plomin, M.J.Owen, and P.McGuffin. 1994. The genetic basis of complex human behaviors. Science 264(5166):1733–1739. Copyright 1994, American Association for the Advancement of Science.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? FIGURE 3–2 MZ and DZ probandwise concordances for common medical disorders. Average weighted concordances were derived from series of studies. SOURCE: Reprinted with permission from R.Plomin, M.J.Owen, and P.McGuffin. 1994. The genetic basis of complex human behaviors. Science 264(5166):1733– 1739. Copyright 1994, American Association for the Advancement of Science. Life Span During the last decade a variety of twin studies have shown that approximately 25 percent of the variation in life span is caused by genetic differences. This seems to be a rather consistent finding in various Nordic countries in different time periods and even so among other species not living in the wild (Herskind et al., 1996; Iachine et al., 1999; Finch and Tanzi, 1997). Reproduction Kohler et al. (1999) studied the genetic dispositions influencing fertility and fertility-related behavior using Danish twins born in the period 1870–1964. It was found that genetic influences on fertility exist, and that

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? FIGURE 3–3 MZ and DZ twin intraclass correlations for personality (neuroticism and extraversion), vocational interests in adolescence, scholastic achievement in adolescence (combined across similar results for English usage, mathematics, social studies, and natural science), specific cognitive abilities in adolescence (memory, spatial reasoning, processing speed, verbal reasoning), and general intelligence. Average weighted correlations were derived from series of studies. SOURCE: Reprinted with permission from R.Plomin, M.J.Owen, and P.McGuffin. 1994. The genetic basis of complex human behaviors. Science 264(5166):1733– 1739. Copyright 1994, American Association for the Advancement of Science. their relative magnitude and pattern depend on sex and on the socioeconomic environment experienced by successive birth cohorts. Genetic effects were most pronounced in periods with consciously controlled fertility, suggesting that the genetic disposition primarily affects fertility behavior and motivation for having children. Analyses of fertility motivation in some of the more recent twin cohorts, measured by age at first attempt to have children, supported this interpretation.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? Health and Diseases For a number of diseases which occur in early or midlife, such as insulin-dependent diabetes and schizophrenia, twin studies have demonstrated the existence of a significant genetic component (Kyvik et al., 1995; Plomin et al., 1994). Genetic factors also influence cardiovascular diseases which occur in early or midlife, while for cardiovascular diseases occurring late in life there is little evidence of a genetic effect (Marenberg et al., 1994). Dementia has a very strong genetic component, not only with regard to early-onset monogenic types but also to late-onset dementia (Breitner et al., 1993; Gatz et al., 1997). Alternatively, twin studies provide little evidence of genetic factors in the etiology of Parkinson disease (Tanner et al., 1999). Twin studies of general physical functioning such as strength (younger individuals) or functional abilities (elderly individuals) show evidence of a considerable genetic influence, an influence that seems to increase with age (Christensen et al., 2000b). Psychology and Behavior The genetic component of physical abilities and diseases is usually broadly accepted. Similar claims regarding psychological and behavioral phenotypes often meet resistance, although a very large body of evidence suggests that heritabilities consistently range from 30 to 50 percent for personality, vocational interests, general intelligence, and scholastic achievement (Plomin et al., 1994). Depression symptomatology (Kendler et al., 1987; McGue and Christensen, 1997) and even proclivity to divorce (McGue and Lykken, 1992) have also been shown to have substantial genetic contributions to their etiology. Summary: Phenotypes and Heritability Twin and adoption studies suggest that a wide variety of phenotypes have a genetic component to their etiology. Naturally, it should be recognized that the heritability estimates are time- and population-specific, i.e., the overall influence of genetic factors depends on the amount of environmental variance in the study population and vice versa. If, for example, more equal access to favorable living conditions and health care is introduced in a population, this is likely to decrease the environmental variance and hence increase the proportion of the total variation attributable to genetic factors (i.e., the heritability). At the same time, an increase in environmental variance as seen in modern societies may also provide the opportunity for genetic effects to be expressed. A substantial heritability for a trait suggests that it may be possible to

