Author: Richard Johnston

One of the overlooked parts of teaching pulmonary function interpretation is developing an appreciation for the number and variety of errors that the equipment, patients and technicians can produce and how they affect the reported test results. I routinely run across a couple dozen errors each week while reviewing reports. Most are minor and do not significantly affect the reported results. Many are mundane because they appear so often and a few are interesting because they point out a particular limitation in the equipment, software or testing standards. I’ve kept a file of the more iconic examples of testing errors for years and a while ago a pulmonary staff physician and I used to hold weekly sessions for fellows and residents where we’d present a number of “zingers” to see if they could figure them out. Unfortunately that physician has moved on to a different institution and I’m no longer as available as I used to be so these sessions are no longer held but I think that they or something like them should be held in all teaching hospitals.

These spirometry results came from a middle-aged woman with sarcoidosis.

	Observed:	%Predicted:	Predicted:
FVC:	3.55	155%	2.29
FEV1:	1.06	60%	1.77
FEV1/FVC Ratio:	30	39%	77

Elevated FVC’s are not all that uncommon (and are a good example of the limitations of reference equations), but one that is 155% of predicted is particularly unusual. This occurs most commonly when somebody has made an error in measuring or entering the patient’s height (I can’t tell you the number of times I’ve seem someone entering 60 inches when they meant 6 feet), but this patient has been seeing pulmonary physicians and having regular spirometry tests for over a decade and the height for this test was the same as it was for the previous visit. In addition the trend report showed that over the last year the patient’s FVC had been between 71% and 65% of predicted.

The flow-volume loop doesn’t look overly unusual although the expiratory flow doesn’t taper off to zero and the patient maintained a low expiratory flow for at least two-thirds of the vital capacity.

The volume-time curve is actually quite unusual and shows a flat section where the exhaled volume steadily increased for over 10 seconds. Most volume-time curves approximate an exponential decay curve which looks quite different. The volume-time curve does flatten at the end, but flattening is abrupt.

The final clue was in the technician’s notes:

“Dx: Sarcoid. Medication: Albuterol, none recently. 2 LPM supplemental O2, transtracheal. Good test quality and reproducibility.”

The majority of patients that receive supplemental O2 get it via a nasal cannula. The patient had been on supplemental O2 for the last couple of years but although her spirometry results aren’t too terrible, her disease process had continue to advance. In between the last visit and the current visit the patient had been fitted with a transtracheal catheter and was now receiving her supplemental O2 through that.

The problem was that during spirometry the supplemental O2 has not been turned off and that oxygen was being added to the patient’s trachea throughout the maneuver. This caused her FVC (and to a much lesser extent her FEV1) to be significantly overestimated. Interestingly, the patient performed four maneuvers and the FVC and FEV1 for three of them met the ATS/ERS criteria for repeatability.

This is a simple procedural error but there were at least two opportunities for the technician to have recognized there was a problem. First, any FVC that is 155% of predicted should be throwing up all sorts of warning flags. Admittedly only a tiny fraction of our patient’s have transtracheal catheters but there are any number of other reasons why that large of an FVC means that there has been error of some kind. Second, the fact that the FVC was at least twice as large as any of the FVCs the patient had performed in the last year and that should be a warning flag all by itself.

The technician that performed the test later said they had never had a patient with transtracheal O2 before and that may well be true. That doesn’t explain the “good test quality” comment in the notes but I think that all too often technicians (and really this applies to all of us) are on “autopilot” and aren’t paying as much attention as they should, particularly when it comes to entering notes.

There are numerous textbooks and manuals that teach PFT interpretation. The possible mechanical, software, procedural, patient and technician errors that can occur with testing is often only marginally discussed when it is discussed at all. To some extent this is understandable because error detection is often far more complex than the actual interpretation process. The most critical component of interpretation that needs to be taught however, is skepticism. When reviewing PFT reports the first step should always be to ask what’s wrong with the results and only you’ve been assured they are reasonably correct or that you know which direction the errors are biased towards should an interpretation be attempted.

Note: One reason that none of computer algorithms for interpreting PFTs have ever been reliable is that they all assume that the test results are completely accurate. Although I doubt that all errors could ever be automatically detected there are numerous routine errors that could be. Sadly however, I’ve never seen any attempt to do this by either the developers of algorithms or by equipment manufacturers.

But even with a sense of skepticism it can often be hard to find the evidence needed to determine that an error is present. Over the last year I’ve had the opportunity to take a look at the on-line signing applications from three of the four largest PFT equipment manufacturers in the USA. There are differences in each approach to on-line signing but one feature they share is that the signing physician can only view the final report. None of them have any easy way for a signing physician to review raw test data or to compare the multiple efforts a patient has attempted and some actually make this quite difficult (I won’t say impossible but having to switch to a different computer and a different application is certainly setting the bar high).

Physicians that use these on-line signing systems are therefore dependent on the choices made by PFT lab staff. To one degree or another this has always been true, but I would have thought that the ability to quickly and easily drill down into the patient’s test data would have been considered to be a critical part of on-line signing. Perhaps at some point it will be but until that time the recognition of errors and the selection of the most accurate patient results most often rests in the hands of the technicians performing the tests.

I used this particular example with at least two dozen groups of fellows and residents over the course of a couple years. So how well did they do at recognizing a problem with this patient’s test? Most of residents didn’t think there was a problem. Most of the pulmonary fellows realized there was a problem but the majority or them thought that the patient height or gender was incorrect and a few thought it might be an equipment problem of some kind. Nobody ever guessed that it was because of transtracheal O2.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

The thyroid gland is located across the front of the upper airway a short distance below the larynx. An enlarged thyroid gland is known as a goiter. The most common worldwide cause of goiter is an iodine deficiency. This is much less common in the western nations where factors such as Hashimoto’s thyroiditis, Graves’ disease, multi-nodular thyroid disease, thyroid cancer, pregnancy and the side effects of some medications are the its primary causes. Common respiratory complaints associated with goiter include cough, hoarseness, shortness of breath and stridor.

