Spirometry in Mixed-Race Civil War Veterans

To the Editor:

The use of race in pulmonary function interpretation is controversial. Plantation-based studies in proslavery publications first suggested racial differences in lung function. Because of its size, Benjamin Gould’s study was a landmark (1). Gould was tasked to lead a large anthropometric survey of over 1.2 million Union soldiers. Gould’s work was published in 1869 by the U.S. Sanitary Commission, a body whose still-famous members included Frederick Law Olmsted, architect of Central Park, and Louisa May Alcott, author of Little Women (1). An influential 1966 publication cited Gould’s work as evidence that observed differences in lung function between Black and White individuals reflect innate features of those races (2). Yet, Gould’s findings have been criticized for failing to account for socioeconomic status or to adequately adjust for height (3).

Gould commented that the “similarity of values for full blacks and mulattoes . . . warrants their consolidation into a single class” (1). This statement highlights a largely neglected but critical group: Americans of mixed race. Political structures starting with the Jim Crow era dichotomized racial designations into White or Black social groups. Although there is no lung function reference equation for mixed-race individuals, studies of genetic ancestry and lung function suggest that their mean vital capacity (VC) should exceed that of African American individuals, who have greater African ancestry (4). The Global Lung Initiative’s other race equation, often used for mixed-race individuals, also yields predicted values higher than those in its reference equation for African American individuals (5).

Using Gould’s work as one time point, comparison with later studies can assess for change in racial VC differences. If genetics is a primary determinant of racial VC differences, we would expect the relationship between race and lung function to be consistent over time. Alternatively, if environmental conditions are more influential, then we would expect that associations with VC evolve alongside social definitions of race. We present an interrogation of the Gould data that aims to distinguish these two possibilities. We hypothesize that because even minimal African ancestry assured low social status in the 19^th-century United States, a formal analysis of Gould’s data will support his assertion of overlapping VC for mixed-race and African American veterans.

Methods

Gould’s report offers instructions for racial designations. Question 16 of examination form EE asks, on the basis of “the appearance and statements of the subject,” investigators report “the race, unless Caucasian” (1). This is further elaborated in Question 30, in which examiners for African American or mixed-race troops were asked to “estimate . . . the proportion of black blood, such as Full Black, Mulatto, Quadroon, Octroon” (1).

Using mean height and VC data for each of Gould’s categories, we performed racially separate linear regressions of VC on height, frequency weighting each category by its number of participants. This approach appropriately accounts for variability in VC across categories by assuming variance is inversely proportional to weight. We compared the African American versus mixed-race regressions graphically and with 95% confidence intervals. Because individual-level variability in VC was not available, we tested for heteroskedasticity in residuals from the category-level data (6). For comparison, we have also included the data for the 15,124 non-Hispanic White soldiers examined by Gould. All significance tests were two sided with a preset threshold for significance of α = 0.05. We performed our estimates using Stata version 16 (StataCorp).

Results

Gould’s data include 1,631 African American and 637 mixed-race participants. Mean VC did not differ significantly between groups in either marginal or height-adjusted regression models (P > 0.15), and we saw no evidence for interaction between race and height (P = 0.71) (Figure 1). We did not see significant evidence of heteroskedasticity of residuals in the category-specific regression models.

Figure 1.

Linear regression of standing measured vital capacity versus standing height for people of “Full Black” (African American), “White” (non-Hispanic White), and “Mulatto” (mixed race) race, from Gould’s Table VI. Each dot represents one of the category values given by Gould in Table VI. CI = confidence interval.

Discussion

In our reanalysis of the historic data of 1,631 African American and 637 mixed-race Civil War veterans, we found no evidence for mean differences in VC. This result concords with Gould’s original conclusion but contradicts modern spirometric studies of mixed-race individuals, which find a consistent inverse association between the degree of African ancestry and VC (4). Key to understanding this result is understanding the genetics versus social status of mixed-race individuals around the time of the Civil War. As hypothesized, despite these two groups likely having distinct genetic ancestry at the time of data collection, there remained no difference in VC, given that they remained in the same social class.

Genetic ancestry studies suggest that individuals identified as African American were unlikely to have the same proportion of African ancestry as those designated as being of mixed race (4, 7). The overwhelming majority of mixed-race individuals were the product of sexual violence against enslaved females. As a result of this pattern and binary American racial classification, in each generation since the transatlantic slave trade, the mean proportion of whole-genome African ancestry in African American individuals has decreased (7). Before this system’s implementation, however, European admixture would have concentrated among those who self-identified as mixed race and had identified non-Hispanic White forebearers. Demonstratively, Sally Hemings, who is famous for her association with Thomas Jefferson, was a third-generation product of forced pairings between slaver and enslaved person.

By contrast, the low social status of African American and mixed-race individuals was very much apparent. According to the rule of partis sequitur ventrum, these children were considered enslaved from birth. Compared with other slave societies, that in the United States was notably less tolerant of racial admixture (6), and mixed-race individuals’ status was not meaningfully different from that of darker-skinned African American individuals (8). Even work explaining later advantages enjoyed by mixed-race individuals demonstrates that these were initially quite small in absolute terms and by comparison with the status of non-Hispanic White individuals (9). Given their similarly poor treatment and high levels of deprivation, it is little surprise that their VC would be similarly low. Elsewhere, Gould’s comment that accelerated height loss among African American veterans may be due to “the condition of the Southern negroes” similarly suggests the importance of social status (1).

Multiple limitations should be appreciated. Gould’s raw data no longer exist for direct interrogation. Age was not recorded for spirometry subjects and could not be used in our regression equations. Investigators from Gould’s study afterward raised concerns about racial misclassification of participants, which could also have influenced our results, and the role of investigator bias influencing measurements is unknown (10). Although racial admixture was not quantified objectively, investigators estimated it on the basis of physical appearance, and we know skin pigmentation is highly correlated to genomic ancestry (11). However, a subjective assessment of skin color is an imperfect assessment of the degree of African ancestry. It is therefore wholly possible that mixed-race individuals simultaneously inherited genes yielding lighter complexion and others that we would associate with lower lung function among those with African ancestry. This study’s quality control techniques for spirometry are not well described. Despite these multiple caveats, it is interesting to note that another contemporaneous study of Civil War veterans similarly failed to identify differences in lung function across racial groups, undermining the notion of important genetic differences in this regard and highlighting the complexity of environmental interacting factors (12).

Our reanalysis of historic data from 2,302 mixed-race and African American Union Army veterans of the U.S. Civil War revealed that both racial groups had highly similar VC. Although caveated by the dataset’s limitations, these results suggest that racial VC differences may be as much social as genetic in origin. Although modern studies have also attempted to adjust for socioeconomic status when considering the role of genetics, socioeconomic status is a multifaceted concept that is notoriously difficult to capture (13). Fading racial dogma has produced an asymmetric reduction in the structural racism needed to support it. This has allowed mixed-race identification to increase and has yielded a gradient of life outcomes midway between those observed among White- and African American–identifying families, which parallels the well-known finding in lung function and could reflect social differences (14, 15). Contrasting a dataset with a context unique from modern-day studies may thus have helped better highlight the role of socioeconomics. Furthermore, though estimated using genetic markers, ancestry strongly links to the environmental, historic, demographic, experiential (racism), and cultural factors associated with race (16). In reconsidering race and ancestry quantification, we must ensure that incuriosity does not lend socioeconomic distinctions a biological sheen.