INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Most pulmonary function laboratories in the United States use prediction equations for normal values based predominantly on populations of European descent (1). A large number of studies have demonstrated, however, that ethnic groups differ in pulmonary function (2), and therefore prediction equations based on European populations may not be accurate for all subjects. Ethnic differences in lung function have been well demonstrated between blacks of African descent and whites of European descent (2, 3, 5). This has been attributed, at least in part, to anthropometric differences between these ethnic groups, with whites having larger chest volumes and shorter leg length (i.e., larger trunk-to-leg ratio) at a given height (4, 9). Accordingly, these studies have shown that whites have lung volumes 10-15% higher than blacks at a given standing height, and a "correction factor" of 10-15% is used by many pulmonary laboratories to reduce white-based predicted values for black subjects (1).

Ethnic differences in lung function have also been suggested in many other ethnic groups (2, 4, 12) including Asians (3, 7, 8, 14). The question of lung function in Asian-Americans (AsAs) is of increasing importance, as this segment of the population in the United States has approximately doubled in size over the last decade, and is projected to double in size again by the year 2040 (13, 17). However, methodologies used in previous studies have not been uniform, and most of these studies suffer from poor definitions of ethnic groups, variability in health and smoking status of subjects, and the testing of subjects with different equipment and techniques. Most studies have concluded that AsAs have spirometry values that are lower than those of European-Americans (EAs) but higher than those of African-Americans (AfAs). No consistent recommendations have been made regarding whether predicted values for AsAs should have a "correction factor" similar to that used for AfAs, and if so how big the correction factor should be.

Only a few studies have addressed differences in diffusing capacity between AfAs and EAs (9, 18, 19), finding that AfA values for single-breath diffusion capacity of the lung for CO (DL_CO) were, on average, 5% lower than those of EAs. No studies have addressed differences in gas transfer between AsAs and EAs.

Considering the diversity found in prediction equations derived from populations of the same ethnicity in separate locations (using dissimilar equipment and technicians), it seems suitable to evaluate ethnic factors only when measurements are made with the same equipment and by the same technicians. The present study was undertaken to investigate ethnic differences in spirometry and gas transfer between AsAs and EAs in a young, nonsmoking population, using the same equipment for all subjects.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The research protocol was approved by the Institutional Review Board. House officers and medical students aged 22-33 yr volunteered for the study, and each gave informed consent before participating. A narrow age range of subjects was selected in order to minimize the effect of age on the respiratory variables. Subjects filled out a questionnaire detailing birthdate, birthplace, ethnic background, length of residency in the United States, smoking history, history or symptoms of lung or heart disease, and level of physical activity. Ethnic background questions entailed recording country of birth of parents and grandparents. All subjects were either U.S. citizens or residents. Subjects were defined as Asian if both parents and all grandparents were born in eastern Asia, or as European if both parents and all grandparents were born in either the United States or a country in western or eastern Europe. Subjects from India or Pakistan were not included in the study. Level of fitness was evaluated by questions asking the average number of hours of exercise performed by each subject per week, with responses grouped into four categories (< 1, 1-3, 4-6, and > 6 h/ wk). Only lifetime nonsmokers without a history of significant lung or heart disease were included in the study.

Because of the difficulties inherent in studying subject groups classified by ethnicity (20), and in particular the difficulties in choosing labels for these groups, the following details how ethnic groups are described in this article. For our own subjects, defined as Asian-American (AsA) or European-American (EA) according to birthplace of parents and grandparents as described above, subjects are referred to as AsA or EA. When referring to ethnic groups from other studies, the identifying terms used in those studies are used, with the exception of the term "Negroes," which was changed to "blacks." Most often, in previous studies, subjects were classified on the basis of skin color, and therefore redefinition of these groups using different labels (i.e., changing the term "blacks" to "African-Americans") may be misleading and untrue.