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? identify specific genetic variants that influence the trait. The chance of identifying, through genetic association or family studies, gene variants affecting a trait depends on the number of gene variants and the size of their effect. At this point in time few important specific gene variants affecting variation in survival, fertility, health, and behavior are known, but this is likely to change within the next few years. Therefore, including genetic material in household surveys is not so much for immediate use as to ensure that the survey will be valuable in a future scenario where genetic covariates are likely to play an important role in the understanding of variations in traits. The most promising aspect emerging from the identification of specific genes influencing various traits is that this may be the basis for insight into the underlying biological processes and, in particular, how genes interact with the environment. An understanding of such gene-environment interactions can lead to environmentally based prevention or treatment of unwanted conditions or diseases with a strong genetic component, as described in the next section. GENE-ENVIRONMENT INTERACTIONS The classic example of gene-environment interaction is phenylketonuria, an inborn error of metabolism. This disease is caused by a mutation in a gene on chromosome 12 (coding for the enzyme phenylalanine hydroxylase). If a child inherits a mutated gene from both the father and the mother, the child will be severely mentally retarded if it consumes an ordinary diet during childhood. However, a child with two copies of this mutation can have a normal development if he or she gets a diet low in phenylalanine and supplemented with tyrosine, thereby bypassing the lack of enzyme problem. Most countries in the industrialized world screen for this defect among newborns, which makes it possible to prevent the effects of a genetic defect through modification of the environment, in this case the diet. Another example is Smith-Lemli-Opitz syndrome (SLOS), first described in 1964. SLOS also requires two mutated copies to develop in a fetus. Craniofacial anomalies are predominant, in addition to limb and genital anomalies, failure to thrive, and mental retardation. SLOS is thought to be the second most common autosomal recessive disorder among white North Americans after cystic fibrosis, with a carrier frequency of 1 to 2 percent. Molecular research has shown that SLOS results in an error in cholesterol synthesis. Insight into the etiology of this genetically determined disease has induced a medical treatment that may reduce some of the postnatal symptoms. The treatment is a diet high in cholesterol (Tint et al., 1994). The two diseases described above are very rare. However, common

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? genetic variants with a moderate or even small effect on prevalent diseases are likely to be of great importance on a population level, although at present few such genetic variants have been documented. The best example is the gene coding for apolipoprotein E (APOE), a protein that ferries cholesterol through the blood stream. Three common variants of this gene (APOE e2, APOE e3 and APOE e4) are found, which are slightly different forms of the protein. Despite the small differences, the e4 variant has been consistently associated with a moderately increased risk of both cardiovascular diseases and Alzheimer disease (Corder et al., 1993). Not only has APOE e4 been shown to be a risk factor for cardiovascular diseases and Alzheimer disease per se, but several studies have found that the e4 variant is involved in gene-environment interaction, making the e4 carrier more susceptible to environmental exposures. For example, an increased risk of chronic brain injury after head trauma has been observed for individuals who carry the e4 gene variant, compared to non-e4 carriers (Jordan et al., 1997). APOE e4 also seems to modulate the effect of other risk factors for cognitive decline. Individuals with APOE e4 in combination with atherosclerosis, peripheral vascular disease, or diabetes mellitus have a substantially higher risk of cognitive decline than those without APOE e4 (Haan et al., 1999). The APOE genotype has also been found to influence the effect of alcohol on blood pressure in middleaged men (Kauma et al., 1998) and to increase the risk of neurological diseases after HIV infection (Corder et al., 1998). It seems likely that APOE e4 is just one of many common genetic variants in the estimated 50,000 to 100,000 genes in the