Become an
Expert in Spirometry

Selecting reference values for spirometry

Selecting reference values Choice affects evaluation

Selecting reference values for spirometry

There is ample choice of prediction equations. Globally, however, only a limited number is widely used: Quanjer ECCS/ERS, Hankinson, Wang, Zapletal, and Polgar [1]. They are based on linear regression, which is unsatisfactory and does not allow to cover the age range from childhood to old age. Hankinson’s equations are a step forward, because it used piecewise polynomial regression, without disjunction at the transition from adolescence to adulthood; they cover the 8-80 year age range for both whites and African Americans. The Quanjer ECCS/ERS predicted values for spirometry are too low, the Polgar and Zapletal equations provide a poor fit to data and thus lead to age-dependent bias (see here and here).  Stanojevic et al. pioneered a new approach, using smoothing splines, producing seamless predicted values (for whites only) covering the 4-80 year age range.

The Quanjer GLI-2012 equations for spirometry produced by the Global Lung Function Initiative are state of the art, covering the 3-95 year age range in whites, African Americans, North East and South East Asians. They have been endorsed by the

They are therefore the reference equations of choice. 

One should at least consider the following points when making a choice:

  1. How well do predicted values fit the data from a representative sample of healthy women and men, and boys and girls, who never smoked. You need at least 150 healthy nonsmoking females and 150 males to check whether they fit the Quanjer GLI-2012 equations. To that end software is available that handles up to 60,000 records. Alternatively use software for calculating predicted values and their LLN for individuals. Most spirometer manufacturers have implemented the equations in their software.
  2. Always use equipment that has been certified as meeting the ATS/ERS minimum requirements.
  3. Perform spirometric tests professionally, as unsatisfactory patient cooperation and suboptimal instruction and supervision, as well as lack of proper quality control, will lead to systematically underestimating true FEV1 and (F)VC.
  4. If your results indicate that in healthy subjects measured values are systematically lower or higher than predicted, check critically whether something is wrong with your equipment or how you administer the tests. However, do note that systematic differences with the Quanjer GLI-2012 equations are likely to occur due to sampling error: z-scores for data sets > 1000 subjects may differ by up to a 0.3 z-scores, and down to 150 subjects by up to 0.4 z-score.

Reference:
1. Polgar G, Promadhat V, eds. Pulmonary Function Testing in Children: Techniques and Standards. Philadelphia, Saunders, 1971.


Top of page | | | ©Philip H. Quanjer