When a goiter is large enough it can press against the trachea and cause a narrowing or deviation of the upper airway. My lab usually gets at least a couple of patients referred to us every year with a diagnosis of goiter and a request that we assess whether it is causing any significant airway obstruction. Decades ago I was taught by my medical director that when this occurs it shows up as an expiratory plateau on a flow-volume loop.

The reality (as usual) is more complex and this is mostly because the thyroid gland lies close to the boundary between the extrathoracic and intrathoracic sections of the trachea. Depending on its size and the which direction the thyroid expands towards, goiter can show up as an extrathoracic or intrathoracic airway obstruction. Even more importantly, as a recent article in Chest showed, the airway obstruction from goiter can be dependent on body position as well.

Note: Interestingly, most of the published studies of goiter and airway obstruction goiter are fairly evenly divided between researchers that detected primarily extrathoracic obstruction and those that detected primarily intrathoracic obstruction. To some extent this appears to be a case of finding what you’re looking for since in many cases only expiratory or only inspiratory parameters are reported.

If the obstruction is intrathoracic, flow limitation will show up as a plateau in flow during exhalation. If it is extrathoracic, flow limitation and a plateau will occur during inhalation. Given the position of the thyroid gland it is also possible for both intrathoracic and extrathoracic flow limitations to be present at the same time.

The effects that extrathoracic and intrathoracic airway obstructions have on flow-volume loops were extensively described by Miller and Hyatt decades ago and the flow-volume loops in their article have been reproduced numerous times since then. The reason for the different flow-volume loop contours is relatively simple however, and has mostly to do with the differences in pressure between the interior of the airway and the surrounding tissue.

During exhalation, the pressure surrounding the airways inside the thorax (intrathoracic) is higher than what is inside the airways and this causes these airways to be narrowed. At the same time, the pressure around the airways outside the thorax (extrathoracic) is lower than what is inside the airways and this causes the airways to widen.

During inhalation, the pressure differences are reversed and this causes the extrathoracic airway to narrow and the intrathoracic airway to widen. Forced inspiratory and expiratory efforts exaggerate this effect. Because a goiter (or tumor or other mass) can press against the airway, this can cause a narrowing that is greater than what is normally expected to occur. Depending on its location this will then show up as a flattening or plateau in either the expiratory or inspiratory portion of the flow-volume loop.

Note: Some goiters extend well into the thorax and sub-sternal goiters have been described in more than one study. These goiters can decrease lung capacity but they are usually detected by imaging (MRI, nuclear medicine uptake scans) and not by lung volume measurements. Changes in FVC, TLC and FRC pre- and post-treatment have occasionally been noted, however.

Flow-volume loops from individuals with goiter are often reasonably normal however but spirometry is also usually performed while sitting upright and a goiter may not narrow the airway enough to affect flow rates in this position. Several studies have shown that some individuals with goiter that had normal flow-volume loops in the upright position showed flattening when they were performed in the supine or lateral positions. Another recent study showed that an individual with goiter that had the Pemberton sign (a somewhat unusual condition where elevating both arms causes venous congestion and cyanosis) also had an expiratory plateau on their flow-volume loop only when their arms were raised.

Interestingly, research has shown that radiological imaging is an unreliable indicator for determining the presence of airway obstruction and flow-volume loops remain the best tool for this purpose. Researchers have used a variety of numerical results such as the MIFR (PIF), the FEV1/PEF ratio and MEF50/MIF50 ratio to assess inspiratory and expiratory flattening. The usefulness of these values is limited however, particularly because their normal variability is high. Visual inspection of the flow-volume loops still seems to be the most accurate way to assess for inspiratory or expiratory flow limitations.

Goiter continues to occur throughout the world for a variety of reasons and upper airway obstruction can be associated with it. One study of a relatively large number of patients referred for thyroid surgery showed that 33% had upper airway obstruction but the actual prevalence amongst all patients with goiter is unknown. For this reasons any individual with goiter, and most particularly those with any respiratory symptoms, should probably be assessed for upper airway obstruction and this should consist of a careful examination of the maximal expiratory and inspiratory flow-volume loops. Individuals with normal flow-volume loops in the upright position should probably also perform supine (and possibly right or left lateral depending on the position of the goiter) spirometry as well, particularly if they complain of orthopnea.

References:

Albareda M, Viguera J, Santiveri C, Lozano P, Mestron A, Bengoa N, Calvet R, Roger A, Pardillo D, Delgado E, Vila LI. Upper airway obstruction in patients with endothoracic goiter enlargement: no relationship between flow-volume loop and radiological tests. Eur J Endocrin 2010; 163: 665-669.

Bright P, Miller MR, Franklyn JA, Sheppard MC. The use of a neural network to detect upper airway obstruction caused by goiter. Am J Respir Crit Care Med 1998; 157: 1885-1891.

Geraghty JG, Coveney EC, Kiernan M, O’Higgins NJ. Flow volume loops in patients with goiters. Ann Surg 1992; 215(1): 83-86.

Gittoes NJL, Miller MR, Daykin J, Sheppard MC. Upper airways obstruction in 153 consecutive patients presenting with thyroid enlargement. BMJ 1996; 312: 484-485.

Karbowitz ST, Edelman LB, Nath S, Dwek JH, Rammohan G. Spectrum of advanced upper airway obstruction due to goiters. Chest 1985; 87(1): 18-21.

Meysman M, Noppen M, Vincken W. Effect of posture on the flow-volume loop in two patients with Euthyroid goiter. Chest 1996; 110: 1615-1618.

Miller RD, Hyatt RE. Evaluation of obstructing lesions of the trachea and larynx by flow-volume loops. Am Rev Respir Dis 1973; 108: 475-481.

Resende PN, Menezes MB, Silva GA, Vianna EO. Pemberton sign. A recommendation to perform arm elevation spirometry with flow-volume loops. Chest 2015; 148(6): e168-e170.

