• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

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

National Research Council (US) Committee on Advances in Collecting and Utilizing Biological Indicators and Genetic Information in Social Science Surveys; Weinstein M, Vaupel JW, Wachter KW, editors. Biosocial Surveys. Washington (DC): National Academies Press (US); 2008.

Cover of Biosocial Surveys

Biosocial Surveys.

Show details

7Comments on Collecting and Utilizing Biological Indicators in Social Science Surveys

and .

These are exciting times for population scientists. This volume describes several innovative population surveys that include biological and genetic information and have shed new light on important questions in the social and health sciences. With the rapid rate of technological change in the collection and measurement of biological information and associated reduction in costs, it is increasingly common for biomarkers to be included in large-scale population-based socioeconomic and demographic surveys. It is unlikely that the benefits to science of these innovations will not be realized until study designs more fully integrate the theoretical insights from the biological, medical, and social sciences. Such integration is likely to yield broad new vistas of inquiry and has the potential to herald a new era of scientific discovery at the interface of the health and population sciences.

The chapters in Part I provide a rich and textured description of some pioneering projects from across the globe that integrate social surveys with biomarker measurement. Each chapter tells a tale of exemplary creativity in the design and implementation of studies that have yielded insights regarding a broad array of indicators of health status, their prevalence in specific populations, and, in some cases, their associations with individual characteristics and the social environment. These studies provide a tantalizing array of intriguing facts and promise extraordinary opportunities to better understand the complexities underlying the interplay between social behavior and biology. The studies also highlight the contributions of research that span different social and ecological environments and provide sources of variation that are necessary to successfully discriminate among hypotheses.


The research in this volume reflects the current state of knowledge and the chapters in Part I also suggest several challenges that are likely to be faced in the future. Studies have clearly demonstrated the feasibility of including biomarkers in large-scale population surveys. Important questions remain about how to select biomarkers for inclusion in particular studies as well as questions about the trade-offs between resources allocated to biomarker measurement relative to other dimensions of a survey. The latter includes resources for the collection of other survey items, sample composition and sample size, including the age range of the sample, whether to include other family members in the study, whether to purposively sample particular subpopulations, and whether to follow subjects over time in order to collect longitudinal information on health and social status. These are not easy choices, and one size does not fit all.

Questions about the selection of biomarkers arise, in part, because there is little clarity on which biological, nutritional, and genetic markers should have the highest priority for inclusion in broad-purpose demographic and socioeconomic surveys. The typical population survey is large in scale and multipurpose in scope and designed to support the testing of an array of different hypotheses. Because health is multidimensional, it is tempting to collect a wide array of different biometric indicators to parallel both the broad-purpose nature of population surveys and the broad-based nature of most self-assessed indicators of health status reported in these surveys. Succumbing to this temptation quickly becomes prohibitively expensive. It is difficult to overstate the contributions of recent developments of simple, low-cost methods for the collection, storage, and analysis of biological samples, and future innovations promise to revolutionize the field. However, progress will be limited if we rely on innovation driven by measurement alone. It is important that theory-driven hypotheses also influence the development of technologies for measurement of biomarkers so they can be included in population surveys.

Some of the most influential social science surveys in the last 40 years have been guided by the integration of theory with measurement across multiple disciplines. For example, in economics, ideas that were pioneered by T.W. Schultz, Gary Becker, and their colleagues have influenced the design of many surveys across the globe. These include highlighting that decisions in any single domain of an individual's life (such as health) affect decisions in other domains (such as work); that choices today affect options tomorrow and are thus affected by choices yesterday; that choices are constrained by the opportunities available to an individual; and that the household and the family play a key role in decision making by individuals.

To fully harness the benefits of integrating biomarker collection with social science surveys, it will be profitable to invest in integrating theory from the biological, biochemical, genetic, and behavioral science fields. This integration will provide direction for data collection and analyses that are hypothesis driven, and the evidence from those studies will serve as guideposts for further inquiries.

An Example: The Work and Iron Status Evaluation

An ongoing nutrition intervention that we have been conducting in Indonesia provides an illustration. The study seeks to test the hypothesis that health has a causal impact on economic and social prosperity. Its design was influenced by one of our earlier studies, the 1997 wave of the Indonesia Family Life Survey (IFLS), in which we collected a series of biomarkers. These included anthropometry, lung capacity (with a puff test), blood pressure, physical mobility, and, key for this discussion, hemoglobin (Hb). Hb is assessed using an in-home Hemocue photometer, which is safe, inexpensive, and simple to field, calling for no more than blood from a finger prick.