human genome that increase susceptibility for environmental exposures. Design Options for Gene-Environment Interaction Studies Within Household Surveys Within the framework of a household survey, there are two major design options for studying gene-environment interactions: the cohort study and the so-called “nested case-control study.” Cohort Studies This design uses the entire study population of persons at risk for the outcome of interest and follows this cohort over time. The cohort study design is feasible if the cost of genotyping (e.g., of APOE variants) is inexpensive. In this case, the APOE genotype can be determined for all participants who have provided a biological sample, and all outcomes studied can be stratified not only on sex and age but also on APOE genotype. For example, if one wanted to study cardiovascular diseases, survey

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? participants within a reasonable age range and free of cardiovascular diseases at intake would be genotyped and followed over time. Traditional survival analysis could then be used to assess the effect of the different APOE genotypes on cardiovascular disease risk. In cohort studies, the assessment of gene-environment interactions can be done by comparing the disease-free survival of individuals stratified on genotype (e.g., +/–APOE e4) and exposure to the environmental factor (+/–), or by using standard techniques such as Cox regression analysis. Nested Case-Control Studies This design is an attractive option if the outcome of interest is a rare disease/condition or if the assessment of the gene variant is expensive (which today is often the case, but this is likely to change within a few years). In the nested case-control study, all new cases of the phenotype of interest (e.g., cardiovascular diseases) are included as well as one to four controls selected from the study base. Including more than four controls results in a limited increase in statistical power (Rothman, 1986). The method for selection of controls from the survey population is essential, and must include consideration of factors such as the ethnic background of the cases and the controls (for more details, see Rothman and Greenland, 1998, and Khoury et al., 1993). Only the biological samples from the cases and the one to four controls per case are analyzed for the gene variants of interest. From the frequencies of the gene variants in the case and the control group, the odds ratio can be estimated, which is approximately equal to the relative risk associated with an allele if the outcome studied is not too common (i.e., less than 5 percent of the population studied will get the disease/ condition). Gene-environment interactions are assessed by stratifying on genotype; i.e., the relative risk associated with the exposure is estimated for each genotype separately—most often using a multivariate logistic regression analysis with interaction terms (Hosmer and Lemeshow, 1989). However, as seen in Figure 3–4, several hundred cases and controls are required to detect modest interaction terms, and some researchers argue that in some scenarios the estimated sample size requirement in Figure 3–4 may be underestimated (Garcia and Lubin, 1999). Household surveys, therefore, will often be one of the best methods to fulfill sample size requirements. Problems in Gene-Environment Studies within Household Surveys Household surveys are most often conducted with lay interviewers, which places certain constraints on which phenotypes can be studied. For

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? FIGURE 3–4 Number of cases required for 80 percent power at a 5 percent Type I error in a case-control study designed to detect gene-environment interaction with two controls per case over a series of frequencies of exposure (e.g., head trauma) and proportions of susceptibility (e.g., proportion of ApoE-4 carriers). The proposed odds ratio of interaction equals 4 (i.e., the exposure is associated with a four times higher risk among susceptibles compared to nonsusceptibles). SOURCE: Reprinted with permission from S.-J.Hwang, T.H.Beaty, et al. 1994. Minimum sample size estimation to detect gene-environmental interaction in case-control designs. American Journal of Epidemiology 140(11):1029–1037. Copyright 1994, Oxford University Press. example, in medically oriented surveys, cognitive abilities and depression symptomatology can be assessed, but a dementia or depression diagnosis according to internationally recognized criteria can be obtained only in follow-up studies with the assessment done by specialists. The biggest challenge to gene-environment interaction studies in the years to come will probably be a multiple comparison problem. With 50,000–100,000 genes being identified, many of which are likely to have several variants, an enormous number of possible gene-environment interactions can be studied. To choose a significance level in statistical

