Abstract
Introduction:
Spirometry is the main pulmonary function test routinely employed in the occupational medicine practice. Its interpretation depends on the choice of the theoretical reference values. Therefore, our objective was to retrospectively evaluate the differences in the spirometric interpretation according to the reference values used.
Methods:
A total of 2462 spirometries performed during health surveillance programs were analyzed. Forced vital capacity (FVC), forced expiratory volume in one second (FEV1), and FEV1/FVC ratio were compared according to three different reference values: the European Coal and Steel Community (ECSC) 1971, the European Respiratory Society (ERS) 1993, and the Global Lung Initiative (GLI) 2012.
Results:
The GLI 2012 provided significantly higher predicted mean FVC and FEV1 values compared to the ERS 1993 and significantly lower compared to the ECSC 1971. The GLI 2012 were able to detect all the obstructive deficits and mixed patterns diagnosed with ECSC 1971 and ERS 1993, in addition to others not diagnosed by these two latter reference values. The number of restrictive patterns identified through the GLI 2012 was significantly reduced and increased compared to those diagnosed using the ECSC 1971 and the ERS 1993, respectively.
Discussion:
In comparison to the GLI 2012, the ERS 1993 values significantly underestimated obstructive and restrictive alterations. Conversely, the ECSC 1971 significantly underestimated obstructive changes, while overestimated restrictive patterns, compared to GLI. Although the GLI reference values may provide a correct spirometric interpretation, their validation in an Italian worker population is necessary to confirm their possible use in routine occupational health programs.
Keywords
Introduction
Spirometry is the main pulmonary function test (PFT) routinely employed in the occupational medicine (OM) practice (Townsend, 2011). The interpretation of spirometry depends on the choice of the theoretical values used for reference. A number of predictive equations have been proposed for different populations in the past years. The European Coal and Steel Community (ECSC) firstly recommended reference values in 1971 (Cara and Hentz, 1971). New reference values have been proposed by the ECSC in 1983 (Quanjer, 1983), subsequently confirmed by the European Respiratory Society (ERS) in 1993 (Quanjer et al., 1993). These latter reference values are still used by most of the Italian occupational physicians as standards selected in the commonly employed spirometry software. However, the American Thoracic Society (ATS)/ERS 2005 standards on the execution and interpretation of spirometry did not recommend any specific series of equations for Europe, suggesting the need for further studies (Pellegrino et al., 2005). In 2012, the Global Lung Initiative (GLI) developed a new set of equations, named GLI 2012 (Quanjer et al., 2012). They obtained approval and validation for their application in many parts of Europe, such as France (Hulo et al., 2016), Germany (Baur, 2013), and Norway (Langhammer et al., 2016). Recently, the ATS standardization of spirometry 2019 update recommended the use of GLI 2012 as default spirometry reference values, although other options may be provided (Graham et al., 2019). In this scenario, it seems important to define which of the reference values proposed may assure a greater diagnostic effectiveness according to the specific characteristics of the investigated populations. From an OM perspective, this may assure physicians the most appropriate approach to the interpretation of the spirometric examinations.
The objective of this study was to retrospectively evaluate the differences in the interpretation of spirometric examinations according to the reference values used. For this purpose, we compared the spirometric outcomes obtained using the ECSC 1971, the ERS 1993, and the GLI 2012 on a population of Italian workers employed in different occupational fields.
Methods
We analyzed spirometry examinations performed between January 2015 and July 2018, during the health surveillance programs, carried out in accordance to the Italian Legislative degree 81/2008, of workers employed in different sectors requiring pulmonary function evaluation for specific occupational risks. The spirometric tests were performed by OM physicians at the Public Health Department of the University of Naples Federico II or directly at the company sites. In all cases, Spirolab III portable turbine spirometers (MIR, Rome) were used. For each test, FlowMIR disposable turbines with cardboard mouthpiece were used. These devices met ATS/ERS 2005 and ATS 2019 standards, and were calibrated and tested before leaving the factory. As explained by the producer, these turbines required no additional calibration on site.
Spirometric tests were considered suitable to be included in the analysis if they were performed in respect of the ATS/ERS 2005 (Pellegrino et al., 2005) and the more recently updated ATS 2019 (Graham et al., 2019) standards of quality, exclusively in the course of health surveillance programs, in Caucasian workers with a known job task. When spirometric duplicates were available for the same subjects, the most recent examination was chosen. Instead, spirometries not complying with the quality standards, performed for reasons other than health surveillance or in non-Caucasian race workers, or with unknown job, were excluded from the analysis. Information on job tasks was considered an inclusion criterion as essential to ascertain the appropriateness of the spirometric examination in relation to occupational risks, particularly as concerns environmental or occupational exposure to airborne pollutants. Data about the pathological history, smoking habit, or information on the current or past exposure to specific occupational risks have been not reported as they did not affect our analysis. This information was considered outside the objective of the study as the aim of the investigation was not to validate a specific reference equation set, but to compare reference values commonly used in the OM practice in Italy. Therefore, it was not necessary to exclusively include tests from healthy subjects, in order to investigate a population representative of those commonly encountered in OM practice.
