Adult reference values for spirometry, body plethysmography and diffusing capacity adjusted for sex, age, weight, height or arm span - the Danish lung function material (DALFUMAT) study

ABSTRACT

Objectives

The study aimed to develop new national reference values for dynamic and static lung volumes, as well as diffusing capacity for carbon monoxide, measured in the same participants. These values were compared with existing reference values from GLI and ECCS. Additional aims included the development of lung function reference values based on arm span, the establishment of post-bronchodilator reference values for spirometry, the creation of reference values for bronchodilator response, and the identification of significant bronchodilator responses using z-scores. Furthermore, the study sought to observe and enhance the quality of Danish lung function laboratories.

Methods

Spirometry, body plethysmography, single-breath diffusion capacity, and bronchodilator testing were performed on 908 healthy non-smokers aged 18–97 years, selected from municipalities served by 13 participating Danish centres. Strict quality control was maintained in accordance with ATS/ERS standards. Sex, age, age squared, weight, and height (or arm span) were used as independent variables in the multiple regression equations. The resulting lung function data were compared with predicted values from GLI and ECCS.

Results

Sex-specific reference equations were established for 29 lung function parameters. Additionally, reference values based on arm span and post-bronchodilator spirometry were calculated, and bronchodilator response was established using four different endpoints, including z-scores. A substantial proportion of the new reference values differed clinically significantly from those predicted by GLI and ECCS equations. These discrepancies were more frequent in females than males, more pronounced for ECCS than GLI, and more common for flow and volume parameters than for diffusion capacity and ratios.

Conclusion

Given that many of the new reference values differ clinically significantly from those provided from GLI and ECCS, we recommend the nationwide adoption of the new DALFUMAT reference values.

Introduction

Lung function testing plays a crucial role in assessing respiratory health, diagnosing lung diseases, and monitoring treatment responses. Valid prediction values are necessary to determine whether a subject’s lung function falls within the reference range and, if not, to assess severity of impairment [1]. Accurate interpretation of lung function tests relies on robust reference values that account for variations in subjects’ size, sex, and age. Variability in measurement techniques and equipment can affect these reference values. Therefore, standardisation is crucial, and the European Respiratory Society (ERS) and the American Thoracic Society (ATS) have recently provided updated technical guidelines for the standardisation of lung function testing [2–5].

Numerous sources for reference values have been published over the years, particularly for spirometry, such as The National Health and Nutrition Examination Survey (NHANES) [6] from the US and Løkke [7] from Denmark. However, fewer publications address the measurement of static lung volumes and diffusing capacity for carbon monoxide (DLco). The predominantly used reference values in Europe are from ERS in 1993. These values are based on the 1983 reference values from the Working Party of European Community for Coal and Steel (ECCS) [8]. They encompass routine lung function testing with spirometry, static lung volumes, and DLco, based on compilations of prediction equations from various studies on adults published between 1948 and 1983 [8–10].

More recently, the Global Lung Function Initiative (GLI) [11] published compilation reference sets, which are the largest resources for lung function reference values based on tests conducted on diverse populations worldwide using different equipment. GLI first produced standardised reference equations for spirometry [12], followed by DLco [13] with an erratum in 2020 [14], and more recently for static lung volumes [15].

The reference values from ERS [9,10] and GLI [12–15] are not based on the same healthy subjects undergoing testing with both spirometry, static lung volumes and DLco. Instead, reference values are compiled from different subjects tested with spirometry (ERS, max. n = 11,000; GLI, max. n = 100,000), others with DLco (ERS, max. n = 960; GLI, max. n = 9,710), and others with static lung volumes (ERS, max. n = 2,960; GLI, max. n = 7,190). Additionally, different equipments and methodologies have been used for testing static volumes, such as both Helium and Nitrogen dilution techniques and plethysmography [15].

The age ranges tested also differ; for example, ERS reference values cover ages 25–70 years [9,10], while GLI reference values for spirometry cover ages 3–98 years [12], DLco 5–85 years [13], with an erratum [14], and static volumes 5–80 years [15].

The ERS/ATS now recommend using GLI or locally validated reference values for the interpretation of lung function tests [16]. Predicted values from GLI have been shown to differ significantly from local reference values in some spirometry studies, such as those from Sweden, the US, and Tunisia [17–19], but may fit well with other populations, such as those from Australia, Algeria, and Norway [20–22]. There are limited data on how well the newer predictions equations from GLI for DLco and static volumes fit different populations, but a recent study from Austria showed a poor fit regarding static volumes [23].

The objectives of this study were to create new national reference values for dynamic and static lung volumes, as well as diffusing capacity, all measured in the same group of healthy participants. These new reference values were validated against existing standards from the GLI [12–15], ECCS [9,10], and Løkke [7]. In addition, the study aimed to develop lung function reference values based on arm span measurements. Reference limits for bronchodilator response were established, and a significant bronchodilator response was defined based on z-scores. Furthermore, the study sought to generate post-bronchodilator reference values for spirometry. Finally, the study aimed to observe and improve the quality of Danish lung function laboratories.

Methods & materials

Study design

A cross-sectional study of randomly selected, never-smoking Danes aged 18 years and older was conducted. A never-smoker was defined as having a total tobacco consumption of less than the equivalent of one pack-year.

Inclusion criteria

• 18 years or older

• Never-smoker

• Both parents of European origin

• Residing in a municipality where one of the participating centres is located

Exclusion criteria

• Individuals with a range of heart and lung diseases, as indicated by ICD-10 codes (see Supplement 5S), were excluded by cross-referencing with the Danish National Patient Register

• Individuals with respiratory symptoms or infections (e.g. cough, pneumonia) in the 30 days prior to the examination day were excluded

• Known pregnancy

Recruiting test subjects

A sample was drawn from the Danish Civil Registration System to ensure that the test subjects were representative of the Danish population in terms of geography, age, and sex. The selected individuals received a letter containing information about the purpose of the project and a request to participate. Additionally, the letter included: a questionnaire regarding the inclusion criteria, the ethics committee’s printed material ‘The Rights of the Research Subject in a Health Science Research Project’, a prepaid response envelope for those wishing to participate, a unique code for completing an online version of the questionnaire. All submitted questionnaires were reviewed, and if the inclusion criteria were met, further information was sent to the test subjects regarding the time and location of the examinations along with more detailed information about the individual procedures.

Recruiting departments

In 2013, 23 Danish departments performing measurements of lung diffusing capacity and body plethysmography in adults were invited to participate.