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? testing of gene-environment interaction seems difficult. The most reasonable strategy probably will be testing of biologically plausible interactions and verification of positive findings in large studies, which again points towards the important role that household surveys can play in future gene-environment studies. FEASIBILITY OF COLLECTING BIOLOGICAL MATERIAL IN HOUSEHOLD SURVEYS The Danish 1995–1999 Experience Since 1995 our group at The Aging Research Center at the University of Southern Denmark and the Danish Center for Demographic Research has conducted six major surveys: The Longitudinal Study of Aging Danish Twins (three waves in 1995, 1997, and 1999, respectively: LSADT-95, LSADT-97, LSADT-99), The Longitudinal Study of Middle-Aged Twins, The Danish 1905-Cohort Survey, and The Danish Centenarian Study, together comprising more than 10,000 individuals. These surveys provide an opportunity to evaluate the logistic impact of including biological material in ongoing household surveys with lay interviewers in populations of middle-aged and elderly persons and the oldest-old, and to compare such surveys with smaller scale studies done by medically trained persons. Table 3–1 gives an overview of the size of the surveys and their participation rates. The Longitudinal Study of Aging Danish Twins (LSADT) In 1995, LSADT began by assessing all cooperating Danish twins aged 75 years and older (a sample of nearly 2,500 individuals). The assessment TABLE 3–1 Surveys Conducted at the Aging Research Center at the University of Southern Denmark and the Danish Center for Demographic Research in the Period 1995–1999   Mean Age Age Range Participants Interview Response Rate % LSADT-95 81 75–102 2,401 77 LSADT-97 77 73–102 2,172 79 LSADT-99 76 70–101 2,709 72 Middle-aged twins 57 46–67 4,314 83 The 1905-cohort 92 92–93 2,262 63 The Centenarian Study 100 100 207 75

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? was repeated in 1997 by including all twins who participated in 1995 as well as a sample of previously unassessed twin pairs who were between 73 and 76 years old in 1997. In 1999, we included all Danish twins aged 70 years and older. By 1999, over 5,000 individuals aged 70 years and older had completed the LSADT intake assessment, nearly 3,500 individuals had completed a two-year follow-up assessment, and nearly 1,000 individuals had completed both a two- and four-year follow-up assessment. Additional assessments are planned for 2001 and 2003. The first two waves (1995 and 1997) included anthropometric measures and questions on sociodemographic factors, lifestyle habits, selfrated health, diseases, sensory deficit, symptoms, medications, physical abilities, depression symptomatology, family history, and social life. Cognitive abilities were assessed using the Mini Mental State Examination (Folstein et al., 1975) and a number of other tests. The 1997 wave used the same survey instrument. In the 1999 wave, the survey instrument was expanded to include physical performance tests, measurement of lung functioning (spirometry), and sampling of DNA by self-administered finger prick or cheek brushes (see details about DNA sampling in a later section of this chapter) (Christensen et al., 1996, 1998, 1999, 2000a, 2000b; Yashin et al., 1998; Andersen-Ranberg et al., 1999; Kohler et al., 1999). The Longitudinal Study of Middle-Aged Twins To complement the LSADT studies with a twin study of middle-aged twins, we identified a random sample of twin pairs born between 1931 and 1952 through the Danish Twin Registry (Kyvik et al., 1996; Skytthe et al., 1998). In 1999, more than 5,000 twins aged 46–67, including monozygotic twins and fraternal twins of same and opposite sex, were invited to participate in an extensive face-to-face interview conducted by lay interviewers, which included tests of physical and cognitive function. The questionnaire, which was based on the LSADT questionnaire, included items on the twins’ current and childhood socioeconomic status, social network, self-rated health, diseases, use of medications, lifestyle habits, physical activity at work and during leisure time, reproductive history, and a brief food frequency questionnaire (Gaist et al., 2000). The same procedure used in LSADT was implemented for DNA sampling. The Danish 1905-Cohort Survey In order to study nonagenarians, all Danes born in 1905 were invited in 1998 to participate in a home-based two-hour multidimensional interview that included cognitive and physical performance tests. The interview instrument was similar to the LSADT instrument and carried out by

