In 2006, the United States census tract celebrates its 100th birthday. This remarkable—but often taken for granted—parcel of government-defined geography occupies an important place in our body politic, with direct relevance for addressing inequities in health and health care. Born at the crossroads of planning for congregation size and public health, census tracts are now widely used by all sectors of government and by myriad academic disciplines to describe, analyze, and improve societal well-being. A centenary offers a useful moment to consider the census tract's underappreciated origins and its significance and implications for population health and the public data required for informed democratic governance.

History: From “Sanitary Areas” to “Census Tracts”

In brief, the idea of establishing small permanent geographic areas was first proposed in 1906 by Dr. Walter Laidlaw, Director of the Population Research Bureau of the New York Federation of Churches (Fig. 1).1 Stymied by the then conventional practice of enumerating populations by political wards—whose boundaries changed with each election—Laidlaw's “strong conviction was that unanticipated population movements and general ignorance of neighborhoods and neighbors was gravely hurting church and synagogue efficiency,”2 , p. 4 since they could neither track nor anticipate changes in their congregation base. Intent on comparing areas' population counts over time, using stable boundaries, Laidlaw succeeded in convincing the Chief of the Division of Population of the Census Bureau of his idea, and in 1910 the Census Bureau created the first ever small census districts for New York City and seven other large cities (Fig. 1). Initially, these districts were called “sanitary areas” because of health departments' role in establishing their boundaries, given the areas' relevance to planning for public health and health services (e.g., by enabling calculation and tracking of neighborhood morbidity and mortality rates). By the mid-1930s, however, as their use by non-health agencies and researchers increased, the standard terminology to describe the census-defined geographic unit became “census tracts.”2

Figure 1
figure 1

100 years of U.S. census tracts: 1906–2006.

Full size image

Health concerns nevertheless remained key to increasing the number of cities with census tracts. Of particular note were the efforts of Howard Whipple Green, a Cleveland health researcher keen on creating small-area Public Health Nursing Districts with permanent boundaries (Fig. 1).2 , pp. 11–16 With Laidlaw's advice, he was able to establish census tracts in Cleveland in 1927 and encouraged similar projects in other cities such that the number of cities with census tracts jumped to 18 in the 1930 census. In all cities, boundaries were established by local census tract committees in keeping with census-defined standards, often with advice from the local public health agencies. Whipple also published several ground-breaking works on infant mortality rates and tuberculosis rates in relation to census tract socioeconomic and racial/ethnic composition.3–6 His methodology of combining census tract and public health data was subsequently adopted by other researchers to study socioeconomic and racial/ethnic disparities in mortality plus the incidence of tuberculosis and cancer; influential analyses include Nathan's 1932 investigation of “Health Conditions in North Harlem 1923–1927,”7 Sheps and Watkins' 1947 study of “Mortality in the Socio-Economic Districts of New Haven,”8 Terris' 1948 paper on “Relation of Economic Status to Tuberculosis Mortality by Age and Sex,”9 and Cohart's 1954 article on “Socioeconomic Distribution of Stomach Cancer in New Haven.”10 Together, this research opened the door to more widespread use of census tract data in the social sciences.2,11

By the 1950s, census tracts were employed in research on health, fertility, immigration, marriage, social stratification, crime, employment, education, residential segregation, and urban ecology, as summarized in the American Statistical Association's monograph on the Golden Anniversary of Census Tracts, 1956,2 and the 1955 landmark publication of Social Area Analysis by Eshref Shevsky and Wendell Bell.11 A major 1959 review paper on “Analysis of Vital Statistics by Census Tract,” written by Elizabeth J. Coulter and Lillian Guralnick, and published in the Journal of the American Statistical Association,12 reported on results of a survey of official health agencies “on the current uses of census tract data with special references to studies of health as related to socioeconomic status.” Results indicated not only strong interest and growing use of these data but also a desire to have census tracts delimited throughout the United States. Looking ahead, the article proposed that census tract data be used to develop a consistent socioeconomic measure that could be used nationwide in order to “permit assessment of mortality rates between and within cities and serve to extend considerably an understanding of conditions producing differential mortality” as well as “help in evaluation of other factors in mortality such as air pollution, climate, or the impact of city services.”12 , pp. 738–739

The Census Tract in the 21st Century: Serving the Body Politic—and Revealing Health Inequities

Jumping ahead to the start of the 21st century, several of the goals articulated by Coulter and Guralnick have been accomplished. In 2000, census tracts were for the first time defined for the entire U.S.13 Scientific use of census tract data continues to expand in the myriad disciplines constituting what may be termed the population sciences, embracing health, social, and geographic sciences alike; listed in alphabetical order, they include: anthropology, criminology, demography, ecology, economics, education, environmental sciences, epidemiology, geography, history, political science, public health, sociology, statistics, and urban planning, to name a few. Census tracts likewise widely serve as the fundamental unit employed by myriad public agencies to create local administrative areas and are used to designate diverse policy-related entities, including “Urban Empowerment Zones,” “Medically Underserved Areas,” “Qualified Census Tracts” for the low-income housing credit, and “poverty areas” (defined as regions where 20% or more of the population is below the U.S. poverty line).13–16 Comprising a fundamental building block for public policy and public planning in health and other sectors, census tracts have thus become a key component of the public data central to documenting need and allocating resources within the body politic.