A launch meeting presenting the study was held, with 19 departments attending. Following confirmation of participation, auditing visits were scheduled. Auditing was carried out by at least two auditors, who inspected the equipment and procedures. Calibration checks were performed using external calibrators, and the execution of the different lung function tests was observed. In case of deviation from the ERS/ATS 2005 standards, corrective actions were implemented. In addition to equipment and measurement techniques, laboratory facilities – including safety, infection control, and data back-up – were also inspected. Calibrations were conducted with the assistance from a representative of the equipment manufacturer when necessary.

The auditors were experienced in lung function testing and had participated in the development of the ERS European Spirometry Driving License course [24], whose training principles were adapted for this study.

Measurements

Equipment and calibration details are reported in the supplementary material. Calibrators were traceable to international standards. The following measuring instruments were controlled as part of the auditing process: Thermometers for room temperature and relative humidity; barometric pressure, obtained from the Danish Meteorological Institute and corrected for height above sea level; and three-litre calibration syringes, which were inspected and manually tested for leaks. If not calibrated within the past year, re-calibration was offered [25].

Stadiometers were checked for malfunction of the head plate and calibrated. Arm span (maximum fingertip-to-fingertip distance) was measured with a wall-mounted metre stick (EU class III) and a plumb line.

Flow recording equipment was calibrated with a decompression calibrator [26] at three levels of forced expiratory volume in 1 second (FEV₁), forced vital capacity (FVC), and peak expiratory flow (PEF) at ATPD (ambient temperature and pressure, dry). DLco equipment was calibrated using a Hans Rudolph simulator at three levels of DLco [27]. Calibration of lung volume (intrathoracic gas volume = ITGV) measurement was performed using a borosilicate bottle filled with copper sponges and a pump simulating tidal breathing [28]. A 100 mL stroke volume was generated using an electric pump (Medical Electronics Construction (MEC), Fosses-la-Ville, Belgium). Vyaire equipment was volume-calibrated by the local distributor using a dedicated Jaeger (CareFusion, Höchberg, Germany) or Vyaire volume calibrator (Vyaire Medical, Mettawa, IL, USA). All equipment was serviced by the local representative, and – with the exception for one laboratory, which was excluded – all passed the quality check and participated in the study. The performance of the laboratory technicians was inspected from the preparation phase, including the arrival and identification of participants, through the measurement of height and weight, and during lung function testing (see supplementary material S3). It was planned to transfer measured results electronically from devices to statistical software; however, this was only possible in laboratories with an updated Vyaire software feature. In laboratories without access to this feature, data were entered manually into Microsoft Excel (Microsoft Corporation, Redmond, WA, USA).

Lung function tests

Subjects were studied within a single day, undergoing anthropometric measurements and extended lung function testing. An overview and details of the testing procedures are provided in the supplementary material (S1-S3). All lung function tests were conducted at the participating centres, following standardised protocols and using local equipment (see supplementary material S4). Most centres (12 out of 13) used the Jaeger MasterScreen PFT Pro system (CareFusion, Höchberg, Germany), while one centre used MEC (Fosses-la-Ville, Belgium).

All participating centres’ equipment was quality-controlled prior to testing by the respective service organisations – namely, the same highly experienced individual from the supplier of all CareFusion/Vyaire equipment, and the constructor of MEC, respectively. Each centre also received a visit from the central audit team (as described above).

Measurements included dynamic spirometry, body plethysmography, DLco, and bronchodilator testing with dynamic spirometry (see supplementary material S2–S3). The following data were obtained and reported: FEV₁, FVC, FEV₁/FVC, total lung capacity (TLC), residual volume (RV), functional residual capacity (FRC), RV/TLC, FRC/TLC, and DLco without and with haemoglobin (Hb) correction (DLco_cor), diffusion constant for CO (transfer coefficient) without and with Hb correction (Kco and Kco_cor), and alveolar volume (V_A). Additional lung function parameters were measured and calculated (see supplementary material S26). Hb measurements from capillary blood samples were analysed using the HemoCue device (Hb 201+; HemoCue AB, Ängelholm, Sweden).

All demographic and lung function data were documented and saved locally (supplementary material S1). Upon completion of all tests at each centre, data were transferred to the principal investigator.

Statistics

Multiple linear regression analyses were performed with lung function parameters as response variables, and age, weight, and either height or arm span as covariates. Preliminary investigations indicated that lung function parameters are either linearly or parabolically associated with age; therefore, age squared was also included as a covariate. The analyses were performed separately for men and women.

Model validation was done by inspecting QQ-plots of the standardised residuals, as well as plots of standardised residuals against fitted values and against each covariate. Several lung function parameters were log-transformed using the natural logarithm (LN).

All analyses were performed using R: A Language and Environment for Statistical Computing, version 4.2.0. A p-value < 0.05 was considered statistically significant.

Ethical approval

The study was approved by the Ethical Committee for the Central Denmark Region (file no. 1–10-72–552-12), in accordance with the Declaration of Helsinki. Both oral and written consent were obtained from all participants prior to participation.

Results

A total of 8000 invitations (supplementary material S7), including a questionnaire, were sent by letter to adults aged 18 years and older with no registered lung and/or cardiac diagnosis (supplementary material S5 for ICD-10 codes), residing in the municipalities of the 13 participating centres (19 centres were audited, but five never recruited any subjects, and one did not meet the study’s quality criteria).

A total of 1209 (15%) individuals returned the letter with a completed questionnaire. More than 20% of these were excluded due to a smoking history exceeding one pack-year. If no other exclusion criteria were met, subjects were booked for testing. Seventeen subjects were tested but later excluded for various reasons (supplementary figure S14). Some participants did not complete all tests; therefore, some analyses included fewer participants than the 908 eligible for analyses.

The baseline characteristics of the final 908 participants (47% male and 53% female) are shown in Table 1. Ages ranged from 18 to 97 years, with a median [IQR] age of 53.9 [31.5] years. Of these, 24.6% (n = 223) were aged 70 years or older (8.8% ≥80 years, and 1.7% ≥90 years). The distribution of age, height, arm span, weight, and BMI is displayed in supplementary figures S15–S19. Height and arm span were highly linearly correlated (r² = 0.84) with the regression line: ‘Height = 0.78 * arm span +36.58’ (supplementary figure S6).

Table 1.

Demographic and lung function data for males and females.