Rios A, Rodriguez JM, Galindo PJ, Cascales PA, Balsalobre M, Parilla P. Spirometric evaluation of respiratory involvement in asymptomatic multinodular goiter with an intrathoracic involvement. Arch Bronconeumol 2008; 44(9): 504-506.

Sterner JB, Morris MJ, Sill JM, Hayes JA. Inspiratory flow-volume curve evaluation for detecting upper airway disease. Resp Care 2009; 54(4): 461-466.

Sugapriya G. Pulmonary function studies in hyperthyroid females with goiter – before and after thyroid surgery. Int J Biol Med Res 2011; 2(3): 661-663.

Torchio R, Gulotta C, Perboni A, Ciacco C, Guglielmo M, Orlandi F, Milic-Emili J. Orthopnea and tidal flow limitation in patients with Euthyroid goiter. Chest 2003; 124(1): 133-140.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

The current ATS/ERS standards for a positive bronchodilator response are an increase in FEV1 or FVC of ≥ 12% and ≥ 200 ml. These standards are largely based on the ability to detect a change that is far enough above the normal variability in FEV1 and FVC to be considered significant. One problem with this is that the amount of variability that is considered to be “normal” is overly influenced by a relatively small number of subjects that have a high degree of variability.

At least one group of investigators has suggested that a way around this is to subject all of an individual’s pre- and post-bronchodilator spirometry to statistical analysis in order to determine their coefficient of variability. Once this is known, the pre- and post-bronchodilator efforts can be assessed as a group to determine whether whether there has been a statistically significant change. Using this approach they were able to show that a rather large number of subjects that did not meet the ATS/ERS criteria did have a statistically significant improvement in FEV1.

But an increase that is statistically significant or one that is greater than normal variability is not the same thing as clinical significance. Numerous investigators have noted that patient can have a post-bronchodilator clinical improvement as shown by a decrease in dyspnea or an increase in exercise capacity without any notable change in FEV1 or FVC. Clinical significance is hard to measure however, particularly since which criteria should be used to measure it are unclear.

Long-term survival is certainly clinically significant and a recent article in Chest (Ward et al) has linked the increase in post-bronchodilator FEV1 to this fact. What these investigators have been able to show was that individuals with a post-bronchodilator increase in FEV1 that was 8% of predicted or greater showed a significantly better long-term survival than individuals with a smaller increase.

Strictly speaking this study is not saying that an 8% increase in FEV1 somehow improves survival. What it is actually doing is separating individuals with asthma (which tends to have a reasonably good prognosis) from those with COPD and IPF (which tend to have a poor prognosis). Because this study only looked at an increase in FEV1 it also doesn’t mean that bronchodilators are of no value to patients with COPD (and possibly some with interstitial diseases since combined obstructive and restrictive disorders aren’t all that uncommon).

There is a qualitative difference between how a post-bronchodilator increase that is 8% of the predicted FEV1 and one that is 12% of the observed FEV1 are assessed.

When 8% of predicted is used as a threshold for reversibility the post-bronchodilator volume increase that is considered significant is the same for all observed FEV1s. This means that individuals with an observed FEV1 less than 67% of predicted (which because it is actually a ratio 67% applies to all ages and both genders) and a post-bronchodilator volume increase greater than 200 ml, a 12% increase in observed FEV1 will be significant when an 8% increase of predicted is not. On the other hand, individuals with an observed FEV1 greater than 67% of predicted and post-bronchodilator increase of 8% of predicted will be significant when a 12% increase is not. Basically this means than the closer an observed FEV1 is to the predicted FEV1, the smaller the relative post-bronchodilator change in FEV1 needs to be for it to be considered significant. This in turn means that more individuals with mild airway obstruction are likely to be considered reversible while individuals with moderate to severe airway are obstruction are going to be less likely to be considered reversible.

Interestingly, as part of their analysis Ward et al showed that that a post-bronchodilator increase that met the ATS/ERS standard was moderately biased towards males. When an absolute increase in FEV1 (>200 ml) was considered by itself, it was highly biased towards males and this due to the fact that males have larger lung volumes than females. When the increase in FEV1 was expressed as a percent of predicted however, results were neutral to both gender and height.

The bias in the current 12%, 200 ml standard can be seen by comparing this graph with the prior graph. Because of the 200 ml threshold, a 12% increase for a male with an observed FEV1 of 40% of predicted or higher is significant, while a female needs to have an observed FEV1 of 52% of predicted or higher for a 12% increase to be considered significant.

Should a post-bronchodilator increase of 8% in percent predicted replace the current ATS/ERS standard of 12% and 200 ml? I think the answer is a qualified yes primarily because it appears to overcome gender bias and additionally because it also makes the 200 ml threshold a moot point. This yes has to remain qualified however, because it will also skew the distribution of individuals that are considered reversible based on how mild or severe their airway obstruction is and although are reasons to believe it be more correct than the current ATS/ERS standard the effect it may have on patient care and treatment has not been studied.

Long-term survival is one measure of clinical significance, and this study makes a strong case that a post-bronchodilator increase in FEV1 greater than 8% of predicted is a clear and significant threshold. This study only addressed changes in FEV1 however, and for this reason it does not mean that post-bronchodilator increases in FVC or IC (or possibly other values such as PIF) and other clinical improvements such as a decrease in dyspnea or cough frequency should not be considered significant.

References:

Brusasco V, Crapo R, Viegi G. ATS/ERS Task Force: Standardisation of lung function testing. Interpretive strategies for lung function tests. Eur Respir J 2005; 26: 948-968.

Hansen JE, Sun XG, Adame D, Wasserman K. Argument for changing criteria for bronchodilator responsiveness. Respiratory Med 2008; 102: 1777-1783.

Hansen JE, Porszasz J. Counterpoint: Is an increase in FEV1 and/or FVC ≥ 12% of control and ≥ 200 ml the best way to assess positive bronchodilator response? No. Chest 2014; 146(3): 538-541.