Measurements

The standing height (Ht) and weight of each subject were measured without shoes. Sitting height (Sh) was measured with the subject seated erect on a stool of fixed height adjacent to a tape measure fixed to the wall. Weight was recorded to the nearest 0.1 kg, and heights were recorded to the nearest 0.5 cm. Before performing spirometry and DL_CO maneuvers, a small amount of blood was obtained from each subject by venipuncture and assayed for hemoglobin (Hb) concentration, using a cooximeter (model 482; Instrumentation Laboratories, Lexington, MA).

The same pulmonary function testing equipment was used to measure all subjects (model 2200; Sensormedics, Yorba Linda, CA), and all tests were performed by the same experienced technician. The equipment was calibrated daily with a 3-L syringe according to manufacturer recommendations. Room temperature was maintained at 21° C, and atmospheric pressure was measured daily. Spirometry results were corrected to BTPS, using a computer algorithm.

All spirometry measurements were performed in accordance with American Thoracic Society (ATS) guidelines (23). Subjects performed maneuvers in the seated position while wearing noseclips. Spirograms were repeated until three acceptable tests were obtained. For FVC and FEV₁, testing was stopped if the largest and second largest values were within 0.2 L of each other. Otherwise, further spirograms were obtained until these criteria were satisfied (23). The FEV₁ and FVC were each recorded as the single best result from all acceptable curves. FEV₁/FVC was recorded from the maneuver with the highest sum of FEV₁ and FVC.

Single-breath DL_CO maneuvers were performed with a gas mixture containing 0.3% CO, 0.3% methane, and 21% oxygen according to standard technique (24). Maneuvers were repeated until three acceptable tests were obtained. Expired CO and methane curves were examined after each maneuver, and washout volume was adjusted manually as necessary to assure dead space clearance (24, 25). The average of the two closest maneuvers was recorded. To confirm that alveolar volume (VA') measurements correlated with total lung capacity (TLC) in this population, five subjects also had TLC measured by nitrogen washout technique.

Data Analysis

The data were analyzed by multiple regression techniques (26). Dependent variables considered were FEV₁, FVC, FEV₁/FVC, VA', DL_CO, and DL_CO/VA'. These were regressed against the independent variables of Ht, age, sex, ethnicity, and Hb, with ethnicity and sex coded as indicator variables. Weight was not used as a variable because of multicolinearity with Ht (i.e., because of the absence of obesity in this subject sample) (3, 4). In the first step, forward and backward stepwise regression models were formed without interaction terms. The model that had each included variable significant at the p  0.05 level, the highest explained variance (r²), and the lowest standard error of the estimate (SEE) was determined, and these independent variables were used for further analysis. The models were then reestimated by including first-order interaction terms for sex and ethnicity (if found to be significant) with the other independent variables. Interaction terms that contributed significantly to the model (p  0.05) were retained in the model only if their addition improved the explained variance and reduced the SEE of the model. Simpler models were chosen if addition of further variables to the equation resulted in only trivial improvement in r² and SEE. Models were estimated separately with sitting height (Sh) in place of Ht to ascertain if use of Sh resulted in higher r² and/or eliminated observed ethnic differences in dependent variables. The effects of level of activity and length of residency in the United States, and their interactions with race, were evaluated by two-way analysis of variance (ANOVA). Interval regression was used to estimate and compare mean hours of exercise per week (27). The Wilcoxon rank sum test was used to compare the means of continuous variables, and the Fisher exact test was used to compare categorical variables (26). Statistical analyses were performed with a statistical software package (Stata, College Station, TX).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Eighty subjects were enrolled in the study. The parents and grandparents of all Asian subjects were also Asian. Their specific countries of origin were China (11 male, 11 female), Korea (3 male, 4 female), Vietnam (2 male, 3 female), Japan (2 male, 1 female), and the Philippines (1 male). One female subject had a Japanese and a Korean parent, and one male subject had a Japanese and a Chinese parent. The baseline characteristics of the study group are shown in Table 1. None of the subjects were obese. Although the EAs tended to be taller and heavier, there were no significant differences between races in age, Ht, Sh, weight, or Hb. EAs exercised 2.0 h/wk more than AsAs, and had lived in the United States an average of 2.4 yr longer, with both of these values reaching significance.