Spirometric parameters considered for the interpretation were forced vital capacity (FVC), forced expiratory volume in one second (FEV1), and FEV1/FVC ratio. The reference values, the lower limits of normal (LLNs), and the relative percentages of the predicted for these parameters were assessed according to the three sets of equations: the ECSC 1971, the ERS 1993, and the GLI 2012. The ERS 1993 values were provided automatically by the spirometer software. The ECSC 1971 were derived in an Excel spreadsheet elaborated by the authors using specific formulas available in the literature (Baur et al., 1999; Cara and Hentz, 1971). The GLI 2012 were extrapolated through an Excel spreadsheet freely provided online by the ERS (https://www.ers-education.org/guidelines/global-lung-function-initiative/spirometry-tools/desktop-sheet-calculator.aspx?idParent=138977).
Although the ECSC equations are based on the slow vital capacity, and not on the FVC, commonly evaluated in the pulmonary functionality assessment, it was possible to include ECSC reference values in our analysis. The two parameters, in fact, did not show significant differences in healthy subject populations (Blagev et al., 2016), comparable to those considered for the elaboration of the reference values themselves. Therefore, in the text and tables, we have employed the term “FVC” for all the three reference values investigated.
The ATS/ERS 2005 standards were employed to define ventilatory alterations (Pellegrino et al., 2005). However, the simple spirometries we performed allowed to make only certain obstructive deficit diagnosis. Restrictive and mixed alterations, the latter intended as a change characterized by a sure obstructive pathway with a possible restrictive component, could be only hypothesized through such kind of simple tests. For their confirmed diagnosis, in fact, global spirometry evaluation, able to assess residual volume (RV) and total lung capacity (TLC), is necessary. Therefore, we did not indicate these latter changes as “deficits,” but as mixed or restrictive “patterns.”
Collected data have been exported to SPSS v16.0. One-way analysis of variance (ANOVA) was performed to test the significance of differences between the predicted values, LLNs, and percentages of predicted determined according to the three reference values. The least significant difference (LSD) test was used for post hoc pair-wise comparisons. In all cases, a p value <0.05 was considered statistically significant.
Results
From the initially available 7763 spirometry tests, 5301 (68.3%) were excluded: 3523 (45.4%) as duplicates, 156 (2%) for belonging to non-Caucasian subjects, 698 (9%) for unknown job, 924 (11.9%) for not fully meeting the ATS/ERS 2005 and ATS 2019 quality criteria. A final number of 2462 examinations (31.7% of the total) was considered suitable for our analysis.
The analyzed population included workers employed in different activities and coming from different Italian regions, and particularly traffic policemen; technical personnel of the University of Naples Federico II in the Campania region; staff assigned to the smoking rooms of gambling bars of Lombardy, Veneto, Piemonte, and Emilia-Romagna regions; assemblers, repairers, and welders in shipyards, in Campania and Calabria regions; artisans in the textile industry; health professionals; chefs and waiters catering in Campania and Lazio regions; cleaners; sailors; swimming pool lifeguards; and bus drivers. Tables 1 and 2 describe the population characteristics and their job assignment, respectively.
Sample population characteristics.a
BMI: body mass index.
a Age is indicated as mean value ± standard deviation, the other parameters as median with minimum and maximum value.
Sample population based on the work assignments.a
a Age is indicated as mean value ± standard deviation, height as median value.
Considering the averages of the FVC predicted values, the GLI reported a mean value 380 mL lower than the ECSC (4.52 L vs. 4.90 L, p < 0.0001) and 330 mL higher than the ERS (4.52 L vs. 4.19 L, p < 0.0001), with a significant difference in both cases. Analyzing the averages of the FEV1 predicted values, the GLI reported a mean value 90 mL lower than the ECSC (3.68 L vs. 3.77 L, p < 0.0001) and 190 mL higher than the ERS (3.68 L vs. 3.49 L, p < 0.0001), with a significant difference in both cases. Instead, analyzing the predicted FEV1/FVC ratio, the average according to the GLI was one point greater than the ERS (81.80% vs. 80.46%, p < 0.0001) and five points greater than the ECSC (81.80% vs. 76.89%, p < 0.0001), in a statistically significant manner. The results of the statistical comparison of predicted values, LLNs, and percentages of predicted values obtained through the three reference values using ANOVA test and post hoc LSD analysis are shown in Table 3.
