Static Lung Volumes: Reference Values from a Latin Population of Spanish DescentCordero P.J.a · Morales P.a · Benlloch E.a · Miravet L.c · Cebrian J.b
aServicio de Neumología y bUnidad de Cuidados Intensivos, Hospital Universitario La Fe, Valencia; cServicio de Neumología, Hospital de Vinaroz, Castellón, Spain
Background and Objectives: The aim of this study was to develop a set of prediction equations and 90% confidence intervals for static lung volumes using the multibreath helium equilibration method from a sample of asymptomatic Caucasian subjects of Spanish descent. Moreover, these equations were compared with those of previous studies. Methods: Measurements of static lung volumes using techniques recommended by the American Thoracic Society and the European Community for Steel and Coal were carried out on a selected sample of 591 healthy nonsmoking volunteers (305 men and 286 women) aged 18–88 years, living in the metropolitan area of Valencia, on the east coast of Spain. Multiple regression analysis using height, age and weight as independent variables were used to provide predicted values for both sexes. These reference values were compared with other sets of prediction equations reported in the literature using an independent sample of 69 subjects (32 men and 37 women). Results: Simple linear regression equations using age, height and body weight predicted all the subdivisions of lung volumes (vital capacity, expiratory reserve volume (ERV), inspiratory capacity, functional residual capacity (FRC), residual volume (RV), total lung capacity (TLC), FRC/TLC and RV/TLC) as well as more complex equational models. The distribution of residuals fulfilled the assumptions of multiple regression analysis (independence, homoscedasticity and Gaussian distribution of residuals), except for ERV, using simple linear models. The derived equations did not differ significantly from most of the previously reported equations and were usually superior in their ability to predict the lung volumes. Conclusions: The use of the present prediction equations is recommended in the Latin population of Spanish descent and in populations with similar Caucasian characteristics.
A review of reference values for static lung volumes shows remarkable discrepancies in predicted values among different authors [1, 2]. Such differences may be ascribed to the selection of subjects, methodological and technical differences in the assessment and variability due to the inclusion of different ethnic groups.
Ideally, reference values should be derived from persons without present or previous conditions that affect ventilatory function . Most studies on lung function in healthy men and women of European descent refer to a rather small number of subjects recruited among men and women without evidence of chest disease, who were usually members of the staff and visitors as well as outpatients attending the other departments of the hospital. Current smokers and former smokers were not, as a rule, excluded [3, 4, 5]. Moreover, since most of the studies were performed at least 2 decades ago, they may not fulfil current quality criteria and may not fit present day populations due to cohort effects . Thus, prediction equations should be derived from a healthy and representative population tested by standardized technical methods and subjected to appropriate statistical analysis [1, 2]. The aim of this study was to develop a set of prediction equations and 90% confidence intervals (90% CI) for lung volumes using the multibreath helium equilibration method from a sample of asymptomatic subjects of Spanish descent and to compare the derived equations with those reported previously.
material and methods
Subjects were selected among a Latin Caucasian population of healthy volunteers (18–88 years of age and nonsmokers), living at sea level in the Valencia Metropolitan area, on the east coast of Spain. Volunteers were recruited to reflect the socioeconomic diversity of the population of Valencia by targeting advertisements to different socioeconomic groups (residential homes for the elderly, large stores, parishes and religious communities, small transportation enterprises, feeding, building, insurance and agricultural workers as well as freelance professionals). We defined nonsmokers as those who had never smoked, those who smoked less than 1 cigarette/day for less than 6 months and those who had not smoked cigarettes for more than 5 years before the onset of the study. The study protocol and rationale were explained to all patients, both verbally and in writing. Informed consent was obtained from all subjects in accordance with the requirements of the local Committee on the Ethics of Human Experimentation.
Each of the subjects filled out a self-administered modified version of the questionnaire cited by the Epidemiology Standardization Project , which formed the basis for assessing the respiratory health status. Specific sociological inquiries included in the questionnaire helped to reduce the possibility of a socioeconomical bias in the sample selection. Moreover, all subjects underwent physical examination, chest radiographic evaluation and an electrocardiogram when the subject was over 45 years old. The following items were considered exclusion criteria: the presence of thoracoabdominal wall deformities, cardiorespiratory and abdominal symptoms in the present or in the past which could lead to study limitations, neuromuscular disease and a known systemic disease, overweight above 20% , pregnancy, high risk occupational lung disease and concomitant treatments with sedative, heart or respiratory drugs.