	Males (N = 430)			Females (N = 478)
	N	Mean ± SD	Range	N	Mean ± SD	Range
Age (year)	430	54.1 ± 18.83	18.3–97.4	478	53.5 ± 19.35	18.2–97.4
Height (cm)	430	177.9 ± 8.12	149.1–197.5	478	166.4 ± 7.55	146.6–196.3
Arm span (cm)	334	180.7 ± 9.07	153.9–202.5	354	166.4 ± 9.01	143.0–201.0
Weight (kg)	430	82.7 ± 13.94	52.0–157.6	478	70.5 ± 13.3	45.1–129.6
BMI (kg/m2)	430	26.08 ± 3.892	18.0–44.4	478	25.45 ± 4.310	16.99–43.28
FEV1 (l)	421	3.96 ± 0.933	1.4–6.3	478	3.04 ± 0.769	1.17–5.48
FVC (l)	421	5.09 ± 1.138	2.0–8.5	470	3.84 ± 0.890	1.74–6.79
FEV1/FVC	421	0.78 ± 0.065	0.6–1.0	470	0.79 ± 0.068	0.57–0.96
RV (l)	414	2.36 ± 0.641	1.0–4.8	470	2.01 ± 0.589	0.83–4.37
ITGV (FRC) (l)	414	3.90 ± 0.857	2.0–6.9	470	3.23 ± 0.730	1.80–6.52
TLC (l)	414	7.38 ± 1.203	4.0–12.2	470	5.74 ± 0.930	3.65–9.42
VC (l)	421	5.02 ± 1.115	1.7–8.0	470	3.72 ± 0.823	1.48–6.43
RV/TLC	414	0.32 ± 0.084	0.1–0.7	470	0.35 ± 0.09	0.18–0.61
DLcocor (mmol/min/kPa)	342	9.8 ± 2.3	3.6–15.6	403	7.70 ± 1.8	3.0–15.0
VA (l)	129	6.54 ± 1.267	3.4–9.5	159	5.39 ± 1.041	2.14–8.81
Kco (mmol/min/kPa/l)	425	1.48 ± 0.262	0.7–2.3	476	1.48 ± 0.241	0.86–2.29
Kcocor (mmol/min/kPa/l)	342	1.49 ± 0.253	0.8–2.2	401	1.49 ± 0.248	0.90–2.20

Legend: Data are presented as Mean ± Standard deviation. Abbreviations: N, Number of subjects; BMI, Body mass index; FEV₁, Forced expiratory volume in 1 second; FVC, Forced vital capacity; FRC, Functional residual capacity; ITGV, Intrathoracic gas volume; TLC, Total lung capacity; VC, Vital capacity; RV, Residual volume; DLco_cor, Pulmonary diffusing capacity for carbon monoxide corrected for haemoglobin; V_A: Alveolar volume; Kco, Transfer coefficient for CO without Hb correction; Kco_cor, Transfer coefficient for carbon monoxide corrected for haemoglobin.

DALFUMAT reference equations

Lung function parameters were dependent on sex, age, age squared, weight, and either height or arm span as covariates, while the test centre was not a significant covariate. Sex-specific prediction equations based on age, age squared, weight, and height for 29 lung function parameters are presented in Table 2. An example of calculation of the predicted value and of lower limit of normal (LLN) for a patient using the DALFUMAT prediction equations is provided in Table 2.

Table 2.

Sex-specific prediction equations based on age (years), age squared, weight (kg) and height (m) for DALFUMAT.