Pellegrino R, Brusasco V. Point: Is an increase in FEV1 and/or FVC ≥ 12% of control and ≥ 200 ml the best way to assess positive bronchodilator response? Yes. Chest 2014; 146(3): 536-537.

Ward H, Cooper BG, Miller MR. Improved criterion for assessing lung function reversibility. Chest 2015; 148(4): 877-886.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

I was reviewing spirometry reports and noticed several patients in a row with an early termination of exhalation that was not reflected in the reported expiratory time. This is far from the first time I’ve noticed this but seeing several examples more or less at the same time got me curious about how the software was determining expiratory time.

The ATS/ERS standards are somewhat mute on how the total expiratory time of a spirometry effort is determined. The closest the standard comes to addressing this is the end-of-test (EOT) criteria:

“The volume–time curve shows no change in volume (<0.025 L) for >=1 s[econd]”

But this is meant primarily as a quality indicator to show the patient has really achieved an adequate exhalation and not as a definition of expiratory time. In fact total expiratory time is not defined in the ATS/ERS standards at all and this leaves it up to individual equipment manufacturers to determine when a spirometry effort is over. This is a problem because the ATS/ERS standards also say that patients should:

“exhale for >=3 s in children aged <10 yrs and for >=6 s in subjects aged >10 yrs.”

So despite not being defined, the length of a spirometry effort is a critical quality indicator.

This spirometry effort was reported at being 6.41 seconds long, but it is clear the patient stopped exhaling well before that time. I downloaded the raw data for the effort into a spreadsheet and by looking at the volume and flow results I was able to see the patient had actually exhaled for 3.96 seconds. During the final 2.45 seconds there was no change in volume (the maximum change was 0.007 L, which is probably due to electronic noise). This makes it apparent that even if there is no change in expiratory volume our spirometry software continues to time the effort.

This spirometry effort was reported as being 8.29 seconds long, but the maximum exhaled volume occurred at 2.98 seconds and exhaled volume decreased slightly thereafter although the maximum decrease was only about 0.011 L. In one sense this is below the EOT threshold, but it is a negative volume and the EOT criteria is for a positive volume. What this does mean is that the software is not using a decrease in volume as a way to determine the end of exhalation, or at least it’s looking for a larger decrease than this.

This spirometry effort was reported as 5.63 seconds. That’s less than 6 seconds but the real end of the effort occurs at around 3 seconds (although the maximum exhaled volume occurs around 4.2 seconds). The maximum change in volume between 3.00 seconds and 5.63 seconds was 0.023 L, which meets the EOT criteria but also means that our spirometry software doesn’t use the EOT criteria to determine expiratory time. This isn’t exactly a surprise, but it’s nice to see it verified.

So how does this software determine where the test ends? By looking at the raw data for a number of spirometry efforts my best guess (and it really is a guess but it fits what I saw) is that after the spirometry effort is over, the software first looks for the maximum exhaled volume. The software then continues to look forward until it finds an inspiratory flow rate of around -0.1 L/sec and uses this as an indicator that inspiration has begun. The software then looks backward from this point until it finds a zero or positive expiratory flow and when it does it marks this as the end of test.

Note: Interestingly, for our lab software it is evident that the computer algorithm used to measure expiratory time is not the same that is used to measure the FVC volume. As mentioned previously, when a patient’s expiratory effort includes a pause and a small inspiration before completing the effort (not a good quality effort of course, but sometimes you have to work with what you can get), the FVC volume is taken from the largest expiratory volume that occurred before the inhalation, and not the largest expiratory volume that occurred during the entire effort. This causes a discrepancy between the expiratory time for the measured FVC volume and the reported expiratory time for the entire effort.

Expiratory time is an important quality indicator for spirometry and many if not most test systems are able to report this value. The problem is that there is no official definition or approach for measuring expiratory time and every manufacturer must decide for itself how to measure it. For the majority of spirometry efforts this is likely not a problem, but when a patient stops exhaling this doesn’t mean that the reported expiratory time reflects this.

It could be argued that the time a patient continues to try to exhale even though there is no significant expiratory flow should be counted as expiratory time. The problem with this thought is that there is no way to determine why expiratory flow has stopped and inspiratory flow hasn’t started. It could be a that there is no more volume to exhale despite a continued effort, it could be glottal closure, or it could be that the patient has just arbitrarily stopped exhaling.

Like many other testing issues, how expiratory time is determined is not explained in our lab system’s manuals. I was able to download raw test data that let me examine how well (or not so well) our software was determining expiratory time but this is not an option on many other manufacturer’s test systems.

In addition, I’ve run across a couple spirometry systems that did not display or print a volume-time curve, and for these systems it is not possible to verify the reported expiratory time in any way. I would agree that a volume-time curve is not as informative as a flow-volume loop, but it is still useful as a way to verify a number of test quality issues that don’t appear (or don’t appear all that well) in a flow-volume loop.

Hopefully, the next version of the ATS/ERS standards for spirometry will include a definition of expiratory time. My personal inclination would be to use the point where the maximum exhaled volume occurred as the expiratory time but I can see a couple circumstances where the results from this approach (particularly since electronic noise can cause flow and volume signal excursions where none really exist) would be debatable, so some definition of a zero flow signal similar to EOT would be needed as well. In any case, I would argue that a period of zero expiratory flow at the end of a spirometry effort should not be included as part of the reported expiratory time and that whenever possible the reported expiratory time should always be verified against a volume-time curve.

References:

Brusasco V, Crapo R, Viegi G. ATS/ERS task force: Standardisation of lung function testing. Standardisation of spirometry. Eur Respir J 2005; 26: 319-338.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

When assessing the response to bronchodilators the change in FEV1 is used far more frequently than any other spirometry result. Other values such as inspiratory capacity (IC) and peak inspiratory flow (PIF) have also been proposed as indicators, but the ATS/ERS standards includes changes in FVC as well as changes in FEV1 and this is often overlooked. Specifically they:

“…recommend using the per cent change from baseline and absolute changes in FEV1 and/or FVC in an individual subject to identify a positive bronchodilator response. Values >12% and 200 mL compared with baseline during a single testing session suggest a ‘‘significant’’ bronchodilatation.”