Of the 621 subjects studied, 591 subjects, 305 males (51.6%) and 286 females (48.4%), were finally included. From these, 396 subjects were natives and 195 (33%) were immigrants from other regions of Spain, 498 subjects (83%) worked and 103 (17%) were retired; 490 subjects (83%) answered they had never smoked and 101 (17%) had stopped smoking at least 5 years before and they had smoked less than 1 cigarette/day during a maximum of 6 months. The remaining 30 (4.7%) were excluded: 12 subjects (1.9%) due to smoking history, 9 (1.4%) because of a poor cooperation when performing the spirometry, 7 (1.1%) due to a suspected clinical history of asthma, and 2 (0.3%) due to a recent infection of upper airways. Subjects were divided into six subgroups according to age: 18–29, 30–39, 40–49, 50–59, 60–69 and ≥70 years, comprising about 50 men and 50 women in each subgroup (table 1).
Table 1. Age distribution of normal subjects subdivided by sex and anthropometric features
Data were gathered from July 1990 to October 1994. Spirometric tests were performed on working days between 9 a.m. and 1 p.m. and the distribution throughout the year was homogeneous. Standing height was measured to the nearest millimeter without shoes, and weight was measured (in kg) without shoes and with light indoor clothing.
Spirometry was conducted using a 10-liter dry spirometer (Mijnhardt, Volugraph 2000), with a helium analyzer incorporated and automatic reading of the different lung volumes that meets requirements of the European Community for Steel and Coal (ECSC) and the American Thoracic Society (ATS) [1, 8]. All measurements were performed on seated subjects, sitting upright, and fitted with a nose clip. The accessory devices used (chair with a straight rigid back, nose clips and mouthpieces) were the same in all the subjects studied. The technical personnel consisted of only two technicians, each with more than 10 years of full-time experience in performing spirometry, and two physicians who were involved in the project.
The tests were carried out in the following order: (1) forced spirometry, (2) determination of static lung volumes by the multibreath helium equilibration method and (3) determination of the inspiratory vital capacity after the helium rebreathing test. Both the calibration of the devices and the performance of the maneuvers were performed following strictly the requirements recommended by the ECSC and the ATS [1, 8]. All volumes and flows were corrected for body temperature and pressure saturated conditions .
The static lung volumes were measured following the closed circuit helium dilution technique . The first step was to flush the spirometer with ambient air, to place the bell in its lowest position and close the circuit. Oxygen was added until a concentration of 25–30% was reached and the volume added recorded, then the helium meter was adjusted to zero when a stable reading was obtained. The system was fit by adding about 2 liters of air and enough amount of helium so as to reach an initial concentration of 10%. The patient breathed room air through the mouthpiece while he had the nose clip placed. The test started once a stable expiratory level at rest was reached, i.e. the functional residual capacity (FRC) position, and after a preliminary period so that the patient became accustomed to the apparatus and attained a stable breathing pattern. At the end of a normal expiration, the valve of the mouthpiece was opened to connect the patient to the spirometer. The patient breathed again consecutively in the closed circuit in which the gas mixture circulated. The CO2 was absorbed by the soda lime contained in a canister while O2 was added through a valve and the flow meter was automatically adjusted to the rate of the O2 consumption of the patient (in adults about 250–300 ml/min). As the helium (contained at the beginning in the device) mixed with the air in the lung, its concentration diminished, as could be seen in the gas analyzer. The stabilization of helium concentration was indicated by a change rate in the concentration below 0.02% during an interval of 30 s indicating the point at which balance was achieved in the whole system . In healthy subjects, the end point of the test is reached in less than 7 min. In this study, the test never took more than 10 min. Measurements were repeated for each subject with a 15-min interval until two FRC determinations did not differ more than 200 ml.
The devices were periodically calibrated according to the following criteria: (1) daily calibration of the spirometer with an accuracy metallic syringe (Sibelmed) of 3 liters. The contents of the syringe were ejected at three different speeds. The maximum differences tolerated were: ±3% or ±50 ml ; (2) weekly calibration with a dynamic signal provided by an explosive decompressor with a capacity of 4 liters (Sibelmed 122), and (3) monthly calibration of the analyzer at two levels and checking of its linearity every 3 months. In order to assess the intrasubject variability, lung volumes were measured on 20 subjects, chosen at random, twice in the same day (at 9 a.m. and at 1 p.m.) and 1 month later (at 9 a.m.). From the data obtained, the coefficient of variation for multiple measurements of the different parameters was calculated [(SD of the differences/mean value)*100].
