• A uniform format for the presentation of PFT results in reports to users and in the medical record can reduce potential miscommunication or misunderstanding.

•  ∘ Only information with validated clinical application should be included.

•  ∘ The normal limit(s) of each test parameter should be displayed.

•  ∘ Consistent with other laboratory values, the measured value should be shown before reference values, ranges, or normal limits.

•  ∘ Report and/or display of the displacement of the result from a predicted value in standard deviation units (z-score) can help in understanding abnormality.

• For spirometry, many parameters can be calculated but most do not add clinical utility and should not be routinely reported.

•  ∘ Only FVC, FEV₁, and FEV₁/FVC need be routinely reported.

•  ∘ Measurement of slow VC and calculation of FEV₁/VC are a useful adjunct in patients with suspected airflow obstruction.

•  ∘ Reporting FEV₁/FVC (or FEV₁/VC) as a decimal fraction, and not reporting it as a percentage of the predicted value for this ratio, will help to minimize miscommunication.

• Lung volumes

•  ∘ The nitrogen washout plot for multibreath tests and the tracings for plethysmograph tests can be shown graphically to aid quality assessment.

• For diffusing capacity the report is consistent with the 2017 European Respiratory Society (ERS)/ATS Technical Standard for this test.

•  ∘ Barometric pressure should be measured and reported and the measured value corrected to the standard pressure of 760 mm Hg.

• Newer collated reference equations for spirometry and diffusing capacity have been developed since prior ATS documents and warrant wide implementation.

•  ∘ The Global Lung Function Initiative (GLI)-2012 multiethnic spirometry reference values are recommended for use in North America and elsewhere for the ethnic groups represented. Their smooth continuity throughout growth is advantageous for laboratories testing children or adolescents.

•  ∘ The National Health and Nutrition Examination Survey (NHANES) III reference values (recommended for North America in 2005 ATS/ERS documents) remain appropriate where maintaining continuity is important.

•  ∘ Regardless of the reference source or lower limit of normal (LLN) chosen, interpreters should be aware of uncertainty when interpreting values near any dichotomous boundary.

•  ∘ For lung volumes and diffusing capacity, no prior ATS recommendation has been made because of the wide divergence of available reference values. A large compilation of international data has been completed for the diffusing capacity of the lung for carbon monoxide (Dl_CO) and is underway for lung volumes. The resulting reference equations should be widely adopted when published.

• Pulmonary function tests that fail to meet optimal standards may still provide useful information. A grading system for test quality can allow for this use, while providing an indication of the uncertainty imposed, and is most helpful if widely standardized.

•  ∘ For spirometry, FVC and FEV₁ are graded separately on an A–F scale either manually or by software. There is evidence that grades A–C are clinically useful, whereas grades D and E may have limited value, and grade F should not be used. The same scale, with different criteria values, is used for children.

•  ∘ For diffusing capacity a similar grading scale is presented on the basis of 2017 ERS/ATS standards.

The following recommendations and rationale are based on developing a format that will be intuitive, will include only information with validated clinical application, will be based on the use of the LLN, and will be consistent with prior recommendations for PFT interpretation and reporting (1–4). Some recommendations are necessarily arbitrary (e.g., the order of rows or columns) but reflect a consensus of current and prior committee members and an informal survey of others (5). Although individual preferences vary, there is wide agreement that the benefit of uniformity outweighs these.

The report format recommended is presented in test-specific units that can be assembled into a report appropriate for a laboratory’s practice or even an individual test session. It is designed so that for simple testing it can be printed, along with interpretive comments, on a single page as a report to a referring physician or for inclusion in the medical record. Of necessity, this contains limited information and is not intended as the only resource for the interpreter, who should have the option of displaying all individual maneuvers from a given PFT session, increasingly done on digital systems. Standardized electronic formats for the saving of all PFT data, including each individual maneuver, are being recommended (6). This will allow reviewers the flexibility to see additional detail or to reanalyze previous PFTs or apply new reference values as they become available. A standardized methodology to incorporate PFT data into electronic medical records is needed, but is beyond the scope of this report. See Appendix EA in the online supplement for a suggested list of test results to save to the electronic medical record.