In all regression models, the coefficient for age was nonsignificant, likely because of the narrow age range of the population studied; age is therefore not a variable in any of the regression equations. Neither length of residency in the United States nor hours of exercise per week were significant predictors of lung function in ANOVA models, and neither variable had significant interaction with ethnicity, indicating that these variables do not contribute significantly to the lung function variables considered, regardless of ethnicity.

Flow and Volume Measurements

There were significant differences in spirometry values between the two ethnic groups. For FVC, significant regression variables were Ht, ethnicity, and sex (Table 2). Stated differently, FVC differed significantly between the two ethnic groups, even when controlling for height and sex. As seen in Figure 1 and Table 2, the regression lines for EA and AsA differed significantly in both slope (p = 0.016) and y intercept (p = 0.029), while the regression lines for males and females differed in slope alone (p = 0.001). The resulting regression model had r² = 0.80 and standard error of the estimate (SEE) = 0.44 L. 

View larger version (14K): [in this window] [in a new window]   Figure 1.   FVC versus height by sex and ethnicity for European-American males (open squares, solid line), Asian-American males (solid squares, dotted line), European-American females (open circles, dashed line), and Asian-American females (closed circles, dotted- dashed line).

For FEV₁, significant regression variables were again Ht, ethnicity, and sex (Table 2), with regression lines for the two ethnic groups differing significantly in y intercept (p = 0.007), and regression lines for males and females also differing in y intercept (p = 0.002; Figure 2). The regression model had r² = 0.72 and SEE = 0.38. 

View larger version (13K): [in this window] [in a new window]   Figure 2.   FEV1 versus height by sex and ethnicity for European-American males (open squares, solid line), Asian-American males (solid squares, dotted line), European-American females (open circles, dashed line), and Asian-American females (closed circles, dotted- dashed line).

There were no significant differences in FEV₁/FVC between the two ethnic groups or sexes, as this variable regressed significantly only with Ht (p < 0.001). The coefficient for the Ht variable showed a decrease of 0.3% in FEV₁/FVC for each 1-cm increase in height.

Reestimation of these regression models using Sh instead of Ht resulted in lower explained variance and higher SEE in all cases. In addition, the use of Sh in the models did not eliminate the ethnic differences in the variables. This is not surprising, as the mean Sh/Ht ratios did not differ significantly in the two ethnic groups (Table 1), suggesting that, in this population of subjects, differences in leg and trunk length are not responsible for the observed differences in lung function.

Alveolar Volume (VA') and TLC Measurements

Similar to spirometry volume and flow measurements, measurement of alveolar volume by inert gas dilution during the DL_CO maneuver revealed significant differences between the two ethnic groups. Significant regression variables for VA' were Ht, ethnicity (significant differences in y intercept, p = 0.025), and sex (significant difference in y intercept, p = 0.006) (Figure 3, Table 2).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Alveolar volume versus height for European-American males (open squares, solid line), Asian-American males (solid squares, dotted line), European-American females (open circles, dashed line), and Asian-American females (closed circles, dotted-dashed line).

To compare VA' from the present study, measured during the single breath DL_CO maneuver, with TLC from other studies, which was typically measured by inert gas washout technique, the reliability of VA' in predicting TLC in our population was determined by measuring TLC in five subjects (three female, two male) by nitrogen washout. For these subjects, the TLC/VA' (mean ± SD) was 1.07 ± 0.03. Therefore, TLC values reported from the present study are VA' values adjusted upward by a factor of 1.07.

Gas Transfer Measurements

Despite significant ethnic differences in spirometry variables, there were no significant differences in DL_CO between the two ethnic groups (p = 0.47). Significant variables in estimation models for DL_CO were Ht and sex (Table 2). Regression lines for males and females differed significantly in slope (p < 0.001), with a resulting r² = 0.75 and SEE = 3.73 (Figure 4). Of note, the regression lines for DL_CO pass through the origin.

View larger version (12K): [in this window] [in a new window]   Figure 4.   DLCO versus height by sex (males = dashed line, females = solid line) for European-American males (open squares), Asian-American males (closed squares), European-American females (open circles), and Asian-American females (closed circles).