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? the same lay interviewers. Population-based registers were used to evaluate representativeness. Participants and nonparticipants were highly comparable with regard to marital status, institutionalization, and hospitalization patterns, but men and rural residents were more likely to participate than women and urban residents. Despite the known difficulties of conducting surveys among the very old, the study showed that it was possible to conduct a nationwide survey of more than 2,000 fairly representative nonagenarians using lay interviewers. Again, the LSADT procedure was used for DNA sampling (Nybo et al., in press). The Longitudinal Danish Centenarian Study This study is a nationwide survey of all persons living in Denmark who celebrated their 100th birthday during the period April 1, 1995, to May 31, 1996. The residence of all centenarians in the study population was identified through the Civil Registration System by the personal identification number of each centenarian. Approximately two weeks after their 100th birthday, all centenarians received a letter explaining the study, and their permission was sought to allow a geriatrician and a geriatric nurse to visit in order to interview them and carry out a physical examination including phlebotomy. More than 200 centenarians participated (Andersen-Ranberg et al., 1999). What Does Inclusion of Biological Material Cost in Terms of Interview Participation Rate? The LSADT studies have response rates of 72–79 percent, which many regard as very good for extensive surveys among the elderly (Table 3–1). The best way to estimate the participation cost of including biological material in an ongoing survey is to compare the 1997 (no biological sampling) response rate for LSADT-95 participants with the 1999 (biological sampling) response rate for individuals who participated for the first time in LSADT-97. For these two subgroups the participation rates were 81 and 80 percent, respectively, indicating that the costs in terms of decrease in interview participation rate in this setting are minimal. It should be noted, though, that Denmark has a free, national health care system, which may enhance participation because insurance and out-of-pocket costs are of less importance compared to the United States, for example. Although the participants in the studies do not receive any information about their health, the surveys are conducted by a medical school, which in itself may make the participants more willing to provide a biological sample. More disappointing was the survey of the 1905-cohort, which had only a 63 percent participation rate, although analyses of register data on

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? all eligible nonagenarians indicated that the responders were fairly representative. There are several probable reasons for the lower response rate in the 1905-cohort. The group is uniformly very old, and it may not be obvious to people born in 1905 (or their relatives) that they could be of interest to science; hence they may be less committed to participating. The centenarian study showed a considerably higher response rate compared to The Danish 1905-Cohort Survey. This is probably due to the “celebrity status” of centenarians and to the fact that the survey was done by a geriatrician and a geriatric nurse. How to Obtain DNA from the Participants by Means of a Finger Prick and Cheek Brushes The same procedure for DNA sampling was used in The Danish 1905-Cohort Study, The Danish Middle-Aged Twins Study, and LSADT-99 without any notable logistic problems. In these investigations, participants were asked to donate a DNA sample in the form of a blood spot or, if they disliked the idea of blood spots, by means of cheek brushes. A full blood sample of approximately 20 ml would have been preferable because the amount and analytical options are much greater, but this would require a second visit to the participants by a trained phlebotomist (usually a hospital technician) and would approximately double the survey expenses. DNA could also have been obtained by other methods such as mouth lavage or urine (see Wallace in this volume), but we found these methods logistically less appealing. Finger Prick The blood spots were made by the participants themselves, guided by the interviewer, using a sterile automated incision device (Tenderfoot®) with a standardized incision depth of 1.0 mm and length of 2.5 mm. This device has a safety clip, which prevents the premature release of the blade, and a retracting blade, which improves safety by protecting both participant and interviewer from injury due to an exposed blade contaminated with blood (Figure 3–5). The participant chose one of the three lateral fingers on either the right or left hand, and the chosen finger was warm, dry, and clean. The Tenderfoot safety clip was removed and the device was positioned longitudinally at the site of the distal part of the chosen finger and triggered. The first drop of blood was wiped away and the next blood drops distributed to the blood spot card. On the blood spot card there are five roundels, and we aimed at collecting three or four drops of blood on each roundel (in total 5 cm2) (Figure 3–6). The blood spot card was air dried and sent to a laboratory the same day, after being