In designing the standardized report, the committee recognized that aspects of data presentation can affect decision-making (7). The use of boldface or colored fonts to highlight measured values below the LLN can draw attention to these, but imposes a binary decision on a continuous variable. The number of variables reported can also have an impact because including a large number of outcomes in the report increases the statistical likelihood of one falling below an arbitrary LLN, with the risk of a false positive result (8).

All reports must begin with unambiguous patient identification, including patient name, medical record number, sex, and date of birth; the latter can be compared with previous records as a check for possible identification errors, as well as for calculating patient age (year to one decimal place for children and adolescents, e.g., age 6.3 yr) (9, 10). Other essential information is height (to the nearest centimeter) and weight, ethnicity, and date of the test. Other useful information includes smoking history, reason for the test, and referring physician’s name. Additional information may include oxygen saturation and barometric pressure.

The display will vary with the testing done, but the suggested order is spirometry, slow vital capacity, and/or lung volume measurement, and diffusing capacity of the lung for carbon monoxide (Dl_CO). Other tests could be added such as forced oscillometry, maximal respiratory system pressure, levels of expired nitric oxide, or other tests, but the philosophy should be similar, that is, reference source, normal limits, graphs that convey quality information, and exclusion of information without clinical value.

The recommended order of the columns in tabular data is the actual value, the LLN, the z-score (optional), and the percent predicted value. The predicted value itself is unnecessary, as it does not aid in the interpretation of abnormality. The z-score of a result is the number of standard deviations it lies away from the mean or, for regression equations, the number of standardized residuals away from the predicted value. Linear graphical displays visualize this in relationship to the normal range and assist in assessing the significance of abnormal values (11, 12). (If newly introduced to the reports, adding a brief explanation may be helpful.) The reference source from which the LLN and percent predicted value are derived must be listed, and whether or not these are adjusted or specific for race/ethnicity must also be stated in technician comments.

As shown in Figure 1, numerical values are given only for the FEV₁, the FVC, and the FEV₁/FVC ratio; the latter should be reported as a decimal fraction and the space for percent predicted value left blank to minimize miscommunications. When appropriate, an additional row can be added for FEV₁/(slow) VC (1, 2). Forced expiratory time (FET) is reported to aid quality assessment. If bronchodilators are given, the LLN column need not be repeated; the absolute and percent change should be given only for FEV₁ and FVC. Other numerical values such as the forced inspiratory flow at 75% of FVC (FEF_75%) and FEF_25–75% have not demonstrated added value for identifying obstruction in adults or children, and therefore are not recommended for routine use (13, 14). The flow–volume curve and the volume–time curve are displayed, from which the peak flow and FET can be seen. These graphs must have sufficient resolution to evaluate the quality of the data. For the volume–time curve, the volume scale should be at least 10 mm/L, the time scale at least 20 mm/s, and 1 second prior to the start of expiration should be displayed (2). On the flow–volume plot, the flow display should be at least 5 mm/L/s, and the ratio of flow to volume should be 2 L/s to 1 L. The scales of the graphs may be adjusted to maximize the image within the available space on the report form, especially for tests on small children. The linear analog scales, where the values for FEV₁, FVC, and their ratio are plotted as z-scores relative to the predicted value (z = 0), give an intuitive sense of severity (12). Because there is always some uncertainty about the application of any prediction to an individual and about the exact LLN, a large star rather than a discrete point is used on the scale to suggest that caution is indicated when interpreting values close to the LLN.

Figure 1. Example of a single-page report for pre- and postbronchodilator spirometry testing. The linear graphic is divided in units of 1 SD, with the LLN shown at a z-score of −1.64. This simplified report is suitable for the medical record or referring physician, but the test interpreter should have access to the data and curves of all acceptable spirometry efforts. FET = forced expiratory time; GLI = Global Lung Function Initiative; IC = inspiratory capacity; LLN = lower limit of normal; Sp_O2 = oxygen saturation as measured by pulse oximetry; ULN = upper limit of normal; VC = vital capacity.

Download Figure | Download Powerpoint

[More] [Minimize]

For slow vital capacity, the graph shows baseline tidal breathing to assess whether inspiration occurred from a stable end-expiratory volume (3). The largest vital capacity is reported along with the inspiratory capacity and, when appropriate, the FEV₁/VC.