Parameter	Intercept	Age	Age squared	Weight	Height	R^2	RSD
Females
Ln FEV1 (l)	−1.0717	0.00201	−9.613E-05	−0.000459	1.432	0.72	0.145
Ln FVC (l)	−1.316	0.00686	−1.199E-04	−0.00111	1.641	0.71	0.132
FEV1/FVC	1.168	−0.00411	2.136E-05	0.000491	−0.158	0.25	0.0589
FEV1/VC	1.054	−0.00520	2.723E-05	0.000443	−0.0484	0.19	0.0916
FEV1/FEV6	1.049	−0.0000368	−1.081E-05	0.000807	−0.165	0.15	0.0527
Ln FEV6 (l)	−1.123	0.00290	−9.002E-05	−0.00179	1.617	0.71	0.132
PEF (l/s)	−9.218	0.0481	−6.977E-04	0.00281	9.580	0.42	1.341
Ln FEF50 (l/s)	−0.0815	0.00414	−1.408E-04	0.00324	0.788	0.40	0.338
MFEF75/25 (l/s)	−1.297	−0.0603	2.390E-04	0.00978	3.610	0.45	0.938
FEV1/PEF (ml/(l/min))	7.712	−0.0498	2.034E-04	−0.00184	0.888	0.20	1.168
FIV1 (l)	−4.0925	0.0100	−2.815E-04	−0.00157	4.669	0.46	0.696
FIF50 (l/s)	−5.0774	0.0359	−5.354E-04	−0.00308	5.352	0.27	1.177
FIF50/FEF50	−0.0308	0.00120	4.578E-05	−0.00339	0.728	0.04	0.579
Ln FEF50/FIF50	1.167	−0.00612	1.624E-05	0.00451	−0.746	0.03	0.469
Ln TLC (l)	−0.817	0.00777	−6.934E-05	−0.00123	1.469	0.49	0.115
RV (l)	−4.464	0.0222	1.201E-06	−0.00406	3.348	0.47	0.430
RV/TLC	0.202	0.000887	2.724E-05	−0.000214	0.0183	0.65	0.0539
VC (l)	−4.923	0.0171	−3.352E-04	−0.00197	5.375	0.63	0.504
Ln FRC (l)	−2.0890	0.00851	−5.121E-05	−0.00560	2.00839	0.36	0.175
Ln FRC/TLC	−1.272	0.000740	1.813E-05	−0.00437	0.539	0.20	0.131
ERV (l)	−3.291	−0.00346	−7.116E-05	−0.0134	3.472	0.54	0.352
DLco (mmol/min/kPa)	−7.095	0.0242	−6.504E-04	0.00875	8.958	0.64	1.047
Ln DLcocor (mmol/min/kPa)	0.0929	0.00568	−1.193E-04	0.000904	1.166	0.70	0.138
Kco (mmol/min/kPa/l)	2.626	−0.00516	−3.127E-05	0.00397	−0.631	0.43	0.182
Kcocor (mmol/min/kPa/l)	2.841	−0.00796	−1.026E-05	0.00402	−0.703	0.44	0.187
VA (l)	−9.140	0.0452	−4.450E-04	−0.00752	8.334	0.53	0.630
TLC-SB (l)	−9.248	0.0451	−4.423E-04	−0.00638	8.436	0.54	0.631
VA/TLC	0.803	0.00222	−3.100E-05	−0.000297	0.062	0.13	0.0687
TLC-SB/TLC	0.850	0.00212	−2.993E-05	−0.000133	0.044	0.12	0.0696
Males
Ln FEV1 (l)	−1.00497	0.00188	−0.0000821	−0.00120	1.471	0.66	0.147
Ln FVC (l)	−1.201	0.00583	−0.000105	−0.00162	1.666	0.68	0.135
FEV1/FVC	1.122	−0.00333	0.0000204	0.000296	−0.143	0.12	0.0611
FEV1/VC	1.361	−0.00313	0.00000838	−0.000433	−0.221	0.25	0.0716
FEV1/FEV6	1.0425	0.00207	−0.0000275	0.000779	−0.193	0.14	0.0491
Ln FEV6 (l)	−0.964	−0.00341	−0.0000297	−0.00158	1.667	0.65	0.139
PEF (l/s)	−15.00917	0.0912	−0.00104	−0.0139	13.701	0.32	1.972
Ln FEF50 (l/s)	0.795	−0.000311	−0.0000783	0.00290	0.369	0.29	0.308
MFEF75/25 (l/s)	1.392	−0.0596	0.000229	0.00523	2.368	0.36	1.063
FEV1/PEF (ml/(l/min))	10.324	−0.0694	0.000369	0.00249	−0.598	0.15	1.361
FIV1 (l)	−5.719	−0.00249	−0.000208	0.00434	5.874	0.47	0.868
FIF50 (l/s)	−5.328	−0.00000191	−0.000269	0.00274	6.159	0.22	1.697
FIF50/FEF50	−0.569	−0.00793	0.000112	−0.00250	1.149	0.03	0.574
Ln FEF50/FIF50	1.094	0.00640	−0.0000746	0.00198	−0.809	0.02	0.478
Ln TLC (l)	−0.841	0.00495	−0.0000423	−0.00147	1.584	0.53	0.115
RV (l)	−3.684	0.0230	−0.00000632	−0.00802	3.0808	0.39	0.502
RV/TLC	0.462	0.000737	0.0000204	−0.000587	−0.111	0.55	0.0568
VC (l)	−9.133	0.0138	−0.000309	−0.00275	8.229	0.62	0.692
Ln FRC (l)	−1.556	0.00754	−0.0000451	−0.00834	1.866	0.37	0.178
Ln FRC/TLC	−0.715	0.00259	−0.00000277	−0.00688	0.282	0.33	0.139
ERV (l)	−3.445	0.00129	−0.000129	−0.0215	3.943	0.50	0.446
DLco (mmol/min/kPa)	−6.858	−0.00758	−0.000576	0.00987	10.194	0.64	1.392
Ln DLcocor (mmol/min/kPa)	0.273	0.00567	−0.000123	0.0000217	1.169	0.66	0.150
Kco (mmol/min/kPa/l)	3.052	−0.00530	−0.0000411	0.00480	−0.872	0.47	0.192
Kcocor (mmol/min/kPa/l)	2.984	−0.00463	−0.0000464	0.00342	−0.774	0.50	0.181
VA (l)	−12.197	0.03639	−0.000392	−0.0125	10.821	0.54	0.806
TLC-SB (l)	−12.436	0.0370	−0.000393	−0.0121	11.0213	0.55	0.807
VA/TLC	0.868	0.000472	−0.0000144	−0.000552	0.0602	0.12	0.0626
TLC-SB/TLC	0.905	0.000491	−0.0000141	−0.000467	0.0482	0.11	0.0627

Legends FEV₁, Forced expiratory volume in 1. Second (l); FVC, Forced vital capacity (l); FEV₆, Forced expiratory volume in 6. Second (l); PEF, Peak expiratory flow (l/s); FEF₅₀, Forced expiratory flow when 50% of FVC has been exhaled (l/s); MFEF_75/25, Forced mid-expiratory flow during the middle half of the FVC (l/s); FIV₁, Forced inspiratory volume in 1. Second (l); FIF₅₀, Forced inspiratory flow when 50% of the forced inspiratory vital capacity has been inhaled (l/s); TLC, Total lung capacity (l); RV, Residual volume (l); FRC, Functional residual capacity (l); VC, Vital capacity (l); ERV, Expiratory reserve volume (l); DLco, Diffusing capacity for carbon monoxide without Hb correction (mmol/min/kPa); DLco_cor, Diffusing capacity for carbon monoxide with Hb correction (mmol/min/kPa); Kco, Transfer coefficient for carbon monoxide without Hb correction (mmol/min/kPa/l); Kco_cor, Transfer coefficient for carbon monoxide with Hb correction (mmol/min/kPa/l); V_A, Alveolar volume (l); TLC_SB, Total lung capacity measured with the single breath technique (l); RSD, Residual standard deviation; R^2 = R², Determination coefficient; Age squared = age*age.

Sex-specific prediction equations based on arm span (as an alternative to height), along with age, age squared, and weight, for the 10 most clinically relevant lung function parameters are presented in Table 3. The prediction equations were equally suitable for predicting lung function whether based on height or arm span (supplementary figure S25).

Table 3.

Prediction equations based on sex, age, age squared, weight (kg) and arm span (cm) for DALFUMAT.