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? FIGURE 3–5 Utensils for obtaining blood spots by finger prick in connection with household surveys. The blood spots can be made by the participants guided by the interviewer using the sterile automated incision device (e.g., TenderfootR) with a standardized incision. marked with the date, the interview subject number, and the number of the interviewer. Only the blood spot was sent to the laboratory, so that the identity of the participant was unknown to the laboratory, ensuring that all analyses and genotyping were done blinded. To avoid catastrophic sample loss, the blood spots were divided in two and are kept in two locked locations. The blood spot cards are stored in boxes at room temperature. The utensil cost was $2–3 per person and the interview time averaged 5–10 minutes depending on age, but with large variation. Cheek Brushes Alternatively, the DNA sample was taken by means of cheek brushes (Figure 3–7). Three cheek brushes were used for each subject. Again the samples were self-collected by the participant, guided by the interviewer. The brush was inserted in the mouth and twirled firmly against the inner cheek for 30 seconds. This process collects cells from which to obtain the DNA. The brush was placed in the original tube and the cap replaced. A

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? FIGURE 3–6 To avoid sample loss, the blood spots are divided in two and kept in two different locations. FIGURE 3–7 Brushes used to obtain biological samples from the inner cheek in connection with household surveys. The samples can be collected by the participant guided by the interviewer. The brush is inserted in the mouth and twirled firmly against the inner cheek for 30 seconds. This process collects cells from which to obtain the DNA.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? new place on the inner side of the cheek was chosen for each brush. Every brush was labeled with the date, the participant number, and the number of the interviewer, and sent to the laboratory the same day (arrived the next day). In the laboratory the brush head was placed in a tube and kept at –80° C until the isolation of DNA took place (within one month). The utensil cost was $3–4 per person and the interview time averaged 2–5 minutes depending on age, but with large variation. Participation in DNA Sampling Naturally, the respondents in these surveys could choose to participate only in the interview and not the biological sampling (Table 3–2). The reason for lower participation in the DNA sampling among the oldest participants is proxy interviews: when a person in the study population was unable to participate (most often due to dementia), we tried to identify a relative who could answer on behalf of that person (a proxy interview). For ethical reasons we did not sample DNA from the participant in connection with proxy interviews. For nonproxy participants, the difference in participation rate between the surveys was modest: 97 percent in the survey of middle-aged twins (mean age 57) versus 90 percent in both LSADT-97 (mean age 76) and the 1905-Cohort Study. As seen in Table 3–2, the percentage of cheek brush samples is highly dependent on age: nearly all middle-aged participants provided DNA by finger prick, while among the oldest-old a fifth of the DNA samples were cheek brushes. TABLE 3–2 Surveys Conducted at the Aging Research Center at the University of Southern Denmark and the Danish Center for Demographic Research in the Period 1998–1999, Which Included Collection of Biological Material by a Finger Prick or Cheek Brush   Biological Samples   Participants No. % of Participants % of Nonproxy Participants Blood Spots % Cheek Brushes % The 1905-cohort 2,262 1,632 72 90 80 20 LSADT-99 2,709 2,319 86 90 91 9 Middle-aged twins 4,314 4,171 97 97 95 5 TOTAL 9,285 8,122 87 93 93 7