I don’t have any particular disagreement with this since post-BD increases in FVC are probably similar in nature to the post-BD changes in IC seen in some individuals with COPD. So when spirometry results like this:

	Pre-BD:	%Predicted:	Post-BD:	%Change:
FVC:	1.82	66%	2.55	+40%
FEV1:	0.66	32%	0.72	+10%
FEV1/FVC:	37	49%	29	-22%

comes across my desk, I’m inclined to consider that the results show a positive bronchodilator response. Post-BD increases in FVC are not usually quite as large as 40% however, so I took a closer look at this particular test. When I did what I saw was that the post-BD test length was significantly longer than the pre-BD test length.

Although FEV1 is often thought of as being a measurement of expiratory flow it is by definition a volume measurement. The time period over which it is measured (i.e., 1 second) is also tightly defined since by default it also includes a definition of how the beginning of exhalation is determined (back-extrapolation). FVC, on the other hand, it much more open-ended. It is simply defined as the maximum amount of air an individual can exhale after a maximal inhalation.

Note: A low FEV1/FVC ratio is one of the primary indicators of airway obstruction but the FEV1/FVC ratio can be falsely elevated when there is a submaximal FVC.  This is at least one of the reasons that the FEV6 and the FEV1/FEV6 ratio have been proposed a substitute for the FVC and FEV1/FVC ratio.

The indications for judging an adequate FVC include the end-expiratory flow rate (which is actually defined as no change in volume more than 0.025 L in 1 sec) and that the effort is at least 6 seconds long. In this case both the pre-BD and post-BD spirometry efforts met the 6 second criteria but neither met the end-expiratory flow criteria. Given the very severe airway obstruction seen in the results this is not surprising and judged by themselves either effort would probably have been considered adequate for routine spirometry.

So is the increase in FVC an indication of a positive reponse to a bronchodilator or is it simply from an increase in expiratory time? Conversely, is an increase in expiratory time itself an indication of a positive response to a bronchodilator?

In order to look at this more carefully I downloaded the raw FVC test data from our database (paired flow and volume signals, 200 samples/sec) and compared what the difference would have been if both efforts had the same expiratory time. What I saw was that at 8.25 seconds (the length of the pre-BD effort) the post-BD volume was 2.02 L, which is a 0.20 L increase from the pre-BD effort, but at the same time only an 11% increase. That’s almost significant by the ATS/ERS standards, but not quite. It’s not clear that this is a fair comparison however, since I’ve not seen any study that measured the post-BD expiratory volume that occurred at the same expiratory time as the pre-BD FVC.

A number of studies have shown that subjects with moderate to severe COPD often have a significant increase in FVC following bronchodilator but the role that expiratory time has in this is often undocumented. In those studies where expiratory time was reported it usually increased post-bronchodilator and in at least one study the investigators were able to show that the increase in expiratory time was most likely attributable to the effects of the bronchodilator and not solely to a learning effect. It appears, therefore, that although a post-BD increase in FVC can be explained in part by an increase in expiratory time, that an increase in expiratory time may also be due to the action of the bronchodilator.

All studies that showed an increase in FVC following bronchodilator thought it was likely due to a decrease in gas trapping. Interestingly more one study indicated that an isolated post-BD increase in FVC (i.e. with little or no change in FEV1) is almost solely seen in severe COPD. Although this does not appear to have been specifically addressed results from one study indicate that those individuals with a significant post-BD increase in FVC are also those who have a significant post-BD increases in IC and decreases in FVC.

Since SVC is frequently larger than FVC in patients with airway obstruction an interesting question is that whether individuals that have a significant post-BD increase in FVC show a similar increase in SVC.  Another question would be that since FEV6 is used as a substitute for FVC does FEV6 show as significant change as FVC or not?

An additional point concerning SVC is that the ATS/ERS standards suggest calculating the FEV1/VC ratio using the largest VC available regardless of its source (i.e. FVC, SVC or FIVC). I’ve seen a lab with software that automatically substituted the largest VC into the reported results for FVC, SVC and FEV1/FVC without documenting which specific test it was taken from. Since SVC is rarely measured post-bronchodilator this kind of automatic substitution could obscure significant post-BD changes in FVC.

The ATS/ERS standards state that a significant increase in FVC is a sign of bronchodilation but do not explicitly consider the effect that expiratory time has on FVC. Even though there can be a significant post-BD increase in expiratory time and this can be considered to be an effect of the bronchodilator the quality of the pre-BD effort should always be considered as well. When a pre-BD spirometry effort ends abruptly or early, particularly when compared to a better quality post-BD effort then I’d be far more inclined to discount it as a significant improvement.

In this particular case, there was an almost significant increase in FVC when the same expiratory time was considered and certainly a significant increase in expiratory time and FVC overall so I’d have to consider this was a significant response to bronchodilator.

References:

Brusasco V, Crapo R, Viegi G. ATS/ERS Task Force: Standardisation of lung function testing. Standardisation of spirometry. Eur Respir J 2005; 216: 319-338.

Brusasco V, Crapo R, Viegi G. Interpretive strategies for lung function tests. Eur Respir J 2005; 26: 948-968.

Cerveri I, Pellegrino R, Dore R, Corsico A, Fulgoni P, Van de Woestijne KP, Brusasco V. Mechanisms for isolated volume response to a bronchodilator in patients with COPD. J Appl Physiol 2000; 88(6): 1989-1995.

Girard WM, Light RW. Should the FVC be considered in evaluating response to bronchodilator? Chest 1983; 84(1): 87-89.

O’Donnell DE. Assessment of bronchodilator efficacy in symptomatic COPD. Chest 2000; 117: 42S-47S.

O’Donnell DE, Forkert L, Webb KA. Evaluation of bronchodilator responses in patients with “irreversible” emphysema. Eur Respir J 2001; 18: 914-920.