Data from IFLS indicate that around a third of reproductive-age women (ages 15 to 40) in Indonesia have low levels of iron, as indicated by Hb levels below 120g/L (the cutoff recommended by the World Health Organization). High rates of iron deficiency among women in this age group have been documented in many countries across the globe, and these women, along with infant children, are the focus of most iron supplementation programs. However, there are reasons to suppose that iron deficiency rates may be high among other demographic groups in Indonesia. The typical diet contains little by way of iron-rich foods (such as red meat), and rice, the primary staple, retards iron absorption (particularly from nonanimal sources). Because there was no evidence on prevalence rates at the population level in Indonesia, in IFLS we collected Hb from over 27,000 male and female respondents across the entire age distribution.

Among women, Hb levels decline with age after menopause and almost half of all women ages 65 and older in Indonesia are iron deficient (as indicated by Hb < 120g/L). The recommended cutoff for men is higher (Hb < 130g/L). The incidence of low iron is lower among prime-age men, with about one in six men ages 20-29 having Hb below the cutoff. However, as with women, Hb levels decline with age beginning around age 30. Slightly over half of all men ages 65 and older are iron deficient (as indicated by Hb < 130g/L) (Thomas and Frankenberg, 2002).

This is important. Iron deficiency is associated with elevated susceptibility to disease and fatigue. Moreover, there is a substantial body of literature that demonstrates that iron deficient anemia—the combination of low Hb and inadequate iron stores—results in reduced work capacity, as measured by, for example, V02max. This has been demonstrated in rigorous treatment-control studies with both humans and animals. Because iron plays a critical role in transporting oxygen through the blood, the mechanisms underlying the relationship between iron and work capacity are well understood right down to the cell level. What is not known is whether increased work capacity translates into higher hourly earnings, more income, and improved well-being. It is possible that as health improves and work capacity increases, hourly earnings increase and, with no change in work effort, income will increase proportionately. However, it is also possible that as hourly earnings increase, individuals will allocate less time to work, so that there is no change in income (or income could even decline). Economic theory suggests that the typical behavioral response will depend on the subject's expectations about the longevity of the change in hourly earnings and is likely to lie between these two extremes.

With this background, we designed the Work and Iron Status Evaluation (WISE) to address these questions. The study follows approximately 5,000 older people (ages 30 through 70) and some 12,000 other household members for a period of six years. The study design and some results are described in Thomas et al. (2006, 2007).

Subjects were randomly assigned to receive a weekly iron supplement or an identical placebo for slightly over a year. During the first three years, subjects were interviewed every four months and administered a very detailed survey instrument that collects information on type of work, hours of work, own farm and nonfarm businesses, earnings, wealth and savings, consumption and spending, transfers to noncoresident family, participation in community activities, time allocation, risk and intertemporal preferences, along with self-assessed physical and psychosocial health.

Considerable time and effort were invested in ensuring that subjects took the tablets every week. Supplements and placebos were provided in blister packs of four tablets and, during the first four weeks, subjects were visited twice a week to ensure compliance with the protocols. The frequency of visits was reduced if subjects appeared to be following the protocols. Blister packs were resupplied every four weeks. At every visit to a household, the number of tablets remaining in the blister packs was recorded. Counting the number of tablets removed from the blister packs can provide only an upper bound on the number of tablets ingested. An advantage of focusing on the impact of iron supplementation is that we can use biomarker assessments as a more direct indicator of compliance. Specifically, at each of the four-monthly interviews in WISE, several biomarkers were assessed, including Hb. Changes in Hb for each subject across waves of the study provide biological evidence on compliance with the treatment protocols, at least among those who were iron deficient and received the treatment. In addition, by monitoring changes in Hb every four months, we are able to trace the time path of the effect of the iron supplementation on iron in the blood. We did this during the year that tablets were distributed and also after the supplements ended, in order to assess the longevity of any effect on Hb levels.

Because low Hb alone does not indicate iron deficient anemia, we also measured iron stores with transferrin receptors using dried blood spots and an Elisa assay (McDade and Shell-Duncan, 2002) that were collected at the same time that Hb levels were measured. These data demonstrate that subjects who received iron and had low levels of iron at baseline, had higher levels of Hb and iron stores (as measured by lower transferrin receptors) relative to baseline levels when supplementation ended. For a subsample of respondents, we also measured work capacity by having each subject pedal to exhaustion (or a maximum of 18 minutes) on a stationary ergocycle with weights, while carefully monitoring the subject's heart. The subjects who received iron supplements and who had low Hb at baseline were able to bike longer after the supplementation than similar controls.