Values derived by body plethysmography or gas dilution are displayed with the same column order (Figure 2). We show a full complement of volume parameters listed in a physiologically rational order; however, some laboratories may choose not to report all. With a multibreath nitrogen (N₂) washout the graph of the fall in N₂ concentration gives an indication of any leaks present (3). For helium dilution functional residual capacity (no graph displayed), equilibration is considered to be complete when the change in helium concentration is less than 0.02% for 30 seconds. The histogram displays the actual volume increments beside the predicted volumes as an indication of severity, and z-scores are shown here in a vertical format. When diffusion capacity is measured, a comparison of total lung capacity measured by both techniques can be a useful quality control measure or an indication of maldistribution.

Figure 2. Examples of the recommended reporting format for lung volume testing in one subject by multibreath nitrogen (N₂) washout and in another subject by plethysmography. The N₂ washout is plotted on a log scale, resulting in a nearly linear profile. On this plethysmography tracing the box pressure has been converted to volume to show the thoracic excursions at a scale of 20 ml per division. The bar graphs on the right show the predicted and observed values of RV, FRC, and TLC, and the arrows show these results in relation to their normal range on vertical linear scales. The graphs depict RV in blue, ERV in orange, IC in gray, and normal range in green. (See Figures E1 and E2 in the online supplement for examples of consolidated reports for full pulmonary function tests.) ERV = expiratory reserve volume; IC = inspiratory capacity; LLN = lower limit of normal; mb = multibreath; pl = plethysmography; QA = quality assurance; RV = residual volume; TLC = total lung capacity; ULN = upper limit of normal; VC = vital capacity; Vtg = thoracic gas volume.

Download Figure | Download Powerpoint

[More] [Minimize]

The display (Figure 3) gives the relevant values, the LLN, and the percent predicted value along with the reference source, a quality assurance indication, and the conditions of the test, in this case post-bronchodilator. The barometric pressure should be given, as well as stating whether the values were corrected to standard barometric pressure (particularly important for laboratories at altitude) (6). Reporting the carbon monoxide transfer coefficient (Kco) is optional, but the term Dl/Va (the ratio of diffusing capacity to alveolar volume) should be avoided as it is commonly misunderstood. If measured, the hemoglobin should be shown as well as the adjusted predicted values for both Dl_CO and Kco. The display shows the washout of both carbon monoxide and the tracer gas and the sample volume. The sample volume is “virtual” and derived from rapid-acting analyzers and can be adjusted for the size of the patient after the maneuver. However, the display must show the sample volume that was used in the calculation (4). Again, there is a linear graph where the result is plotted as z-scores away from the predicted value.

Figure 3. Example of the recommended reporting format for the single-breath diffusing capacity test. The 2017 Technical Standard for Dl_CO (6) requires that the CO and tracer gas concentrations be graphed versus exhaled volume, rather than versus time as shown here. When hemoglobin is measured, it should be shown on the report with a note indicating whether the predicted value has been adjusted for this. CH4 = methane (tracer gas); Dl_CO = diffusing capacity of the lung for carbon monoxide; GLI = Global Lung Function Initiative; Kco = carbon monoxide transfer coefficient; LLN = lower limit of normal; Pb = barometric pressure; pred adj Hb = hemoglobin adjusted for predicted value; sb = single breath; TLC = total lung capacity; Vi/Vc = inspired volume/vital capacity.

Download Figure | Download Powerpoint

[More] [Minimize]

There is a place for technician comments on the test session, any quality issues, and other relevant information that may aid in interpretation. The accompanying figures show how the format units may be combined into one-page reports for both spirometry (Figure 1) and more complete test sessions (Figures E1 and E2). The one-page form has space for only a brief interpretive summary, and therefore laboratories preferring a more detailed interpretation and/or less crowded components may opt for a two-page report. Test tracings are shown for spirometry and other tests as well, but some laboratories may choose not to include all of these for the end user as long as the interpreter has full access to them. Although comparison with prior values is important to interpretation, a format for this is not addressed in this document.