	Females
Parameter	Intercept	Age	Age squared	Weight	Arm span	R^2	RSD
Ln FEV1 (l)	−0.763	0.00209	−0.000101	−4.690E-05	0.0124	0.73	0.145
Ln FVC (l)	−0.927	0.00758	−0.000132	−6.256E-04	0.0139	0.74	0.130
FEV1/FVC	1.105	−0.00460	0.0000265	4.307E-04	−0.00114	0.24	0.0624
Ln TLC (l)	−0.390	0.00798	−0.0000770	−4.335E-04	0.0119	0.51	0.117
RV (l)	−3.206	0.0223	−0.0000228	−8.605E-04	0.0252	0.44	0.459
RV/TLC	0.281	0.000836	0.0000259	2.984E-05	−0.000336	0.64	0.0549
VC (l)	−3.778	0.0165	−0.000342	−7.0574E-04	0.0467	0.66	0.487
DLco (mmol/min/kPa)	−4.579	0.0108	−0.000577	1.528E-02	0.0744	0.65	1.078
Kco (mmol/min/kPa/l)	2.408	−0.00718	−0.0000114	3.828E-03	−0.00475	0.45	0.185
VA (l)	−6.971	0.0433	−0.000470	−2.890E-03	0.0695	0.58	0.621
	Males
Parameter	Intercept	Age	Age squared	Weight	Arm span	R^2	RSD
Ln FEV1 (l)	−0.744	0.00174	−0.0000884	−0.000622	0.0129	0.66	0.150
Ln FVC (l)	−0.833	0.00596	−0.000116	−0.00113	0.0143	0.67	0.141
FEV1/FVC	1.0443	−0.00356	0.0000242	0.000371	−0.00103	0.11	0.0616
Ln TLC (l)	−0.293	0.00458	−0.0000484	−0.000930	0.0125	0.46	0.125
RV (l)	−1.591	0.0216	−0.0000206	−0.00624	0.0187	0.34	0.532
RV/TLC	0.558	0.000831	0.0000193	−0.000571	−0.00162	0.54	0.0590
VC (l)	−7.0925	0.0123	−0.000338	−0.000544	0.0694	0.59	0.726
DLco (mmol/min/kPa)	−4.347	−0.0168	−0.000569	0.0125	0.0877	0.64	1.404
Kco (mmol/min/kPa/l)	2.617	−0.00695	−0.0000218	0.00448	−0.00586	0.48	0.190
VA (l)	−8.770	0.0339	−0.000437	−0.00922	0.0873	0.52	0.817

Legends: FEV₁, Forced expiratory volume in 1 second (l); FVC, Forced vital capacity (l); VC, Vital capacity (l); TLC, Total lung capacity (l); RV, Residual volume (l); DL_CO, Pulmonary diffusing capacity for carbon monoxide (mmol/min/kPa); Kco, Transfer coefficient for carbon monoxide (mmol/min/kPa/l); V_A, alveolar volume (l); RSD, Residual standard deviation; R^2 = R², Determination coefficient; Age squared = age*age.

Comparison of DALFUMAT with GLI, ECCS, and Løkke

The prediction equations for eight clinically relevant lung function parameters (FEV₁, FVC, FEV₁/FVC, TLC, RV, RV/TLC, DLco, and Kco) from the DALFUMAT reference material were compared with prediction equations from GLI [12–15], ECCS [9,10], and, for spirometry, also with the Danish spirometry predictions by Løkke [7]. DALFUMAT data were applied to the different prediction equations, and z-scores were calculated and displayed by sex (Figures 1 and 2, and supplementary figures S20–S22 and supplementary tables S23–S24).

Figure 1.

Comparison between prediction equations from DALFUMAT (model), GLI, ECCS and Løkke for FEV₁, FVC and FEV₁/FVC. Density plot (y-axis) and z-score (x-axis). Left figures for females. Right figures for males.

Figure 2.

Comparison between prediction equations from DALFUMAT (model), GLI and ECCS for TLC and DLco. Density plot (y-axis) and z-score (x-axis). Left figures for females. Right figures for males.

The predicted lung function values for lung volumes using the other three reference materials were lower than those predicted by DALFUMAT (model). This is reflected in z-scores for predicted dynamic volumes (FEV₁ and FVC) (Figure 1 and supplementary figure S20) and static volumes (TLC and RV) (Figure 2 and supplementary figure S21), which were skewed to the right compared to DALFUMAT – less so for GLI and more so for ECCS. For example, in females, the z-score for FEV₁ was +0.6 for GLI, +0.8 for Løkke, and +1.0 for ECCS, while in males, deviations were smaller, with FEV₁ z-scores of approximately +0.4 for GLI and Løkke, and +0.7 for ECCS (Figure 1 and supplementary tables S23–S24).

The same trend – DALFUMAT predicting higher values than the others – was even more pronounced for FVC and TLC in females, with z-scores for ECCS being +1.7 for FVC and +1.2 for TLC, while differences were smaller in males (Figures 1 and 2 and supplementary tables S23–S24).

Predictions for RV were closer to DALFUMAT, with z-scores of +0.4 for ECCS and +0.6 for GLI in females, and +0.2 for ECCS and +0.5 for GLI in males (supplementary figure S21 and supplementary tables S23–S24).

Predictions for the ratios FEV₁/FVC and RV/TLC showed different patterns (supplementary figures S20–S21 and supplementary tables S23–S24). For FEV₁/FVC, differences were small (≤−0.3 z), but negative, indicating that DALFUMAT predicted lower ratios. For RV/TLC, z-score for GLI was approximately +0.5, while it was only +0.2 to +0.4 for ECCS.

For DLco, the predicted values from ECCS were skewed to the left (ECCS predicted higher values than DALFUMAT and GLI), with z-scores ranging from −0.3 to −0.5, while GLI was skewed to the right, with z-scores from +0.1 to +0.5 (supplementary figure S22 and supplementary tables S23-S24). GLI and DALFUMAT predicted almost identical values for Kco.

Bronchodilator testing

Prediction equations for post-bronchodilator spirometry parameters, based on both height and arm span, are presented in Table 4.

Table 4.

Post-bronchodilator prediction equations for spirometry for females and males based on (a) height and (b) arm span.

Parameter	Intercept	Age	Age squared	Weight (kg)	Height (m)	R^2	RSD
(a) based on height (m)
Females
Ln FEV1 post (l)	−1.0487	−0.0000431	−0.0000790	−0.000463	1.464	0.739	0.138
Ln FVC post (l)	−1.228	0.00401	−0.0000936	−0.000976	1.620	0.691	0.134
FEV1/FVC post	1.133	−0.00361	0.0000150	0.000409	−0.124	0.314	0.0552
PEF post (l/s)	−9.907	0.0474	−0.000687	0.00473	9.922	0.452	1.269
Ln MFEF75/25 post (l/s)	−0.400	−0.000635	−0.000125	0.00195	1.0726	0.505	0.329
Males
Ln FEV1 post (l)	−0.986	0.000574	−0.0000702	−0.000925	1.480	0.684	0.139
Ln FVC post (l)	−1.0794	0.00319	−0.0000779	−0.00111	1.603	0.665	0.132
FEV1/FVC post	1.0542	−0.00230	0.00000825	0.000128	−0.0939	0.177	0.0550
Ln PEF post (l/s)	−0.412	0.0129	−0.000138	−0.000969	1.414	0.342	0.196
Ln MFEF75/25 post (l/s)	−0.0930	−0.00134	−0.0000938	0.00111	0.943	0.425	0.310
(b) based on arm span (cm)
Females
Ln FEV1 post (l)	−0.749	0.000109	−0.0000836	−0.0000795	0.0128	0.756	0.136
Ln FVC post (l)	−0.906	0.0047	−0.000104	−0.000723	0.0142	0.724	0.130
FEV1/FVC post	1.109	−0.00405	0.0000194	0.000509	−0.00111	0.309	0.0574
PEF post (l/s)	−9.0515	0.0481	−0.000721	0.006346	0.0946	0.511	1.243
Ln MFEF75/25 post (l/s)	−0.189	−0.00141	−0.000123	0.00223	0.00956	0.511	0.331
Males
FEV1 post (l)	−3.129	−0.0160	−0.000126	−0.00196	0.0475	0.655	0.555
Ln FVC post (l)	−0.630	0.00302	−0.0000855	−0.000626	0.0132	0.629	0.141
FEV1/FVC post (l)	1.000353	−0.00271	0.0000136	0.000204	−0.000644	0.160	0.0560
Ln PEF post (l/s)	−0.0856	0.0134	−0.000148	−0.000491	0.0119	0.331	0.196
Ln MFEF75/25 post (l/s)	0.0768	−0.00115	−0.0000991	0.00162	0.00815	0.421	0.312