Comparative analysis of the mean ± standard deviation of predicted values, LLNs, and percentages of predicted values assessed through the three reference values.
LLN: lower limit of normal; ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; FVC: forced vital capacity; FEV1: forced expiratory volume in one second; LSD: least significant difference.
a p < 0.0001, significant difference between ECSC 1971 and ERS 1993 (post hoc LSD analysis).
b p < 0.0001, significant difference between ERS 1993 and GLI 2012 (post hoc LSD analysis).
c p < 0.0001, significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
Concerning the mean percentages of predicted values, as calculated from spirometric results, the average FVC% according to the GLI was seven points significantly greater than the ECSC (91.09% vs. 84.23%, p < 0.0001) and seven points significantly lower than the ERS (91.09% vs. 98.35%, p < 0.0001). The average FEV1% according to the GLI was two points greater than the ECSC (94.12% vs. 92.19%, p < 0.0001) and five points lower than the ERS (94.12% vs. 99.11%, p < 0.0001), a significant difference in both cases.
The FVC%, FEV1%, and FEV1/FVC% of the predicted reference values were analyzed also with respect to age ≤40 or >40 years (Table 4). The 40-year cutoff was chosen because it was the mean age of the total investigated population. This subgroup analysis confirmed the significant differences between the three reference values determined in the total sample, with the exception of FEV1% that did not show significant difference between the GLI 2012 and the ECSC 1971 in workers ≤40 years (95.51% vs. 95.06%, respectively, p = 0.355). This latter result was not demonstrated in workers >40 years, in which a significantly higher FEV1% was determined with the GLI 2012 compared to the ECSC 1971 (92.41% vs. 88.69%, respectively, p < 0.0001), according to the total sample trend.
Comparison of the three reference values based on the age of the investigated population.a
ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; FVC: forced vital capacity; FEV1: forced expiratory volume in one second; LSD: least significant difference.
a Workers ≤40 years old: 1355 (55.04% of the total sample); 704 male (52%) and 651 female (48%). Workers >40 years old: 1107 (44.96% of the total sample); 858 male (77.5%) and 249 female (22.5%). b p < 0.0001, significant difference between ECSC 1971 and ERS 1993 (post hoc LSD analysis).
c p < 0.0001, significant difference between ERS 1993 and GLI 2012 (post hoc LSD analysis).
d p < 0.0001, significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
e p > 0.05, no significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
Also, when a gender-based analysis was performed (Tables 5 and 6), significant differences could be determined between the reference values for all the three parameters considered, with a trend comparable to that demonstrated for the total population.
Comparison of the three reference values in male workers according to the age.a
ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; FVC: forced vital capacity; FEV1: forced expiratory volume in one second; LSD: least significant difference.
a Male workers ≤40 years old: 704 (52% of the total workers ≤40 years). Male workers >40 years old: 858 (77.5% of total workers >40 years).
b p < 0.0001, significant difference between ECSC 1971 and ERS 1993 (post hoc LSD analysis).
c p < 0.0001, significant difference between ERS 1993 and GLI 2012 (post hoc LSD analysis).
d p < 0.0001, significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
Comparison of the three reference values in female workers according to the age.a
ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; FVC: forced vital capacity; FEV1: forced expiratory volume in one second; LSD: least significant difference.
a Female workers ≤40 years old: 651 (48%). Female workers >40 years old: 249 (22.5%).
b p < 0.0001, significant difference between ECSC 1971 and ERS 1993 (post hoc LSD analysis).
c p < 0.0001, significant difference between ERS 1993 and GLI 2012 (post hoc LSD analysis).
d p < 0.0001, significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
e p > 0.05, no significant difference between GLI 2012 and ECSC 1971 (post hoc LSD analysis).
In the total sample considered, it was possible to identify an obstructive deficit or a mixed pattern in 36 (1.5%), 56 (2.3%), and 78 (3.2%) spirometric tests according to the ECSC, ERS, and GLI equations, respectively. The detected differences were significant both at the testing group analysis (p value ANOVA <0.0001) and when individual groups were compared between each other. All the obstructive deficits/mixed patterns present according to the ECSC and the ERS were confirmed by the GLI (Table 7).