In regression models for the ratio of DL_CO with VA' (DL_CO/ VA'), the only significant independent variable was Hb (p  0.001, Figure 5). There were no significant differences for DL_CO/VA' between EAs and AsAs, or between the sexes. If DL_CO/VA' is "corrected" for Hb by dividing the DL_CO/VA' of each subject by their Hb (DL_CO/VA'/Hb), the resulting regression line is flat, with overall F test showing no difference from a slope of zero (p = 0.06, Figure 6). Thus, in our subjects, DL_CO/VA'/Hb is a "constant" value of 0.37 ml CO/min/mm Hg/L/g/dL (SEE = 0.01), and does not vary with Ht, age, ethnicity, or sex.

View larger version (10K): [in this window] [in a new window]   Figure 5.   DLCO/alveolar volume versus hemoglobin for European-American males (open squares), Asian-American males (closed squares), European-American females (open circles), and Asian-American females (closed circles).

View larger version (11K): [in this window] [in a new window]   Figure 6.   DLCO/alveolar volume/hemoglobin versus height for European-American males (open squares), Asian-American males (closed squares), European-American females (open circles), and Asian-American females (closed circles).

Magnitude of Ethnic Differences and Comparison with Other Studies

To compare the present study with those previously reported in the literature, the prediction equations from several widely accepted studies based on subjects of European extraction (4, 28), as well as from the present study, were used to calculate values for FVC, FEV₁, TLC, and DL_CO based on characteristics of the subjects studied here. These values, calculated at the mean Ht and age of the subjects in the present study, are shown in Table 3. In general, the predicted values for European males and females in the present study showed excellent agreement with previous studies. In particular, the predicted values for FVC and FEV₁ in EA males agree well with the other studies listed, while for females, the present study agrees well with Morris and coworkers (30) and Crapo and colleagues (31) for FVC and FEV₁. Female values for FVC and FEV₁ tended to be higher than those of Knudson and co-workers (34). TLC values for EA agree closely with those of Goldman and Becklake (28) and Crapo and coworkers (33) for both males and females. FEV₁/VC values for both males and females agreed closely with those of Crapo and colleagues (31) (data not shown). Predicted DL_CO values from the present study agree closely with Cotes (4) and Miller and coworkers for males, but are lower then other predicted values for other series.

As can be seen from Table 3, AsA values for lung volumes and FEV₁ in the present study averaged 7% (range, 4-11%) lower than EA values.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this population of young, healthy, nonsmoking subjects, the significantly lower FVC, FEV₁, and VA' in AsAs than EAs for the same height, independent of level of fitness or length of residence in the United States, represents a true physiologic difference between these two ethnic groups. However, at a given height, DL_CO, DL_CO/VA', or DL_CO/VA'/Hb were not different between the two ethnic groups.

AfAs have consistently been found to have 10-15% lower spirometric values than whites (2, 3, 5). Damon (6), who found FVC and FEV₁ values (determined using the same equipment) to be 13% lower in black than white soldiers, concluded that the 3% shorter trunks and 20% smaller chest expansion of blacks did not adequately explain their reduced lung size. A later study by Rossiter and Weill (9), using common equipment, found 13% lower lung function volumes and 8% lower DL_CO in blacks than whites. Using anthropometric data from a Rhodesian study (36), they concluded that it was appropriate to use scaling factors. While not all studies have recommended the use of scaling factors (10), both the American Thoracic Society and American Medical Association (37, 38) recommend a reduction of 12% in predicted values for disability evaluation of African-Americans.

Most studies evaluating pulmonary function differences between Asians and EAs measured lung function in Asian subjects and compared the resultant prediction equations with those previously derived from other studies of EAs. Although Asian values were lower than EA values (14, 39, 40), measurements performed with differing equipment and with differing techniques make quantitative comparisons problematic.