The following prediction equations corresponding to vital capacity (VC), inspiratory capacity (IC), expiratory reserve volume (ERV), functional residual capacity (FRC), residual volume (RV) and total lung capacity (TLC) in liters and to the ratios FRC/TLC and RV/TLC were obtained. The equations were calculated separately for men and women. Age (years), height (cm), weight (kg) and body mass index (BMI; kg/m2) were taken as predictors or independent variables. Regression equations were also calculated using the following transformations of both the independent and dependent variables: (1) logarithmic (y* = ln y); (2) square root (y* = ✓ y); (3) quadratic (y* = y2); and (4) interacting variables: (y*= y1*y2).
In order to compare with other prediction equations, an independent sample of 69 subjects (32 men and 37 women) was studied between October 1994 and October 1995. All patients fulfilled the inclusion criteria mentioned above. The procedure and the technicians who made the determinations corresponded to those of the previous phase, but lung volumes were determined with a sealed-water spirometer (Collins DS II/PLUS) that meets ECSC and ATS requirements [1, 8]. In this group of patients, the values observed were compared with those obtained after applying different sets of prediction equations reported in the literature [3, 4, 5, 11, 12, 13] including those of the present study. TLC, RV, FRC and VC were analyzed.
All continuous data were tested for Gaussian distribution using the Kolmogorov-Smirnov test. The different variables were compared regarding sex by Mann-Whitney U test. Intrasubject variability was studied by applying Friedman’s test for repeated measures. Spearman’s test was applied to establish correlations between dependent and independent variables. This analysis was repeated by applying the different transformations to the variables. Prediction equations were obtained following the multiple linear regression technique by the stepwise procedure. In all cases, the goodness of fit was studied by: (1) multiple correlation coefficient (R), (2) standard error of estimate (SEE) and (3) residual behavior (independence, Gaussian distribution and homoscedasticity). For all pulmonary volumes and ratios, 90% CI were calculated as the product between the SEE of each equation times 1.64. Prediction equations of TLC, RV, FRC and VC reported by some authors [3, 4, 5, 11, 12, 13] as well as those of the present study were compared using the Wilcoxon’s matched-pairs signed-rank test. A 5% significance level was used in all statistical tests.
Table 2 shows the results obtained from a comparative analysis of the variables regarding sex. Statistically significant differences were found in all the variables except in BMI and in the FRC/TLC ratio. Therefore, prediction equations were derived separately for both sexes.
Table 2. Mean values and standard deviations of age, height, weight, BMI and lung function variables for women and men
In men, variables such as age, height, ERV and the FRC/TLC and RV/TLC ratios did not show a Gaussian distribution. All these variables, except age, became Gaussian following logarithmic transformation. In women, age, weight, RV/TLC did not show a Gaussian distribution. After applying the logarithmic transformation and square root all the variables, except age, became Gaussian.
The parameters analyzed showed no significant differences (table 3). The lowest variation coefficient was found in VC (2.3%) while the highest was in ERV (21.0%), being 14% in FRC.
Table 3. Global analysis of intrasubjectvariability and coefficients of variation
The results are shown in table 4. In some cases the correlation coefficients improved, although very slightly, after simple transformations.
Table 4. Linear simple correlation coefficients
Tables 5 and 6 show the equations obtained for men and women as well as the 90% CI of the equations obtained for each variable and 95% CI of their coefficients. The equations obtained both in men and women by introducing the variables transformed did not significantly improve the goodness of fit regarding the equations obtained with the nontransformed variables.
Table 5. Prediction equations for lung volumes in men
Table 6. Prediction equations for lung volumes in women
All the variables studied fulfilled the application conditions of the model of multiple regression except in ERV which did not fulfil the assumption of Gaussian distribution of the residuals not even when they were repeated with the transformed variables. The exclusion of the outliers did not significantly modify the prediction equations so they were finally included.
Table 7. Comparison of the values foundin VC, FRC, RV and TLC in anindependent sample of subjects withthose estimated by the different equations
The mean (standard deviation) of the differences between observed and predicted values showed remarkable discrepancies among the different authors considered in the analysis (fig. 1). Our prediction equations best estimated all the variables, no significant differences occurred between the values observed and those estimated. Then, the prediction equations that showed a better goodness of fit were those of the ECSC  and those derived by Crapo et al. ; nevertheless, both of them showed significant differences regarding the values observed in the variables VC, FRC and RV for women. The prediction equations reported by Grimby and Soderholm  showed results more distant from the values observed, being statistically significant in VC, FRC, RV and TLC for women and in RV for men and tending to their underestimation (117±886 ml for TLC in men, and 656 ± 432 ml for TLC in women).
