Deriving ‘all age’ reference values for spirometry
Let us start by looking at FEV1 as a function of age in Caucasian females, aged 3-95 years. The FEV1 increases until about age 20, and then declines steadily. There is very considerable scatter about the mean, with a maximum at 20-40 year. Although the scatter diminishes with age in adults, it does not do so in proportion to the predicted value.
Obviously we cannot satisfactorily describe the relation between FEV1, age and height over the entire age range with a simple linear relationship of the form
log(Y) = a + b•log(height) + c•log(age)
However, let us do it anyway and see how the predicted values relate to the measured ones. To this end we have first derived the relationship of height with age. We then substituted age, and height for age, into the regression equation to obtain a continuous line.
As expected, the fit is very poor (right figure). It is obvious that we must make allowance for the rise and fall in FEV1 by making an age-specific adjustment. This can be done by adding an age-specific spline. Think of it as an age-specific correction to the above linear equation. The equation then has the following form:
log(Y) = a + b•log(height) + c•log(age) + spline
For those who take a keen interest in how this is achieved in practice, consult 'GAMLSS in action'.
The figure on the left shows the spline for the predicted value of FEV1 in females as a function of age. It discloses that at about age 5 yr the predicted value based upon age and height to needs to be lowered by about 20%. At about ages 11 and 50 years no correction is required. At age 20 years the predicted value needs to be increased by about 20%, and at age 90 years diminished by about 40%.
Now let us see how this works out in practice. The regression line (black, figure on the right) now looks realistic.
Oviously, part of the scatter shown is due to between subject differences in standing height. In the data on Caucasian females, height in 3-20 year old subjects ranged between 85 and 188 cm, a twofold range. In adults the height range was 123-192 cm; obviously the shortest subjects are outliers compared to the rest. In the adult women average height is 164.6 cm at age 20, and 152.6 cm at age 90, a 12 cm difference. These age-related height differences have two causes:
- A tendency to become shorter in adulthood occurs among all races and both sexes. Height loss is related to aging changes in the bones.
- Secular trends: a person born, say 60 years ago, is likely to have been shorter at age 20 than someone born 20 years ago. This relates to changes in socio-economic conditions.
It is therefore worthwhile to explore how much of the scatter is due to differences in height (next).
|Ref. 1 - Fixed correction factor for ethnic group|
|The following studies suggest that a fixed correction factor for an ethnic group is not based on a firm scientific basis.|
|1||Schoenberg JB, Beck GJ, Bouhuys A. Growth and decay of pulmonary function in healthy blacks and whites. Respir Physiol 1978; 33: 367-393. PubMed|
|2||Corey PN, Ashley MJ, Chan-Yeung M. Racial differences in lung function: search for proportional relationships. J Occup Med 1979; 21: 395-398. PubMed|
|3||White NW, Hanley JH, Lalloo UG, Becklake MR. Review and analysis of variation between spirometric values reported in 29 studies of healthy African adults. Am J Respir Crit Care Med 1994; 150: 348-355. PubMed|
|Ref. 2 - Review of ethnic differences|
|Cotes JE, Chinn D, Miller MR. Lung Function: Physiology, Measurement and Application in Medicine. Blackwell Scientific Publications 2006.|
|Ref. 3 - Leg length and sitting height|
|Negroid subjects in general have longer legs for trunk length:|
|1||Verghese KP, Scott RB, Teixeirea G, Ferguson AD. Studies in growth and development. XII. Physical growth of northern American Negro children. Paediatrics 1966; 44: 243-247. PubMed|
|2||Van de Wal BW, Erasmus LD, Hechter R. Stem and standing heights in Bantu and white South Africans: their significance in relation to pulmonary function values. S Afr J Lab Clin Med 1971; 45 (suppl.): 568-570. PubMed|
|3||Rossiter CE, Weill H. Ethnic differences in lung function: evidence for proportional difference. Intern J Epidem 1974; 3: 55-61. Manuscript|
|4||Quanjer PH, Kubota M, Kobayashi H, Omori H, Tatsumi K, Kanazawa M, Stanojevic S, Stocks J, Cole TJ. Secular changes in relative leg length confound height-based spirometric reference values. Chest 2015; 147(3): 792-797. Manuscript|
|5||Quanjer PH, Capderou A, Mazicioglu MM, et al. All-age relationship between arm span and height in different ethnic groups. Eur Respir J 2014; 44: 905–912. PubMed|
|Standing height leads to a slightly higher explained variance than sitting height:|
|1||Damon A. Negro-white differences in pulmonary function. Vital capacity, timed vital capacity and expiratory flow rates. Human Biol 1966; 38: 380-393. PubMed|
|2||Van de Wal: see above.|
|Sitting height cannot be measured with the same accuracy as standing height, so that it is not practically useful in prediction equations:|
|Tanner JM. Growth at adolescence. Ed. 2. Blackwell, Oxford, 1978.|
|Ref. 4 - Differences in body build|
|1||Bibi H, Goldsmith JR, Vardi H. Racial or ethnic variation in spirometric lung function norms. Recommendations based on study of Ethiopian Jews. Chest 1988; 93: 1026-1030. PubMed|
|2||Massey DG, Fournier-Massey G. Japanese-Americans pulmonary reference values: influence of environment on anthropology and physiology. Env Res 1986; 39: 418-433. PubMed|
|Ref. 5 - Genetic and ethnic factors|
|1||Reed TE. Caucasian genes in American Negroes. Science 1969: 165: 762-768. PubMed|
|2||Repapi E, Sayers I, Wain LV, et al. Genome-wide association study identifies five loci associated with lung function. Nat Genet 2010; 42(1): 36-44. PubMed|
|3||Kumar R, Seibold MA, Aldrich MC, et al. Genetic ancestry in lung-function predictions. N Engl J Med 2010; 363(4): 321-330. PubMed||4||Quanjer PH. Lung function, genetics and socioeconomic conditions. Eur Respir J 2015; 45(6): 1529-1533. Manuscript|