Interpretation of PFTs requires comparison with reference values because lung function depends on body dimensions and physiological changes throughout growth and aging. Reference equations use such factors as height, age, sex, and race/ethnicity to predict the average lung function as well as the range of expected values, with the goal of distinguishing the effects of disease from normal variability among healthy individuals.

PFT laboratories must select appropriate reference values for the patients being tested. These should be generated from high-quality data collected from a large sample of healthy asymptomatic individuals who have never smoked or been affected by other respiratory illness or significant exposures. Not all published reference sources meet these criteria; therefore careful consideration of the advantages and disadvantages of available reference values is necessary.

Lung function reports should identify the source of reference values, because the same measured values may be interpreted differently based on the reference source used. Furthermore, manufacturers need to be transparent when reference values have been combined from various sources. In the event that a laboratory changes its selected reference equations, this should be noted on the report, and percent predicted values for prior lung function data should be recalculated, if possible. It is preferred that there be no discontinuity between pediatric and adult equations in the reference values selected, and extrapolation of values beyond the age range of the equations should not be done during growth and will increase uncertainty in the elderly (9, 15). Any such extrapolation must be noted in technician comments.

In 2005, the 1999 NHANES III spirometry reference equations (16) were recommended for use in North America (1). This study provided values for whites, African Americans, and Mexican Americans living in the United States. The age span was 8–80 years in two sets of equations with a break at age 18–20 years; a separate recommendation was made for children under age 8. As there was uncertainty whether these equations were a good fit for various European populations, no recommendation was made.

Subsequently, the GLI group was formed with ERS sponsorship and with the participation of the ATS PFT Committee. The goal was to merge available data sets, including the NHANES III data, to develop more broadly applicable reference equations. This effort resulted in new spirometry reference equations using data collected from more than 74,000 individuals, ages 3–95 years, from 26 countries (12). The GLI established reference values for whites, African Americans, North East Asians, and South East Asians. The equations for the white population were shown to be applicable not only in the United States and Europe but in other parts of the world, including Hispanic regions, and for Hispanic Americans. These findings confirm a reanalysis of the NHANES III data, which showed no need for separate reference equations for Hispanic and non-Hispanic whites (17). For individuals not represented by these four groups, or who are of mixed ethnic origin, a composite equation is provided. (See Appendix EB for guidance on the choice of equations.) The GLI data found the FEV₁/FVC ratio to be generally independent of ethnic group, and thus its LLN is a useful indication of airflow limitation even when ethnicity is uncertain.

More recently, the Canadian Health Measures Survey has published population-based spirometry reference equations for whites, using a format similar to NHANES III (18). Average adjustment factors are given for several indigenous and immigrant groups. Of note, they found that individuals of Chinese ancestry living in Canada had values intermediate between white values and those predicted by GLI equations from data collected in China.

Since the GLI-2012 publication, these white reference values have been compared with those of NHANES III. In large clinical populations from Australia and Poland, the values predicted by GLI-2012 and NHANES III were similar and rates of airflow limitation (FEV₁/FVC < LLN) were similar in both men (GLI, 34.5 and NHANES III, 33.3%) and women (GLI, 27.9 and NHANES III, 25.4%) (19). Similar findings have been demonstrated in additional clinical populations from Australia (20), the United States (21), and in children and adolescents (22). The Canadian predicted values compared somewhat more closely with GLI-2012 than NHANES III values, but differences among the three were minor and not likely to be clinically important (18).

In a simulation of NHANES III and GLI-2012 predicted values across a broad range of age and height, the FEV₁ prediction differences were within the recommended repeatability criterion of ±150 ml across a wide range of heights and ages (21). There were larger differences at the extremes of height in older individuals, where more uncertainty would be expected due to relatively few subjects of advanced age in the GLI-2012 data and extrapolation of the NHANES III data beyond the age of its subjects.

Both the ATS and ERS recommend the use of the LLN, or the upper limit where appropriate (e.g., lung volumes), to delineate between health and suspected disease. These are set at the fifth percentile (equivalent to a z-score of −1.645) so that 95% of a healthy population falls within the normal range and the lowest 5% would be false positives. However, clinical PFTs are typically done when disease is suspected, increasing the pretest probability of an abnormal result so that the false positive rate is much lower in this setting. The LLN does not necessarily need to be the fifth percentile but, with adequate outcome data, could be adjusted higher when the pretest probability is high or lower for population screening (23, 24). More important than the applicability of a particular LLN is recognition of the uncertainty that lies near any dichotomous boundary and where caution is required, especially when results are limited to a single test occasion.