Fig. 1. Examples of comparisons of reference values for RV from the present study and some previous studies [3, 4, 5, 11, 12, 13]. Calculations were made for a 40-year-old person. Weights for the data of Grimby and Soderholm  and the present study are average weights from actuarial tables for each height .
This study reports a set of prediction equations and 90% CIs for static lung volumes using the multibreath helium equilibration method from a sample of asymptomatic Caucasian subjects of Spanish descent. The subjects represent a wide range of heights and ages and were drawn from various occupational and social classes.
The following basic criteria required to obtain the reference values of the static lung volumes were fulfilled in our study: (1) a sufficient number of subjects for each sex; (2) a defined reference population: healthy nonsmoking individuals living in an urban area at sea level; (3) standardization of measurements and quality control procedures; (4) accurate statistical analysis, and (5) internal consistency of the reference equations proposed.
In order to be able to extend the results obtained, the sample must represent the population studied. The ideal design of a reference sample demands a randomization of the components selected prior to the application of the inclusion and exclusion criteria. Nevertheless, due to its complexity, most of the studies reported in the literature do not fulfil such randomization criteria. On the other hand, Van Ganse et al. , in a revision of the medical criteria for selecting ‘normal’ subjects, reported that the differences among different models of prediction equations which are currently used in respiratory physiology, mainly depend on the different biological characteristics of the individuals and/or the technical factors used, but not on the selection criteria of the volunteers. The patients included in our study were not randomly selected. To try to avoid this inconvenience, we considered the following aspects: (1) to diversify the sources of the volunteers included in the study, therefore diminishing the sample bias, and (2) to accurately define the individual as well as the inclusion and exclusion criteria. Individuals were initially subdivided regarding their sex in all the assessments based on the literature [3, 4, 5, 11, 12, 13] and later because of the significantly higher values in our series in men regarding women.
According to Stocks and Quanjer , published reference values are unsatisfactory by today’s standards. Discrepancies among sets of prediction equations for static lung volumes in Caucasians for either men or women reflect differences: (1) in the selection of the reference subjects , and (2) in the techniques applied and/or in the quality control of the measurements .
The selection criteria used by many authors were very different. Therefore, while Crapo et al.  only included nonsmoking volunteers, others included health care personnel, not excluding smokers. According to Clausen , the achievement of independent prediction equations for nonsmokers, former smokers and smokers has resulted in statistically significant differences among them, so that ATS recommends obtaining different reference values in nonsmokers .
Other differences are related to the ethnic group since lung volumes are lower in members of the black race and in Asiatic subjects than in the white population. The differences found among races are mainly due to the TLC which has shown to be up to 15–20% higher in Caucasians than in the Chinese or Indian population [17, 18]. FRC and RV showed no significant differences among different races. The differences in the distribution of the body fat, the thorax dimensions and different pressures produced by the respiratory muscles may partially account for the differences observed .
Physical activity during childhood, especially those which develop the shoulder girdle, e.g. rowing, swimming and diving, contribute to the development of larger lung volumes [1, 20]. There are doubts whether reference values must be adjusted in athletes. In our study, professional athletes were not included although some sportsmen and sportswomen were included.
Subjects who live at high altitude have larger lung volumes than those who live at sea level. Malik and Singh  proved that the VC was 15–18% larger in adolescents who lived in >3,500 m compared with the population with similar characteristics who lived at a lower altitude (1,500–2,000 m). Although in most studies lung volumes were determined at sea level, Goldman and Becklake  carried out their determinations at altitudes of 1,750 m, while Crapo et al.  performed them at 1,520 m, this fact could account for larger volumes.
Finally, the socioeconomic status could also affect the lung function since a low social level is usually related to other factors, e.g. unfavorable environmental conditions. Living in urban areas results in increased environmental and occupational exposure to harmful substances, higher contamination levels, higher rate of respiratory disorders, more difficulties in having medical care and malnutrition .
In order to obtain static lung volumes, we strictly followed the ECSC and the ATS recommendations [1, 8, 22]. Since the existence of circadian  and seasonal  variations has been demonstrated in the determination of lung volumes, all the studies were performed in the morning and homogeneously all over the year.