Distribution of the obstructive/mixed spirometry according to the different reference values considered.a
ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; LSD: least significant difference.
aOne-way ANOVA was performed to test the significance of differences between the number of obstructive/mixed tests in the three groups.
b p <0.0001, significant difference between the number of ECSC 1971 obstructive/mixed tests and those detected by the ERS 1993 (post hoc LSD analysis).
c p < 0.0001, significant difference between the number of ERS1993 obstructive/mixed tests and those detected by the GLI 2012 (post hoc LSD analysis).
d p < 0.0001, significant difference between the number of GLI 2012 obstructive/mixed tests and those detected by the ECSC 1971 (post hoc LSD analysis).
In the entire sample population, there were 1072 (43.5%) tests interpreted as restrictive patterns using the ECSC, 367 (14.9%) according to the GLI, and 161 (6.5%) for the ERS, with a statistically significant difference both at the testing group analysis and when individual groups were compared between each other (Table 8).
Distribution of restrictive patterns according to the three different reference values considered.a
ECSC: European Coal and Steel Community; ERS: European Respiratory Society; GLI: Global Lung Initiative; ANOVA: analysis of variance; LSD: least significant difference.
aOne-way ANOVA was performed to test the significance of differences between the number of restrictive patterns in the three groups.
b p <0.0001, significant difference between the number of ECSC 1971 restrictive patterns and those detected by the ERS 1993 (post hoc LSD analysis).
c p < 0.0001, significant difference between the number of ERS1993 restrictive patterns and those detected by the GLI 2012 (post hoc LSD analysis).
d p < 0.0001, significant difference between the number of GLI 2012 restrictive patterns and those detected by the ECSC 1971 (post hoc LSD analysis).
Discussion
This is the first study that simultaneously compares the ECSC 1971, the ERS 1993, and the GLI 2012. In fact, several studies have compared the ERS 1993 with the GLI 2012 (Baur et al., 2016; Liistro et al., 2017; Quanjer et al., 2013) and only one Italian investigation compared the ECSC 1971 with the GLI 2012 (Innocenti et al., 2014).
Significant differences have been detected for all the parameters investigated, FEV1, FVC, and FEV1/FVC, both in terms of predicted values, LLNs, and percentages of the predicted values, with respect to the different reference values employed. This underlines how a non-correct choice of reference values may lead to a misleading interpretation of the spirometric examinations.
The GLI showed the highest frequency of obstructive diagnosis in the analyzed sample. In fact, these reference values were able to confirm all the obstructive tests found through the ECSC and the ERS, but also to identify obstructions considered normal for the ERS and normal or restrictive for the ECSC. Considering restrictive patterns, a notable overestimation was evident in our sample when the ECSC were employed with respect to the other two reference values. Conversely, the ERS lost an important percentage of suspected restrictions, instead identified by the GLI.
The differences between the ERS and the GLI, emerged in our findings, are similar to those reported by Baur et al. (2016), who showed a good correlation in the diagnosis of obstructive deficits and a higher frequency of restrictive patterns using GLI, and Liistro et al. (2017), who found a mean FEV1% four points greater in ERS. Our study also confirmed the data of Quanjer et al. (2013), in which 1- to 2- and 5- to 10-point greater percentages of obstructive deficits and restrictive patterns, respectively, could be detected by the GLI with respect to the ERS. Conversely, comparing the ECSC with the GLI, our results are different from Innocenti et al. (2014), who found an agreement between the two reference values and recommended the use of the ECSC.
However, this study has some limitations to be considered for a correct interpretation of the results. The first one relates to the intrinsic nature of the investigation. In fact, this is a retrospective study that analyzes spirometry tests performed by several operators over a long period of time, thus full uniformity in procedures could not be guaranteed. Additionally, restrictive and mixed patterns could be only hypothesized and not confirmed, as the simple spirometries employed did not provide static volume measurements (i.e. RV, TLC) that are necessary to diagnose restrictive deficits. This aspect prevents to extrapolate definite conclusions on such alterations, and require further research.
In conclusion, our investigation pointed out some interesting differences in the interpretation of the spirometric tests according to the variable reference values adopted in the OM practice in Italy. In fact, we could demonstrate that, compared to the GLI 2012 equations, the ERS 1993 provided an underestimation of the presence of obstruction and restriction, while the ECSC 1971 showed a considerable overestimation of the presence of restrictive patterns. This study therefore demonstrates the importance of a suitable choice of spirometric reference values for a correct interpretation of the PFTs performed during the health surveillance programs and the need for a validation study of the GLI 2012 in the Italian worker population to obtain reliable guidance on their use in OM practice.
Footnotes
Acknowledgements
The authors would like to thank Dr Fabio Sito and Dr Bartolomeo Boggia for providing data on spirometry carried out during their health surveillance activity.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