The respiratory community is familiar with using the percent predicted value to describe lung function results; however, this value should not be used to define abnormality. The true LLN is age- and/or height-dependent and therefore will occur at varying percent values in different individuals. The fixed values commonly used (e.g., 80% predicted for FVC, 0.70 for FEV₁/FVC) are estimates based on middle-aged adults, and therefore erroneous clinical decisions based on these fixed cutoffs are more likely to occur in children and in older or shorter adults. Using fixed cutoffs also introduces a sex bias into clinical assessments (25). Interpretation of individual results relative to the range of values expected can be more appropriately incorporated into PFT reports using the recommended linear analog scale.

For spirometry, the GLI-2012 reference values are recommended for use in North America for the ethnic groups represented, as well as in Europe, Australia–New Zealand (26), and other areas with represented populations. For laboratories wishing to maintain continuity, the NHANES III equations also remain recommended for whites (including Hispanics) and African Americans. Use of GLI-2012 is recommended for clinical research studies to facilitate comparisons with international studies (27, 28). The GLI-2012 equations are also preferred for laboratories testing children or adolescents because they permit tracking during this time of rapid lung growth and development without discontinuities due to switching reference sources. The Canadian reference values also provide a useful resource (18). Whatever reference source is used, interpretations must be based on a parameter-specific lower limit determined from the distribution of the reference data.

For Dl_CO, no prior ATS recommendation has been made because of the wide divergence of available reference values. With ATS and ERS sponsorship a GLI group has assembled data from more than 12,000 individuals in 14 countries to develop new (white-only) reference equations from age 5 to 85 years. Publication of these is expected in 2017 (29) and their rapid adoption is recommended.

For plethysmographic or dilutional lung volumes, no recommendation can be made for reference values at this time. A new international project to address this need is underway. (Values from a Canadian study [30] are used as examples in the accompanying figures.)

Whereas considerable attention has been given to guidelines for the procedures to conduct spirometry (2), guidelines to assess the quality of the testing are still needed. The purposes of quality review are to provide feedback to the technicians and, in the clinical setting, to indicate any limitations to the interpretation of the results. In clinical research, quality review helps determine whether a subject can be included in a trial and whether data at any time point can be used in the analysis.

Various quality-grading systems have been reported (31–38) and others have been provided by spirometer manufacturers, but users would benefit from standardized methodology. The system recommended for adults and children is shown in Tables 1 and 2. It is modified from a system that has been used in research and epidemiological studies (16, 18, 38). For younger children, 2–6 years of age, the criteria are modified on the basis of the 2007 ATS/ERS recommendations for spirometry testing in preschool children (39). These systems can be used manually, or as part of spirometry software, assigning a grade (A through F) separately for the quality of FVC and FEV₁. In general, tests with a grade of A, B, or C are usable; tests with grade D are suspect; tests with grade E might be used by the interpreter only to show values “within the normal range” or “at least as high as,” without demonstrated repeatability; and tests with grade F should not be used.

The grading system consists of acceptability and repeatability components. An ideal test session conforms to prior ATS/ERS recommendations (2, 39) with at least three acceptable maneuvers and repeatable FVC and FEV₁ values. These criteria were intended to guide technicians to achieve the best possible results, and the goal should be to exceed them because many technicians can achieve better quality tests. However, their strict application may also lead to the exclusion of useful results. A grading system allows the user to evaluate the likelihood that spirometry results are representative of true values in the face of test performance that is not ideal. Failure to achieve optimal tests may be due to underlying disease; thus bias may be introduced into clinical research studies by eliminating subjects with more severe disease. The reviewer must consider whether or not the subject’s effort in a maneuver was maximal, and/or whether lack of repeatability could be due to lung disease, using all available information including technician comments. When a test session with a poor-quality grade is used as a baseline for pre/post-bronchodilator or longitudinal comparisons, an apparent improvement in values may be the result of better effort or technique