Schermer T, Heijdra Y, Zadel S, van den Bemt L, Boonman-de Winter L, Dekhuijzen R, Smeele I. Flow and volume responses after routine salbutamol reversibility testing in mild to severe COPD. Respiratory Medicine 2007; 101: 1355-1362.

Tashkin DP, Celli B, Decramer M, Liu D, Burkhart D, Cassino C, Kesten S. Bronchodilator responsiveness in patients with COPD. Eur Respir J 2008; 31: 742-750.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

A while back one of our technicians brought a helium dilution FRC graph to my attention and wanted to know if it showed a system leak. At that time my response was that it definitely wasn’t a leak (leaks don’t show increases in helium) and was probably due to too much oxygen being added to the system at the beginning of the test.

A couple of days ago a technician brought a similar graph to me and again I was asked why it looked unusual. I’ve had time to think about this issue since the last time and I’ve come up with an alternate explanation that I think fits the facts a bit better.

A normal helium dilution curve looks something like this:

which shows the helium decreasing with what is more or less an exponential decay curve. What’s unusual about the other curve is that it shows a relatively rapid fall to the lowest helium concentration near the beginning of the test and then a slow rise to the final concentration.

The helium dilution FRC test uses a closed-circuit consisting of a volume-displacement spirometer, a blower motor, a canister of CO2 absorbant (soda lime), tubing and a helium analyzer. The total volume of the circuit is roughly 10 liters. A small amount of helium is introduced into the circuit and after the blower motor mixes it thoroughly, the system starts off with about 7% helium (although the initial concentration doesn’t matter as much as the ability to measure it accurately).

An important point however, is that it takes time for the helium to equalize throughout the circuit. The blower motor pushes air through the circuit at about 30 LPM, which means that it takes 20 seconds or so for the full system volume to move around the circuit. In addition the gas flow through the helium analyzer is only about 250 ml/min and the helium analyzer is relatively slow to respond to changes in helium concentration. This means that after the helium has been added, it takes about a minute for the helium concentration to equalize in throughout the circuit and for this to register on the helium analyzer signal.

When somebody starts breathing a gas mixture containing a tracer gas like helium it also takes a certain amount of time for the tracer gas concentration to equalize within the lung. The rate at which this occurs is partly due to tidal volume and respiratory rate (alveolar ventilation) and partly to the degree of ventilation inhomogeneity. Everybody has a certain amount of ventilation inhomogeneity but the amount varies from one person to another and generally increases with age and airway obstruction.

When somebody with a low ventilation inhomogeneity breathes with a high alveolar ventilation on the circuit, there is a rapid increase in helium concentration within their lung.

While this is happening however, it takes time for their exhaled air to be incorporated and mixed within the gas circuit.

This means that for at least the first 20 seconds of the test the person breathes a gas mixture that contains the initial concentration of helium and by the time the helium concentration in the gas circuit decreases because of their exhaled air, the concentration of helium in their lungs is now higher than it is in the circuit.

From this point on, instead of the person’s lung coming into the equilibration with the helium concentration in the gas circuit, the gas circuit has to come into equilibration with the helium concentration in the person’s lung.

The issue with the somewhat unusual helium concentration graph has to do with the differences in time constants in the breathing circuit and an individual’s lung. When an individual’s alveolar ventilation is low or when there is a high degree of ventilation inhomogeneity, the time constant for equilibration is longer in the lung than it is in the breathing circuit and the helium concentration decreases with a normal decay curve. When, on the other hand, ventilation inhomogeneity is low and alveolar ventilation is high, the time constant for the equilibration of helium in the lung can be lower than it is for the breathing circuit and when that happens, the reverse curve will be present.

Time is a factor for all physical processes. When a complex system has components with different time constants unusual patterns can emerge. In this case even though the helium concentration pattern may be considered to be unusual it’s actually just a consequence of time constants being different than what we usually expect.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

My hospital’s Oncology division treats a number of patients with lymphoma and leukemia. It also has an active bone-marrow transplant program and for all of these patients diffusing capacity measurements are a critical part of assessing treatment progress. Since these patients are also frequently anemic, correcting DLCO results for hemoglobin is also critical.

For a factor that has as much importance for the interpretation of DLCO results as it does the effect of hemoglobin on DLCO has actually been studied a relatively small number of times. Part of the reason for this is the problem of finding an acceptable model. A reduced or elevated hemoglobin is a consequence of many diseases and conditions. When studying patients longitudinally it is often difficult to separate the changes in DLCO that occur from the disease process and those that occur from changes in hemoglobin. For this reason changes in hemoglobin pre- and post-treatment in anemia and polycythemia have been studied most frequently.

The ATS/ERS currently recommends correcting DLCO for hemoglobin (although notably they recommend that the predicted DLCO be corrected, not the observed value) using the equations developed by Cotes et al in 1972. Cotes’ work was based on subjects with iron-defficienty anemia but just as importantly on theoretical considerations involving Roughton and Forster’s equation on the relationship between the membrane and hemoglobin components of the diffusing capacity:

Several other approaches towards correcting DLCO for hemoglobin have been developed. Dinakara et al, studied a group of patients with chronic anemia and found a curvilinear relationship between hemoglobin and DLCO that although generally similar to Cotes, had a significantly greater amplitude.

Mohsenifar et al studied a diverse group of patients with anemia or polycythmia as well as normals and instead found a linear relationship between hemoglobin and DLCO.

Marrades et al studied a relatively small group of patients with anemia and also found a linear relationship between hemoglobin and DLCO that was quite similar to Mohsenifar.

The amount by which each approach corrects DLCO depends variously on the observed DLCO, predicted DLCO, observed hemoglobin and the presumed value for normal hemoglobin. For this reason direct comparison of the correction equations is difficult and must be taken on a case by case basis.

The relatively significant difference in results between studies of what should be a fundamental property of the diffusing capacity makes it difficult to assess the different correction factors. Comparison is also complicated by the fact that DLCO was measured using different breath-holding times (Ogilvie vs Jones-Meade) and FIO2’s (0.21 vs 0.18). Admittedly there is often a fair amount of overlap between correction factors, however which specific factors overlap changes as the observed DLCO and hemoglobin change.