Regarding the position in which the test is performed, it has been proven that when the position is gradually changed from standing to sitting, semirecumbent and supine decubitus, there is progressive FRC and ERV decrease and IC increase [25, 26]. TLC, VC and to less extent RV diminish, though slightly. Most authors made these determinations with the subject in a sitting position, However, Boren et al.  studied their subjects in a semirecumbent position which could account for the lower volumes obtained.
Regarding the different methods used to measure static lung volumes, all of them showed a high correlation in healthy subjects . Thus, comparison of TLC determination methods by means of Pearson’s correlation coefficient ranged from 0.87 to 0.96 between body plethysmography and radiological methods [28, 29, 30], and from 0.93 to 0.94 between the radiological method and the closed circuit helium dilution technique [27, 31]; it was 0.929 between body plethysmography and closed circuit helium dilution technique , and 0.99 between the latter and the single-breath helium dilution method . The single-breath methods may slightly underestimate the lung volumes in healthy subjects. In subjects with chronic obstructive pulmonary disease, the underestimation of the true lung volume may become very large due to uneven distribution of inhaled gas . Therefore, the European Respiratory Society does not recommend it for routine use . From the equations reported, we must emphasize that in the equations obtained by Goldman and Becklake  the closed circuit using hydrogen as a test gas was used, in those of Crapo et al. , the single-breath helium dilution method, and in those of Boren et al. , the closed circuit helium dilution technique and the open circuit nitrogen dilution method. Therefore, we assume that possible differences may be partially due to the gas measurement method chosen by other authors. These differences are even more marked when the volumes obtained by gas dilution methods are compared with other methods, e.g. body plethysmography , in which FRC determination may elicit values which are even 10% higher, or radiological methods, although they are not usually employed for this purpose.
In order to control the quality of the determinations, protocols were designed to enable us to prove the accuracy and precision of the measurements. To ensure the accuracy of the measurements, the devices were periodically calibrated. To assess the precision, the intrasubject variability was assessed in a group of subjects in whom repeated determinations were performed on different days. The coefficients of variation obtained for the different variables showed no significant differences regarding those reported in literature [4, 12].
In the statistical analysis, simple linear models of prediction equations for their goodness of fit and their simplicity were used. The mathematical adjustment of our equations represented by the multiple correlation coefficient and the standard error of estimate is very similar to other models reported in the literature. Furthermore, unlike previously reported series [3, 4, 5, 12, 13], our reference sample is larger and uniform in the different age groups.
The equations obtained in the present study were compared with those from other studies [3, 4, 5, 11, 12, 13] evaluating a sample of 69 subjects, independent of the initial sample. Comparison was performed only with lung volume equations selected by the ATS in 1991 . Only the equations of Crapo et al.  strictly fulfilled all the recommendations of the ATS , since the remaining studies included smokers. The differences occurring in the criteria used in the sample selection, smoking habit, height above sea level and the type of technique used account for the differences found among different authors. The equations which showed worse fit were those of Grimby and Soderholm , mainly in women. The equations proposed by Boren et al.  for men, although they were calculated from a large sample (n = 422), showed a trend to underestimate the lung volumes. This fact may be explained by the inclusion of ‘patients’ in the reference population as well as testing in a semirecumbent position. Overall, our equations showed the best fit, and there were no significant differences in any of the variables analyzed between the estimated and observed values.
In conclusion, the present study provides original prediction equations for lung volumes using the multiple breath helium dilution technique in 591 nonsmoking Caucasian adults (from 18 to 88 years) living in Valencia at the east coast of Spain, at sea level, and with no evidence of disease. The prediction equations presented are recommended as most suitable for the Latin population of Spanish descent and for populations with similar Caucasian characteristics.
Dr. Pedro J. Cordero
Lluis Vives 8, 3
E–46700 Gandia (Spain)
Tel. +34 96 2876282, Fax +34 96 1704106
Received: Received: June 5, 1998
Accepted after revision: October 1, 1998
Number of Print Pages : 9
Number of Figures : 1, Number of Tables : 7, Number of References : 35
Respiration (International Review of Thoracic Diseases)
Founded 1944 as ‘Schweizerische Zeitschrift für Tuberkulose und Pneumonologie’ by E. Bachmann, M. Gilbert, F. Häberlin, W. Löffler, P. Steiner and E. Uehlinger, continued 1962–1967 as ‘Medicina Thoracalis’
Vol. 66, No. 3, Year 1999 (Cover Date: May-June 1999)
Journal Editor: C.T. Bolliger, Cape Town
ISSN: 0025–7931 (print), 1423–0356 (Online)
For additional information: http://www.karger.com/journals/res