Two studies have made direct comparisons between AfAs, AsAs, and EAs. Oscherwitz and coworkers (7) found highest values of FVC and FEV₁ in EAs, intermediate values in Asians, and lowest values in blacks for the same height, age, and sex. Seltzer and colleagues (8), in a population of more than 65,000 subjects (82% whites, 14% blacks, and 4% Asians), found FVC and FEV₁ to be highest in EAs and lowest in blacks. Both studies are important but suffer from poor definitions of ethnic groups, the confounding factor of smoking in a large proportion of the populations, the relatively small number of AsAs in each study, and the use of multiple types of spirometers in each study.

The present study differs from these in that a young, healthy, nonsmoking population of clearly established ethnic background were all measured with the same pulmonary function equipment. In addition, the decline in pulmonary function normally seen with aging was not a factor. As in other studies (11, 34) we found little variation in lung function in this age range (22- 33 yr), eliminating potentially confounding variables.

The definition of ethnic groups in studies of this type is problematic because, even within defined ethnic groups, there may be wide variations in cultural, social, and economic factors (20). This has been particularly problematic in most of the previous studies of ethnic differences in pulmonary function, in which ethnic groups were determined on the basis of skin color alone. In this study, an attempt was made to carefully determine the ethnic lineage of subjects by identifying the birthplace of parents and grandparents and classifying subjects as "Asian" and "European" on the basis of this information. It is recognized that combining subjects from widely diverse eastern Asian and European cultures into groups is artificial, and that further variation between groups might be seen if examined closely (i.e., between Chinese and Korean subjects). However, the intention of the study was to evaluate differences between ethnic groups in the way that these groups are commonly perceived and clinically evaluated in our society.

The magnitude of reduction in FVC, FEV₁, and TLC of approximately 6-7% seen in AsA males and females in our study is similar to that of other similar studies (8, 14, 15), and supports the use of a scaling factor for AsA subjects, using prediction equations based on EA subjects. This 6-7% scaling factor is less than the 12-15% factor generally accepted for AfAs (2- 4, 6, 9, 11, 37, 38).

An important finding was the lack of a significant difference of DL_CO between AsAs and EAs, as this measure has not previously been reported. DL_CO has been found to be 5-8% lower in AfAs than EAs (9, 18, 19), a reduction approximately half of the lung volume change. Considering the approximate 7% reduction in AsA lung volumes from EA values, we might have expected a DL_CO reduction in AsA of approximately 3- 4%. Possibly, our relatively small sample size resulted in insufficient power to detect a significant difference in DL_CO between AsAs and EAs.

While DL_CO differed significantly between males and females (Figure 4), the DL_CO/VA' did not differ between sexes or ethnicity. Although VA' was lower in AsA than EA and DL_CO was not significantly lower, DL_CO/VA' was not significantly higher in AsAs than EAs, as might be expected. Again, as is the case for DL_CO, this study may have lacked sufficient power to detect a significant difference in DL_CO/VA' between the ethnic groups. Indeed, when DL_CO is corrected for Hb level (i.e., DL_CO/VA'/Hb), the resulting regression line is flat (Figure 6), suggesting that this physiologically important measure of gas transfer is constant across sexes and ethnic groups at this age. As the "capillarity" of the lung, reflected in the DL_CO/VA', was similar in ethnic groups, hemoglobin appears to be the true determinant of gas transfer in the healthy lung.

Our study population was intentionally narrow in terms of age as well as socioeconomic status in order to minimize the effects of these variables on lung function and allow a more direct evaluation of differences due to ethnicity. Further studies might be performed to broaden these findings to wider age groups as well as groups of differing socioeconomic status.

Although the number of subjects in this study was limited to 80, the fact that age, obesity, level of fitness, and tobacco use could be ignored as variables made it simpler to identify and quantify differences between sex and ethnicity. We found that FEV₁, VC, and TLC values are approximately 7% less for AsAs than EAs for the same height, while DL_CO values average only 2% less for AsAs than EAs, an insignificant difference. The study emphasizes the key importance of hemoglobin in gas transfer measures, as the DL_CO/VA'/Hb is 0.37 ± 0.01 ml CO/min/mm Hg/L/g/dL regardless of ethnicity, sex, or height in this young but mature healthy population.