As importantly, recent research has begun to accomplish Coulter and Guralnik's hope for improving documentation and analysis of and interventions to address social inequalities in health and health care. New work, for example, has shown that the census tract poverty measure can validly be used for monitoring and analyzing U.S. socioeconomic inequalities in health, thereby overcoming the problem posed by the lack of socioeconomic data in most U.S. public health surveillance systems and hospital data systems.17–25 Building on a research legacy extending back to Laidlaw and Whipple's time, albeit with greatly improved computer technology,26 the basic method is to characterize both cases and the population from which they arise by the socioeconomic characteristics of their immediate residential area (i.e., the census tract). This is accomplished by geocoding each case's residential address, obtained from their health records, to the corresponding census tract, whose population counts (e.g., by age, gender, and race/ethnicity) in turn are obtained from U.S. census data. Rates of outcomes among persons living in, say, poorer census tracts can thus be compared to the corresponding rates among persons living in more affluent census tracts.

Importantly, new methodologic research on census tract socioeconomic measures has shown that they (1) can be used with any population, regardless of age or employment status (and hence, unlike education and income, can be used with persons still in school or not earning an income, whether because they are unemployed or retired), (2) provide estimates of socioeconomic gradients on par with those obtained with individual-level socioeconomic measures (hence indicating ecologic fallacy is not, empirically, a major concern), and (3) compared to block group socioeconomic measures, have the advantage of yielding a higher proportion of addresses geocoded to the relevant census-defined geographic unit.17–24 Additionally, they are not affected by problems unique to ZIP Code level data, such as the spatiotemporal mismatch of census and ZIP Code geographic boundaries, which has been shown to introduce sufficient bias to reverse, wrongly, the direction of the socioeconomic gradient detected using ZIP Code-level socioeconomic data.25

Currently, work using geocoding and census tract area-based socioeconomic measures not only demonstrates the existence of widespread socioeconomic disparities in health within diverse U.S. racial/ethnic groups for outcomes spanning from birth to death, but also underscores the contribution of socioeconomic inequity to racial/ethnic disparities in health.17–28 Census tract data are likewise increasingly being used to analyze inequities in access to health care, providing insight into the contribution of economic and geographic barriers to obtaining adequate care.28–30 At a time when fully 70% of the 467 U.S. public health objectives for Healthy People 2010 have no socioeconomic targets because of a lack of socioeconomic data in the relevant health data sources, the methodology of geocoding and using census tract area-based socioeconomic measures provides a promising solution to the longstanding problem of unavailable routine data at the national, state, and local level on U.S. socioeconomic health and health care inequities.17–24 Further underscoring the attractiveness of this approach, starting in 2010, the American Community Survey is scheduled to be releasing census tract data annually, based on five-year rolling estimates, thereby no longer necessitating reliance on only decennial data.31

Additionally, rising use of census tract data in the population sciences has also spurred growing awareness of—and empirical evidence documenting—the significance of social context in shaping the societal distribution of myriad outcomes, including population health.32–34 Specifically, multilevel analyses, incorporating both individual-level and census tract socioeconomic data, have shown that both levels independently, and in some cases synergistically, are associated with risk of diverse health outcomes, once again ranging from birth to death.21,32–35 The implications are that determinants of health cannot be analyzed exclusively at the individual level; context matters. As for what aspects of this context are germane, although much remains to be elucidated, studies indicate relevant aspects include at one level, the presence of harmful or beneficial neighborhood physical characteristics, themselves in part reflecting zoning ordinances (e.g., liquor stores and fast-food restaurants versus supermarkets and parks; mixed-use neighborhoods versus neighborhoods without commercial or industrial sites, etc.), and at another, the impact of concentrated poverty on risk of adverse social interactions, including violence.32–35 Tellingly, a recently released federal report, Shaping a Vision of Health Statistics for the 21st Century,36 includes as its conceptual model a figure that literally places population health—and its distribution—at the center, with determinants framed in terms of “place & time,” “context,” and “community attributes.” In this model, population health is clearly shaped by the social and physical conditions of neighborhoods, workplaces, and society at large—and the use of census tract data has played an important part in revealing these connections. For this reason, the report recommended that “all ongoing health statistics data sets should be geocoded to at least the census block group level”—which is a subdivision of the census tract—thereby enabling linkage of the health data to census-derived community characteristics, including at the census tract level.

Centenary Reflections: Census Tracts, Governance, and Social Responsibility

In summary, the census tract centenary is grounds for both celebration and reflection. Celebration, because the census tract—as a unit of geography at once spatial, social, and political—has usefully thrived for a century. And reflection, because the census tract can achieve its objective precisely because it is a national creation of democratic governance informed by local input, in accord with uniform standards, rather than driven by the imperatives of short-term political gain (per the prior use of wards to tally populations) or market distortions. Laidlaw's envisioned areas, critical for his hoped for “convenient and scientific city map system”1—have met and exceeded initial expectations. Census tracts not only provide a stable geographic unit for estimating the number and characteristics of the people and housing units located within them, but they have also expanded scientific understanding of the impact of context on the social patterning of the public's well-being, with obvious policy relevance. Census tracts thus stand as an important testament to the useful role of government—and scientists—in assuring the existence of population data necessary for understanding and improving the health and well-being of our bodies and the body politic. We do well to honor their 100th anniversary.