Legends: FEV₁, Forced expiratory volume in 1. Second (l); FVC, Forced vital capacity (l); PEF, Peak expiratory flow (l/s); MFEF_75/25, Forced mid-expiratory flow during the middle half of the FVC (l/s); RSD, Residual standard deviation; R^2 = R², Determination coefficient; Age squared = age*age.

The change in dynamic spirometry parameters following bronchodilator administration was calculated in four different ways: 1) As an absolute change in litres, 2) As a percentage of the pre-bronchodilator value, 3) As a percentage of the predicted value, and 4) As a z-score (supplementary figures S10–S13 and supplementary table S9).

1) FEV₁ increased by a mean of 87 ± 147 ml (with a 95th percentile upper limit of 329 ml), 2) FEV₁ increased by a mean of 2.6 ± 4.6% of the pre-bronchodilator value (95th percentile upper limit: 10.2%), 3) FEV₁ increased by a mean of 2.4 ± 4.2% of the predicted value (95th percentile upper limit: 9.3%), and 4) FEV₁ increased by a mean z-score of 0.0047 ± 0.30 (95th percentile upper limit: 0.50 z).

Hence, the 95th percentile upper limit for ΔFEV₁ of 9.3% predicted corresponds to a change in z-score of 0.50. Similarly, the 95th percentile upper limit for ΔFVC is 6.3% predicted and a z-score of 0.50, while for ΔFEV₁/FVC it is 9.4% predicted and a z-score of 0.87.

Discussion

This study is the first to provide prediction equations for all clinically relevant routine lung function parameters, obtained from a random sample of healthy individuals aged 18–97 years. Strict quality assurance was applied, and the results were compared to the widely used GLI [12–15] and ECCS reference values [9,10], as well as the local reference values from Løkke [7], revealing clinically significant differences. Additionally, reference values based on arm span and post-bronchodilator measurements were included. Furthermore, we propose new criteria for bronchodilator response using z-scores as a supplement to the recently suggested criteria based on percentage of predicted values.

The ATS recommends that reference equations should be evaluated in a representative sample of local, healthy subjects prior to their implementation in a lung function laboratory [29]. In the present nationwide Danish study of lung function in a random sample of healthy adults, we found an overall agreement with other prediction equations, but also differences of potential clinical significance.

The GLI network has suggested that a goodness-of-fit between measured data and the predicted values, with an average residual (z-score) <±0.50, is not physiologically or clinically meaningful [15]. Others have proposed a lower threshold, suggesting that differences in z-score <±0.30 are not clinically meaningful [30].

We compared our data with GLI predictions for eight of the most clinically important lung function parameters from spirometry, static lung volumes, and DLco (Figures 1 and 2, supplementary figures S20-S22 and supplementary tables S23-S24). In females, ‘large’ z-score deviations (>±0.50) were found in 4 out of 8 equations, and ‘moderate’ deviations (>±0.30) in an additional 2 out of 8. In males, deviations were ‘large’ in 1 out of 8 and ‘moderate’ in 3 out of 8 equations. Thus, approximately half of the GLI prediction equations did not fit our data well, with a poorer fit observed in females.

Similarly, when comparing our data with the prediction equations from ECCS [9,10] and Løkke [7], we identified even more and larger deviations. For ECCS, ‘large’ deviations in 5 out of 8 and ‘moderate’ deviations in 2 out of 8 equations in females. In males, ‘large’ and ‘moderate’ deviations were found in 2 out of 8 and 3 out of 8 equations, respectively. Comparing spirometry equations with Løkke, we found that 2 out of 3 had ‘large’ deviations in females, and all 3 equations showed ‘large’ or ‘moderate’ deviations in males.

Hence, our DALFUMAT lung function data do not align well with several reference equations from GLI, and even less so with ECCS and Løkke, where substantial deviations were observed in most equations. The poorest fit was found for parameters related to spirometry and static lung volumes, whereas parameters for diffusing capacity showed better agreement with GLI and ECCS predictions, particularly in males.

These findings are consistent with a recent large study from Austria involving over 5000 respiratory-healthy subjects aged 6–80 years. In that study, z-score fits for static lung volumes (TLC, RV, and RV/TLC) deviated more than ±0.50 from GLI predictions in both sexes, with more pronounced deviations in females, leading to the development of new Austrian reference equations [23].

Similarly, a Belgian study found that GLI predictions generally fit most lung function parameters well in subjects with average height and age. However, z-score deviations correlated positively or negatively with height and age away from the average, with the largest deviation observed in RV, especially in females [30].

There might be several explanations for the differences between the predicted values from DALFUMAT, GLI [12–15], ECCS [9,10], and Løkke [7]. These include technical differences, true biological variation (including secular trends) between populations, and sampling error. Likely contributing factors include the use of outdated (ECCS/Løkke) versus modern equipment and quality control (GLI/DALFUMAT), differences in staff training (DALFUMAT), older (ECCS/Løkke) versus more recent data (GLI/DALFUMAT), heterogenous equipment types (ECCS/GLI) versus uniform equipment (DALFUMAT/Løkke), and heterogeneous (ECCS/GLI) versus homogeneous populations (DALFUMAT/Løkke).