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? Storage and DNA Quality Control The success of any data-intensive project is dependent on the ability to store and retrieve data. It is important to have a facility that is used to handling large numbers of biological samples. In the surveys mentioned above, the Department of Clinical Biochemistry at Odense University Hospital takes care of the registration, storage, and retrieval of the samples. This department handles 3 million biochemical analyses each year, and therefore an addition of some 10,000 samples did not impose major logistic problems. As mentioned, the biological samples are kept anonymously and apart from the information obtained during the interview, but the information is linkable through the participants’ ID numbers. For quality control we have determined APOE gene variants on a subsample of the blood spots and the cheek brushes with a success rate of 99 percent. Based on our experience, the DNA isolated from cheek brushes is of a sufficient quantity and quality to perform at least 200 genotypings for each brush, while each blood spot card can provide DNA for approximately 2,000 genotypings. DNA can be obtained from blood spot cards stored for decades at room temperature (Strobel et al., 1998; Makowski et al., 1996). Although experiences with storing cheek brushes for longer time periods are few, it seems likely that storage at –80° C will preserve the DNA for decades. CONCLUSION Collecting biological material in household surveys is feasible. From the perspective of the interface between epidemiology and genetics, collection of genetic material is valuable because genetic factors do play a role in many traits of interest to demographers and social scientists. The identification of genes of importance and their interaction with the environment can provide the basis for prevention and treatment of unwanted conditions through modification of environmental factors. ACKNOWLEDGMENT This research was supported by a grant from the Danish National Research Foundation and the National Institute of Aging (P01-AG08761). David Gaist, M.D., Ph.D., Hanne Nybo, M.D., and Karen Andersen-Ranberg M.D. coordinated the surveys, and Lise Bathum, M.D., Ph.D., established the biobank. Bernard Jeune, Matt McGue, and James W. Vaupel played a major role in planning these studies.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? REFERENCES Andersen-Ranberg, K., K.Christensen, B.Jeune, A.Skytthe, L.Vasegaard, and J.W.Vaupel 1999 Declining physical abilities with age: A cross-sectional study of older twins and centenarians in Denmark. Age and Ageing 28:373–377. Breitner, J.C., M.Gatz, A.L.Bergem, J.C.Christian, J.A.Mortimer, G.E.McClearn, L.L. Heston, K.A.Welsh, J.C.Anthony, M.F.Folstein, et al. 1993 Use of twin cohorts for research in Alzheimer’s disease. Neurology 43(2):261–267. Christensen, K., D.Gaist, B.Jeune, and J.W.Vaupel 1998 A tooth per child? Lancet 352:204. Christensen, K., N.V.Holm, M.McGue, L.Corder, and J.W.Vaupel 1999 A Danish population-based twin study on general health in the elderly. Journal of Aging and Health 11:49–64. Christensen, K., N.V.Holm, and J.W.Vaupel 1996 Alzheimer’s disease in twins. Lancet 347:976. Christensen, K., M.Kristiansen, H.Hagen-Larsen, A.Skytthe, L.Bathum, B.Jeune, K. Andersen-Ranberg, J.W.Vaupel, and K.H.Ørstavik 2000a X-linked genetic factors regulate hematopoietic stem-cell kinetics in females. Blood 2000 95(7):2449–2451. Christensen, K, M.McGue, A.I.Yashin, I.A.Iachine, N.V.Holm, and J.W.Vaupel 2000b Genetic and environmental influences on functional abilities among Danish twins aged 75 years and older. Journals of Gerontology: Medical Sciences 55A(8):M446-M452. Corder, E.H., K.Robertson, L.Lannfelt, N.Bogdanovic, G.Eggertsen, J.Wilkins, and C.Hall 1998 HIV-infected subjects with the E4 allele for APOE have excess dementia and peripheral neuropathy. Nature Medicine 4:1182–1184. Corder, E.H., A.M.Saunders, W.J.Strittmatter, D.E.Schmechel, P.C.Gaskell, G.W.Small, A.D.Roses, J.L.Haines, and M.A.Pericak-Vance 1993 Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261:921–923. Finch, C.E., and R.E.Tanzi 1997 Genetics of aging. Science 278(5337):1996–1998. Folstein, M.F., S.E.Folstein, and P.R.McHugh 1975 “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research 12:189–198. Gaist, D., L.Bathum, A.Skytthe, T.K.Jensen, M.McGue, J.W.Vaupel, and K.Christensen 2000 Strength and anthropometric measures in identical and fraternal twins: No evidence of masculinization of females with male co-twins. Epidemiology 11(3):340– 343. Garcia, C.M. and J.H.Lubin 1999 Power and sample size calculations in case-control studies of gene-environment interactions: Comments on different approaches. American Journal of Epidemiology 149:689–692. Gatz, M., N.L.Pedersen, S.Berg, B.Johansson, K.Johansson, J.A.Mortimer, S.F.Posner, M. Viitanen, B.Winblad, and A.Ahlbom 1997 Heritability for Alzheimer’s disease: The study of dementia in Swedish twins. Journal of Gerontology: Medical Sciences 52(2):M117–125. Haan, M.N., L.Shemanski, W.J.Jagust, T.A.Manolio, and L.Kuller 1999 The role of APOE epsilon 4 in modulating effects of other risk factors for cognitive decline in elderly persons. Journal of the American Medical Association 282:40– 46.