While strict application of the grading criteria can be done by computer software, the reviewer’s role is to apply judgment by reviewing the individual curves, which may change the scoring and allow interpretation. For example, if a maneuver in the session is unacceptable only because of excessive back extrapolation volume, it can still be used to confirm repeatability of FVC. Determining whether the subject has exhaled completely is difficult, and one report has suggested that many subjects are unnecessarily excluded by the 2005 ATS/ERS end-of-test criteria (38). The FET is used to determine whether the subject has tried to exhale long enough, and the end of the volume–time curve is assessed to determine whether expiratory flow has ceased, defined as a volume change less than 0.025 L in 1 second. Often the FET is less than 6 seconds, because the software has stopped data collection once this low flow criterion is met, or an artifact during exhalation may be falsely perceived as a plateau, thus stopping data collection and underestimating FET and FVC. Subjects should be verbally encouraged to continue the expiratory effort at the end of the maneuver to obtain optimal results (2). The technician and/or reviewer must make a determination as to whether or not end-of-test criteria were met and whether the data are useful. Some subjects, especially children and adolescents, cannot exhale for the required 3 or 6 seconds. If these subjects have a 1-second plateau, and the reviewer judges that these maneuvers represent a maximum FVC, the grade should be adjusted higher. Subjects with airflow obstruction may never reach a plateau even at the suggested 15-second limit and may have nonrepeatable FVC values only because of a difference in FET. Similarly, subjects with restrictive lung disease may reach an early plateau and may not be able to maintain a 6-second effort. The results may still be considered acceptable in such cases and an appropriate comment by the reviewer should be made.

FEV₁ is graded separately because even test efforts that are clearly unacceptable for FVC may contain a valid measurement of FEV₁ (or FEV_0.75, the forced expiratory volume exhaled in the first 0.75 s of the FVC maneuver, and a recommended measure in preschool children) (2, 39). Anything that occurs later in the flow–time tracing (e.g., early termination, cough artifacts) does not affect the FEV₁ or FEV_0.75 value, and thus these may be used even with end-of-test errors.

Despite the attention paid to expiratory parameters, the most common reason for low FVC, FEV₁, and PEF values is an incomplete inhalation. Achievement of maximal inhalation is best assessed by measures of repeatability and also by the consistency of the shape of the flow–volume or volume–time curve (40).

FVC maneuvers that have lower PEF values compared with others in a session may produce higher FEV₁ values due to the negative effort dependence of flow. The technician should coach forceful initiation of the FVC maneuver (i.e., blast), to achieve the fastest/highest PEF (41). Rounded peaks on the flow–volume curve may reflect submaximal blasts that can increase variability in both FVC and FEV₁.

Quality review of lung volume measurement is more challenging as a variety of methods are available, including body plethysmography, nitrogen washout, helium dilution, and radiographic imaging. We are not aware of any quality-grading systems that have been validated for the measurement of absolute lung volumes. Until a validated system is available, we recommend adherence to the 2005 ATS/ERS recommendations for the measurement of lung volumes (3). If the acceptability and/or repeatability criteria are not met but these data are reported, a comment should be included to caution users of the test results.

The Dl_CO (or Tl_CO) test is complex, involving a number of technical factors, and variability can be high. In the absence of quality-grading systems that have been validated using the new recommendations (6), a grading scheme based on maneuver acceptability is proposed (Table 3). The average Dl_CO value from at least two grade A Dl_CO maneuvers that are repeatable within 2 ml/min/mm Hg, or 0.67 mmol/min/kPa, should be reported. If only one grade A maneuver is acquired, the Dl_CO value from that maneuver is reported. If only maneuvers with grades B to D are available, Dl_CO values from these might still have clinical utility; therefore the average of the two best graded of these maneuvers should be reported, but this must be noted to caution the interpreter of the test results. Maneuvers with grade F are not usable.

In the clinical pulmonary function laboratory, technicians, supervisors/managers, or computer software can assign quality grades, whereas in the clinical research setting, an independent quality reviewer is commonly used. This reviewer should be an expert in pulmonary function testing and have extensive experience, both in direct testing and in monitoring testing performed by others. If more than one reviewer is used, comparisons across reviewers should be done to ensure consistency, for example, with a blinded sample of good and bad test sessions.