Standing height (stature) as a proxy for chest size is an indispensable variable for predicting lung function. However, measuring standing height is not always possible or reliable – for example, in individuals who use a wheelchair or have severe scoliosis or kyphosis. In such cases, predicted height can be inferred from arm span measurements. However, if standing height is predicted from the arm span and used to estimate lung function, it is important to note that the ratio between arm span and standing height differs with age, sex, and geographical origin [31–33]. To overcome this bias, we present lung function prediction equations based directly on arm span, in addition to those based on height. Using arm span instead of height has potential advantages, such as saving time by avoiding the need to remove shoes. The time required to correctly position the participant for arm span measurement may be the equivalent to the time needed to position the head in the Frankfurt plane during height measurement. Nevertheless, more data and better anthropometric instruments are needed to optimise the measurement process.

We found that the 95th percentile threshold for bronchodilator response for FEV₁ was a change of 329 ml and 10.2% when calculated from the pre-bronchodilator value, which is similar to findings of 320 ml and 13.3% in a study of 2371 healthy subjects [34]. However, this differs from the ERS/ATS 2005 [35] recommendation of a change of > 200 ml and >12%. In 2023, ERS/ATS revised these criteria [16], suggesting a threshold change of >10% of predicted, which is consistent with our findings of >9.3% predicted and aligns with the study of Quanjer [34] (>11.6%) and the large study of burden of obstructive lung diseases by Tan [36].

The ERS/ATS 2023 standard [16] did not specify the threshold as a change in z-score, even though the grading of abnormality is recommended to be expressed in z-scores. Based on our data, we suggest a threshold z-score change of >0.50. Accordingly, the bronchodilator response of >10% predicted, as suggested by ATS/ERS 2023, would correspond to a z-score change of >0.54 for FEV₁ and >0.80 for FVC, based on our DALFUMAT data. We propose that future studies of bronchodilator response should include thresholds for both % predicted and z-score and evaluate their respective clinical utility.

Traditionally, the FEV₁ value after bronchodilator testing is compared to reference values derived from pre-bronchodilator reference materials. However, using post-bronchodilator reference values may help avoid falsely elevated percent predicted FEV₁, as the predicted FEV₁ (and predicted FEV₁/FVC) after bronchodilation is higher than the pre-bronchodilator value. This may prevent underestimation of disease severity and reduce the risk of misclassifying airway obstruction [37–40].

The application of post-bronchodilator spirometry reference values may also be advantageous in identifying individuals with pre-COPD – i.e. those who do not exhibit obstruction according to pre-bronchodilator reference values but are at risk of developing COPD [38–40].

Currently, there are no internationally accepted post-bronchodilator reference values, and neither GLI nor ECCS include such prediction equations. However, recent publications have presented post-bronchodilator reference values [37,39,40], and we now contribute with our findings from DALFUMAT.

The prevalence of overweight and obesity is increasing globally. In 2022, 43% of adults were overweight and 16% were living with obesity, according to the World Health Organization (WHO) [41]. However, some lung function reference materials exclude subjects with overweight and obesity, as excess body weight may reduce static and dynamic lung volumes – especially expiratory reserve volume and FRC. Measures of diffusing capacity may also be affected, such as reduced V_A and increased Kco [42]. Sensitivity analysis both in GLI reference values for static volumes and DLco, comparing datasets with and without overweight/obese individuals, did not show large differences (e.g. <75 ml and <2.5% in TLC, RV, and FRC). Consequently, both GLI reference materials included overweight/obese individuals [13,15]. Similarly, to improve generalisability of our reference equations in DALFUMAT, we did not exclude overweight/obese individuals, provided they were lung-healthy individuals. Notably, in contrast to the ECCS and GLI, weight was an independent predictor for all reference values in DALFUMAT, alongside sex, age, age squared, and height (or arm span).

Strengths and limitations

This study has several strengths: 1) A high degree of quality control strictly following the ATS/ERS 2005 standards [2, 3–5] was ensured, both before and during the data acquisition period. 2) Uniform equipment and service organisation were used at all except one of the many test centres, likely improving quality of lung function testing across the country and providing nationwide generalisability. Jaeger equipment was used in 12 of 13 centres comprising 89% of tests, and MEC was used in the remaining 11%. 3) Unlike ECCS and GLI, multiple-lung function parameters from spirometry, body plethysmography, and diffusion measurements were obtained from the same subjects, ensuring internal consistency between the prediction equations. 4) A considerable number of elderly participants were included, allowing prediction equations to cover adults up to 97 years of age, higher than covered by GLI and ECCS. 5) Both normal-weight and overweight individuals were included, enhancing generalisability, especially given the increasing prevalence of overweight in many populations. 6) We provided prediction equations for post-bronchodilator spirometry and proposed new cut-off values for a positive response. We provided both pre- and post-bronchodilator prediction equations for spirometry. The latter has relevance assessing ventilatory impairment after bronchodilation treatment. 7) In addition to the traditional prediction equations based on height, we also provided equations based on arm span. This is particularly useful for subjects for whom height measurements is not feasible, and arm span may even be quicker to measure (no need to remove shoes).

The main limitations of our study are as follows: 1) Children were not included, so we cannot determine how well the current reference materials apply to Danish children. This highlights the need for further studies. 2) Only individuals of European origin were studied – similar to the GLI reference values for static lung volumes [15] and DLco [13] including the erratum [14]. Consequently, data for individuals of other ethnic origins are still required. 3) The sample size of 908 subjects may be viewed as a limitation, as it is comparable to or smaller than that used in ECCS [9,10] (ranging from 960 to 11,000 across different equations), and significantly smaller than in GLI (ranging from 7200 to 100,000 for various lung function parameters). However, with approximately 900 subjects, the error due to sample size is predicted to be minimal (±0.3 z-score), and our population is three times the minimum of 300 subjects required for prediction equations to be considered reliable [43]. 4) We used multiple linear regression rather than the more sophisticated lambda-mu-sigma (LMS) GAMLSS method employed by GLI [15]. The LMS method is particularly valuable when compiling large data sets spanning a wide age range, from children to the elderly. However, both we and ECCS opted for multiple linear regression because it provided sufficiently accurate predictions for adults and is simpler to interpret. Furthermore, since model fit was also evaluated by plotting residuals against predicted values and covariates, no need was found to vary the SD across covariates. Therefore, LMS analysis was not deemed necessary.

Conclusion

New prediction equations have been developed for all clinically relevant routine lung function parameters – including spirometry, body plethysmography, and diffusing capacity – based on a random, nationwide sample of healthy Danish adults. In addition, reference values have been established for arm span, post-bronchodilator spirometry, and bronchodilator response.