The selection of the Cotes equation by the ATS/ERS is to some extent understandable since it tends to be the most conservative of the correction factors. It’s reliance on the Roughton and Forster equation also makes it the most “scientific” of the correction factors but neither of these reasons necessarily make it the most correct. In particular, the Cotes equation has been criticized because a basic assumption is that the ratio between DMCO (the membrane component of diffusion) and Vc (the capillary blood volume) is a constant of 0.70 over a wide range of DLCO and hemoglobin values.

I found it interesting that the Mohsenifar and Marrades studies generated almost identical correction factors from two different subject groups but I am concerned that the linear relationship that they indicate exists between hemoglobin and DLCO may be too simplistic. Having said that, a comprehensive analysis of CO uptake by red blood cells indicates that DLCO does change linearly with hemoglobin but this observation was based primarily on mathematics and not on empirical data.

Finally, although the Dinakara equation was most often an outlier from the other equations, at least one study has indicated that it more accurately predicted the risk from bone marrow transplantation.

All studies can be criticized to one degree or another for their relatively small subject populations and the lack of a good study model which is unfortunate given how important hemoglobin is to DLCO. For relatively small differences in hemoglobin there is no significant different in the correction factors from the different studies but then same could probably be said of the uncorrected DLCO. At both moderate and extreme differences in DLCO and hemoglobin however, there are significant differences between the correction factors and this has implications for the interpretation of DLCO results.

This is an area that definitely needs more research but it’s far from clear how this should be done. There are reasonably significant physiological differences between acute and chronic anemia (and polycythemia) that make it difficult to develop a reliable human or animal model. In addition the changes in DLCO and hemoglobin that occur due to medications or disease and their treatment are difficult to disentangle. Although there is intriguing evidence that the correction factors developed by Marrades and Mohsenifar may be more accurate than the others, at the present time DLCO should probably continue to be corrected for hemoglobin using the Cotes equation but I say this mostly for the sake of standardization and not because there is any overwhelming evidence that it is the most accurate approach.

Source:	Gender:	Units:	Formula:
Cotes	Male	Ratio	((1.7 x Hb) / (10.22 + Hb))
	Female	Ratio	((1.7 x Hb) / (9.38 + Hb))
Dinakara	Both	Ratio	1 / (0.06965 x Hb)
Marrades	Male	DLCO (ml/min/mmHg)	1.4 x (14.6 – Hb)
	Female	DLCO (ml/min/mmHg)	1.4 x (13.4 – Hb)
Mohsenifar	Both	DLCO %predicted	1.35 x (44 – Hct)

References:

Brusasco V, Crapo R, Viegi G. ATS/ERS Task Force: Standardisation of lung function testing. Standardisation of the single-breath determination of carbon monoxide uptake in the lung. Eur Respir J 2005; 26: 720-735.

Burgess JH, Bishop JM. Pulmonary diffusing capacity and its subdivisions in polycythemia vera. J Clin Invest 1963; 42(7): 997-1006.

Chakraborty S, Balakotaiah V, Bidani A. Diffusing capacity reexamined: relative roles of diffusion and chemical reaction in red cell uptake of O2, CO, CO2 and NO. J Appl Physiol 2004; 97: 2284-2302.

Coffey DG, Pollyea DA, Myint H, Smith C, Gutman JA. Adjusting DLCO for Hb and its effects on the hematopoietic cell transplantation-specific comorbidity index. Bone Marrow Transplantation 2013; 48: 1253-1256.

Cotes JE, Dabbs JM, Elwood PC et al. Iron-deficiency anaemia: its effect on transfer factor for the lung (diffusing capacity) and ventilation and cardiac frequency during sub-maximal exercise.  Clin Sci 1972; 42: 325–335.

Cotes JE, Chinn DJ, Miller MR. Lung Function, Sixth Edition. Blackwell Publishing, 2006.

Dinakara P, Johnston RF, Kauffman LA, Solnick PB. Am Rev Resp Dis 1970; 102: 965-969.

Herbert SJ, Weil H, Stuckey WJ, Urner C, Gonzalez E, Ziskind MM. Pulmonary diffusing capacity in polycythemic states before and after phlebotomy. Chest 1965; 48(4): 408-415.

Marrades RM, Diaz O, Roca J, Campistol JM, Torregrosa JV, Barbera JA, Cobos A, Felez MA, Rodriguez-Roisini R. Adjustment of DLCO for hemoglobin concentration. Am J Respir Crit Care Med 1997; 155: 236-241.

Mohsenifar Z, Brown HV, Schnitzer B, Prause JA, Koerner SK. The effect of abnormal levels of hematocrit on the single breath diffusing capacity. Lung 1982; 160: 325-330.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

I was reviewing reports today when I ran across one with some glaring errors. There were several things that immediately told me that the reported plethysmographic lung volumes were way off; the VA from the DLCO was almost a liter and a half larger than the TLC and the SVC was only about half the volume of the FVC.

When I took a look at the raw test data I saw at least part of the reason why the technician had selected these results to be reported and that was because the SVC quality from most of the efforts was poor. They mostly looked like this:

It is apparent that the patient leaked while panting against the closed shutter and this caused the FRC baseline to shift upwards. I’ve discussed this problem previously, but when this happens the RV is larger than the FRC, there is a negative ERV and the TLC is overestimated. There is no way to fix this problem from within the software. The FRC is determined by the tidal breathing before the shutter closes and cannot be re-measured afterward.

For this reason the technician had selected the only effort that had no leak and a reasonably clean SVC maneuver.

The problem with this is that the SVC volume was also the smallest of all of the efforts and substantially less than the FVC so although the SVC quality was better in one sense, the overall quality of the lung volume measurement was still quite poor. Strictly speaking at this point the technician should either had the patient perform a couple more SVC maneuvers and combined them with the existing FRC measurements, or should have just deleted the results and not reported them at all. Neither of these things were done however, so I needed to try to fix them.