The evidence from WISE thus far essentially replicates the results in the nutrition literature. The biomarkers provide direct evidence that among iron deficient subjects who received iron supplements, iron levels have increased. We have not ruled out that part of the low Hb is driven by a hemoglobinopathy. A potentially important candidate is inflammation. Because Hb tends to be depressed by inflammation and, at least in the United States, inflammation rates rise with age, it may be that the age profile of Hb reflects elevated levels of inflammation among older Indonesians. To investigate that issue, we draw on the dried blood spots that were collected as part of the survey and measured a marker for inflammation, C-reactive protein (Cordon, Elborn, Hiller, and Shale, 1991). While inflammation is probably a factor in the observed low levels of Hb in the Indonesian population, the evidence from our data clearly demonstrates that the age profile of C-reactive protein cannot possibly explain the age profile of Hb (Crimmins, Frankenberg, McDade, Seeman, and Thomas, 2007).

Armed with the evidence from biomarker measures, which indicate that the treatment-control intervention has been implemented successfully, we turn to measurement of the impact of a randomly assigned health shock on social and economic prosperity. Consider, first, productivity at work, which we measure by hourly earnings. It is unlikely that someone whose hourly wage is not directly related to productivity would see any increase in wages if their iron levels were improved by the treatment. However, people whose hourly earnings are a reflection of productivity are likely to have higher hourly earnings if they were iron deficient at baseline and received the treatment. We therefore examined self-employed men—most of whom are farmers or construction workers or run bicycle taxis—whose earnings are likely to be closely related to their own productivity. After six months of supplementation, relative to a year before, the hourly earnings of men who received the treatment and who had low iron at baseline were significantly higher than comparable controls. There were no differences among those working in the wage sector. Among men working in the wage sector at baseline, there were no productivity differences between treatments and controls but there were differences in how they allocated their time.


This example provides an illustration of the potential value of integrating theoretical insights from the nutrition and biochemistry literatures with those from the social sciences. The credibility of this study relies crucially on the combination of direction from the health sciences on the likely impact of iron on health and well-being, knowledge of the appropriate markers to assess changes in iron (and thus health) status, and the insights suggested by economic theory on likely responses to changes in productivity associated with improved health. It is hard to imagine designing and implementing a study like this without guidance from theory and practice in all of these disciplines. It is also hard to imagine successfully completing a project like this without an interdisciplinary team of scientists involved in all aspects of the study.


  1. Cordon S, Elborn J, Hiller E, Shale D. C-reactive protein measured in dried blood spots from patients with cystic fibrosis. Journal of Immunological Methods. 1991;143:69–72. [PubMed: 1919037]
  2. Crimmins E, Frankenberg E, McDade T, Seeman T, Thomas D. Inflammation and socio-economic status in a low income population. University of California, Los Angeles; 2007. Unpublished manuscript.
  3. McDade T, Shell-Duncan B. Whole blood collected on filter paper provides a minimally invasive method for assessing human Transferrin Receptor level. Journal of Nutrition. 2002;132:3760–3763. [PubMed: 12468620]
  4. Thomas D, Frankenberg E. The measurement and interpretation of health in social surveys. In: Murray C, Salomon J, Mathers C, Lopez A, editors. Summary measures of population health: Concepts, ethics, measurement, and applications. Geneva, Switzerland: World Health Organization; 2002. pp. 387–420. Chapter 8.2.
  5. Thomas D, Frankenberg E, Friedman J, Habicht JP, Hakimi M, Ingwersen N, Jaswadi, Jones N, McKelvey C, Pelto G, Seeman T, Sikoki B, Smith JP, Sumantri C, Suriastini W, Wilopo S. Causal effect of health on labor market outcomes: Experimental evidence. Los Angeles: California Center for Population Research, University of California; 2006. CCPR working paper #2006-70.
  6. Thomas D, Frankenberg E, Friedman J, Habicht JP, McKelvey C, Jones N, Pelto G, Sikoki B, Sumantri C, Suriastini W. Iron supplementation, iron deficient anemia, and hemoglobinopathies: Evidence from older adults in rural Indonesia. University of California, Los Angeles; 2007. Unpublished manuscript.
Copyright © 2008, National Academy of Sciences.
Bookshelf ID: NBK62445
PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (7.4M)

Related information

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...