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? Herskind, A.M., M.McGue, N.V.Holm, T.I.Sorensen, B.Harvald, and J.W.Vaupel 1996 The heritability of human longevity: A population-based study of 2872 Danish twin pairs born 1870–1900. Human Genetics 97:319–323. Heston, L.L. 1966 Psychiatric disorders in foster home reared children of schizophrenic mothers. British Journal of Psychiatry 112:819–825. Hosmer, D.W., and S.Lemeshow 1989 Applied Logistic Regression. New York: John Wiley and Sons. Hwang, S.J., T.H.Beaty, K.Y.Liang, J.Coresh, and M.J.Khoury 1994 Minimum sample size estimation to detect gene-environment interaction in case-control designs. American Journal of Epidemiology 140(11):1029–1037. Iachine, LA., N.V.Holm, J.R.Harris, A.Z.Begun, M.K.Iachina, M.Laitinen, J.Kaprio, and A.I.Yashin 1999 How heritable is individual susceptibility to death? The results of an analysis of survival data on Danish, Swedish and Finnish twins . Twin Research 1(4):196–205. Jordan, B.D., N.R.Relkin, L.D.Ravdin, A.R.Jacobs, A.Bennett, and S.Gandy 1997 Apolipoprotein E epsilon 4 associated with chronic traumatic brain injury in boxing. Journal of the American Medical Association 278:136–140. Kauma, H., M.J.Savolainen, A.O.Rantala, M.Lilja, K.Kervinen, A.Reunanen, and Y.A. Kesaniemi 1998 Apolipoprotein E phenotype determines the effect of alcohol on blood pressure in middle-aged men. American Journal of Hypertension 11:1334–1343. Kendler, K.S., A.C.Heath, N.G.Martin, and L.J.Eaves 1987 Symptoms of anxiety and symptoms of depression: Same genes, different environments? Archives of General Psychiatry 44:451–457. Khoury, M.J., T.H.Beaty, and B.H.Cohen 1993 Fundamentals of Genetic Epidemiology. New York/Oxford: Oxford University Press. Kohler, H.P., J.L.Rodgers, and K.Christensen 1999 Is fertility behaviour in our genes? Findings from a Danish twin study. Population and Development Review 25:253–288. Kyvik, K.O., K.Christensen, A.Skytthe, B.Harvald, and N.V.Holm 1996 The Danish twin register. Danish Medical Bulletin 43:467–470. Kyvik, K.O., A.Green, and N.H.Beck 1995 Concordance rates of insulin dependent diabetes mellitus: A population based study of young Danish twins. British Medical Journal 311:913–917. Makowski, G.S., E.L.Davis, and S.M.Hopfer 1996 The effect of storage on Guthrie cards: Implications for deoxyribonucleic acid amplification. Annals of Clinical and Laboratory Science 26:458–469. Marenberg, M.E., N.Risch, L.F.Berkman, B.Floderus, and U.de Faire 1994 Genetic susceptibility to death from coronary heart disease in a study of twins. New England Journal of Medicine 330:1041–1046. McGue, M., and K.Christensen 1997 Genetic and environmental contributions to depression symptomatology: Evidence from Danish twins 75 years of age and older. Journal of Abnormal Psychology 106:439–448. McGue, M. and D.T.Lykken 1992 Genetic influence on risk of divorce. Psychological Science 3:368–373. Neale, M.C., and L.R.Cardon 1992 Methodology for Genetic Studies of Twins and Families. Dordrecht: Kluwer Academic Publisher.

OCR for page 42
Cells and Surveys: Should Biological Measures be Included in Social Science Research? Nybo, H., D.Gaist, B.Jeune, L.Bathum, M.McGue, J.W.Vaupel, and K.Christensen in press The Danish 1905-Cohort. A genetic-epidemiological nationwide survey. Journal of Aging and Health. Plomin, R., M.J.Owen, and P.McGuffin 1994 The genetic basis of complex human behaviors. Science 264:1733–1739. Rothman, K.J. 1986 Modern Epidemiology. Boston/Toronto: Little, Brown and Company. Rothman, K.J., and S.Greenland 1998 Modern Epidemiology. Second Edition. Philadelphia: Lippincott-Raven Publishers. Skytthe, A., N.V.Holm, J.W.Vaupel, and K.Christensen 1998 Establishing a population-based register of middle-aged twins in Denmark (abstract). Twin Research 1(2):110. Sørensen, T.I., G.G.Nielsen, P.K.Andersen, and T.W.Teasdale 1988 Genetic and environmental influences on premature death in adult adoptees. New England Journal of Medicine 318(12):727–732. Strobel, E., C.Emminger, G.Mayer, J.Eberle, and L.Gurtler 1998 Detection of HIV 1 infection in dried blood spots from a 12 year old ABO bedside test card. Vox Sanguinis. 75:303–305. Stunkard, A.J., T.I.Sorensen, C.Hanis, T.W.Teasdale, R.Chakraborty, W.J.Schull, and F. Schulsinger 1986 An adoption study of human obesity. New England Journal of Medicine 314:193– 198. Tanner, C.M., R.Ottman, S.M.Goldman, J.Ellenberg, P.Chan, R.Mayeux, and J.Langston 1999 Parkinson disease in twins: An etiologic study. Journal of the American Medical Association 281:341–346. Tint, G.S., M.Irons, E.R.Elias, A.K.Batta, R.Frieden, T.S.Chen, and G.Salen 1994 Defective cholesterol biosynthesis associated with the Smith-Lemli-Opitz syndrome. New England Journal of Medicine 330:107–113. Yashin, A.I., I.A.Iachine, K.Christensen, N.V.Holm, and J.W.Vaupel 1998 Genetic component of discrete disability traits: Liability models with age dependent thresholds. Behaviour Genetics 28:207–214.