Given that a substantial proportion of these reference values differ clinically significantly from those found in the GLI and ECCS materials, we recommend the nationwide adoption of the new DALFUMAT reference values.

Future research should focus on comparing and validating our equations against other existing models to determine which best correlate with clinical outcomes across diverse patient populations.

	Intercept	Age (yr)	Age squared	Weight (kg)	Height (m)	R2	RSD	sum equation	Predicted	Predicted LLN
Ln FEV1	−1.0717	0.00201	-9.613E-05	−0.000459	1.432	0.715	0.1452	0.995	2.706	2.131

Calculation of predicted FEV₁ for a 70-year-old female, height 1.70 m, weight 80 kg using the equations from Table 2. These three parameters are inserted together with the respective coefficients in the equation:.

−1.0717 + (0.00201 × 70 yr) + (−9.613 × 10⁻⁵ × 70 × 70 yr²) + (−0.000459 × 80 kg) + (1.432 × 1.70 m) = 0.995.

Predicted value for FEV₁ is antilog 0.995 = EKSP(0.995) = 2.706 L.

Lower limit of normal (LLN) is calculated as ‘Predicted value −1.645 × RSD’ = EKSP(0.995-(1.645 × 0.1452)) = 2.131 L.

If her measured FEV₁ is 2.50 L ~ 92.4% predicted (100%× 2.50 L/2.71 L) and z-score = −0.55 ((Ln(2.50 L)-0.995 L)/0.1452). This z-score indicates a normal value, as it is greater than −1.645.

Calculation of predicted DLco and interpretation.

Since DLco is not log-transformed in this model, no antiLOG step is required, making the calculation more straightforward. Predicted value is 7.345 mmol/min/kPa and LLN is 5.623 mmol/min/kPa.

If her measured DLco is 5.10 mmol/min/kPa ~69.4% predicted (100% × 5.10/7.345) and z-score is −2.14 ((5.10–7.345)/1.0466). This z-score falls within the range of mildly reduced values, defined as between −2.5 to −1.645.

Funding Statement

This work was supported by the Danmarks Lungeforening; The Central Region of Denmark; Foundation for Professional Development of specialist medical practise; The participating Danish hospitals; Boehringer-Ingelheim Denmark; GlaxoSmithKline Denmark; Novartis Denmark; Alere Denmark; IntraMedic Denmark.

Disclosure statement

• Elisabeth Bendstrup: Honoraria for lectures from Boehringer Ingelheim & Astra Zeneca. Support for attending meetings and/or travel from Boehringer Ingelheim. Participation on a Data Safety Monitoring Board or Advisory Board: Simbec-Orion – Molecure.

• Bente Grønlund: Support for attending meetings and/or travel from Astra Zeneca and Chiesi Farmaceuti.

• Ole Hilberg: Participation on Advisory Boards: Astra Zeneca, Sanofi, Boehringer Ingelheim & Glaxo Smith Kline.

• Lars Kristensen: Honoraria for lectures from Astra Zeneca, Chiesi & Glaxo Smith Kline. Support for attending meetings and/or travel from Boehringer Ingelheim & Chiesi.

• Flemming Madsen: Donation of 6 vials of Provocholine for another method study from Birk NPC AS.

• Ingrid Louise Titlestad: Honoraria for lecture from Astra Zeneca. Support for conference meeting and travel from Boehringer Ingelheim. Participation on a Data Safety Monitoring Board or Advisory Board: Astra Zeneca.

• Ulla Møller Weinreich: Grants from The Independent Danish Research Foundation. Honoraria for lectures from Astra Zeneca, Resmed, Fisher and Paykel Healthcare, Sanofi & Glaxo Smith Kline. Participation on a Data Safety Monitoring Board or Advisory Board: Astra Zeneca, Sanofi & Glaxo Smith Kline. Chair, Danish Respiratory Society.

No potential conflict of interest was reported by the author(s).

Author contributions

Jann Mortensen: conception, design, audition, data collection, interpretation, supervision, first draft, revisions, approval of the final manuscript and first author.

Lars Kristensen: conception, design, audition, data collection, revisions and approval of the final manuscript.

Birgitte Hanel: audition, data collection, revisions and approval of the final manuscript.

Mathias Munkholm Larsen: audition, data collection, revisions and approval of the final manuscript.

Jan Abrahamsen: data collection, revisions and approval of the final manuscript.

Kirsten Sidenius: design, data collection, revisions and approval of the final manuscript.

Bo Martin Bibby: statistics, revisions and approval of the final manuscript.

Bente Grønlund: data collection, revisions and approval of the final manuscript.

Ole Hilberg: conception, design, revisions and approval of the final manuscript.

Niels Maltbæk: data collection, revisions and approval of the final manuscript.

Ingrid L. Titlestad: conception, design, revisions and approval of final manuscript.

Ronald Dahl: conception, design, revisions and approval of final manuscript.

Sofie Kryger: Control of data entry, revisions, and approval of the final manuscript.

Ulla Møller Weinreich: data collection, revisions and approval of the final manuscript.

Johannes Martin Schmid: data collection, revisions and approval of the final manuscript.

Elisabeth Bendstrup: data collection, revisions and approval of the final manuscript.

Jens Peder Dreyer Paludan: data collection, revisions and approval of the final manuscript.

Charlotte Hyldgaard data collection, revisions and approval of the final manuscript.

Lisbeth Mariager Danielsen: data collection, revisions and approval of the final manuscript.

Lene Sønderskov Dahl: data collection, revisions and approval of the final manuscript.

Elin Jørgensen: data collection, revisions and approval of the final manuscript.

Torben Tranborg Jensen: data collection, revisions and approval of the final manuscript.

Tilde Kinket Ellingsgaard: data collection, revisions and approval of the final manuscript.

Dan Fuglø: data collection, revisions and approval of the final manuscript.

Peter Hovind: data collection, revisions and approval of the final manuscript.

Ronan Martin Griffin Berg: data collection, revisions and approval of the final manuscript.

Flemming Madsen: conception, design, audit, data collection, interpretation, supervision, first draft, revisions, approval of the final manuscript and last author.

Data availability statement

The data underlying our findings can be shared upon reasonable request directed to the corresponding author.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20018525.2025.2606556

Take-home message

We provide comprehensive lung function reference values for individuals aged 18 to 97 years. Unlike ECCS and GLI, all measurements in our study were obtained from the same cohort of subjects, ensuring consistency across more than 30 prediction equations.