The good news was that the FRC measurements from all of the patient’s plethysmography efforts were almost exactly the same (less than 5% difference) so I knew I had a good quality FRC to work with. Even excluding the baseline shifts however, the largest of the SVC measurements was still substantially less than the FVC so none of them were salvageable. I could use the FVC in place of the SVC but in order to derive TLC and RV I needed an IC or an ERV. These volumes aren’t measured during an FVC, but the tidal breathing that occurs before the actual FVC maneuver is stored and displayed. Fortunately, our software lets me download the numerical data that generates the flow-volume loop (paired volume and flow data points at 200 samples per second) into a spreadsheet. From the spreadsheet I was able to isolate the volume of the end-exhalation tidal breath (FRC) and use this to determine both the IC and ERV.

Even if I wasn’t able to download the raw test data in this way, I could still have gotten a reasonably accurate IC and ERV from the FVL graph itself. Every graphic program I use shows the X and Y coordinates of the cursor. By placing the cursor on the left and right ends of the volume scale and measuring the difference in the x-coordinates it’s possible to determine how many pixels there are per liter.

By placing the cursor on the end-exhalation (FRC) side of the tidal loop it’s also possible to measure how many pixels there are from there to the zero marker. After that, dividing by the number of pixels per liter gives the IC. This isn’t as accurate as being able to use the actual numerical data but all of the FVL graphs I work with have at least 50 pixels per liter and that means that the error can’t be more than 0.02 L (not perfect of course, but still better than reporting grossly incorrect results).

The IC I retrieved from the FVC test data was 3.66 L. I averaged the FRCs from the plethysmographic efforts with the best quality loops and got 2.21 L. Since:

TLC = IC + FRC

That meant that the TLC had to be 5.87 L. And since:

RV = TLC – VC

and

ERV = FRC – RV

That meant the RV was 1.79 L and the ERV was 0.42 L. I created a new plethysmographic lung volume record by manually entering these results and now the lung volumes on the patient’s report looked like this:

It is still possible that the patient’s TLC is underestimated, but most importantly the report now showed that the TLC was within normal limits which is at least reasonably correct rather than showing a mild to moderate restrictive ventilatory defect, which was most definitely not correct at all.

This is not the first time I’ve done this. I don’t do it all that often because these kind of errors are usually small enough not to make a significant difference in the reported results but at the same time I have no particular hesitation about manually correcting a report when I know the way in which the software derived the results wasn’t correct. I am concerned however, that I have never been able to get our technicians to make these kinds of fixes on their own. Having said that, this is probably not a fair criticism because even after I’ve given detailed explanations on what I’ve done to fix reports and why I’ve done it this is still probably seen as a supervisory-level problem and beyond their level of expertise. To some extent I understand this since there really isn’t a straightforward way to fix these kind of errors and it requires a good understanding of the software to know what approach needs to be taken in each case.

We spend a fair amount of time training technicians on testing quality and I also annoy them reasonably often with emails when I find an error (it’s not a problem, it’s an educational opportunity!) but even so I think there is an attitude that if the computer says that’s what it is, then that’s what it is. This is an endemic problem in our field (and to be fair, others as well) and it’s not isolated to just technicians. I’ve had (pulmonary!) physicians show me reports with significant mathematical and testing errors that came with their patients from other labs and then ask why my lab isn’t giving them the same results.

What I don’t see as much as I’d like is an appreciation for how the different pieces of a PFT report fit together. I am glad that our technicians can usually recognize suboptimal test quality when they see it but I would be happier if it was also realized how a low SVC affects the TLC and the RV.

I have similar criticisms for our lab software. It seems that at least a couple times a week we have patients who are able to perform a good FVC maneuver but for whatever reason never seem to be able to perform an SVC that is anywhere near as good. Although the FVC is assessed for a number of ATS/ERS quality indicators there isn’t a similar process of any kind for the SVC. Admittedly, the differences between SVC and FVC are usually small and don’t make a significant difference in the TLC and RV, but it’s not possible to assess this within any of the lung volume test modules. The only way we can compare the FVC and SVC is on a report and if there is a serious discrepancy there’s no easy way to correct it.

This report was a real fixer-upper. Like all fixer-uppers it took a bit of effort to find and fix the errors, but the end result made it worthwhile. I know that IC and ERV are not supposed to be measured as part of an FVC maneuver and I also understand the reasoning for this, but despite the “forced” part, an FVC and an SVC are much the same. When an SVC is too flawed to use then there is no reason not to “borrow” information from an FVC.

The software for every test system has its own quirks but one way or another it should be possible find and correct errors before a report is finalized (even if only in the testing notes). We should all remember that computers are only tools, patients are only human and testing systems can get whonky despite our best efforts to the contrary. A report is not etched in stone and there are numerous ways to cross-reference test results and verify their integrity. If you don’t look for problems you’ll never find them, and if you don’t find (and fix!) problems then you aren’t doing your patients, your physicians or your lab any favors.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

I’d spent some time researching single-breath tests a while back and of course ran across the Fowler method for measuring anatomic dead space. It’s a relatively simple test but assessing its results as well as the results of alternate dead space measurement techniques turns out to be more complicated than I had remembered.

The official definition of anatomic dead space is that it is that part of the inhaled volume that remains in the airways at the end of inhalation and does not participate in gas exchange. An accurate estimate of this volume is important because respiratory dead space (Vd/Vt, discussed previously) is composed of both anatomical and physiological dead space. The physiological component of the respiratory dead space cannot be determined without knowing the anatomical dead space.

Anatomic dead space is usually considered to be the physical volume of the airways but static measurements of airway volume do not take into consideration the dynamic aspects of respiration. The most commonly used method for measuring anatomic dead space in a research setting is the single-breath technique developed by Fowler in 1948. In this method, after an inhalation of oxygen, the nitrogen concentration in an individual’s exhalation is plotted against exhaled volume.

During exhalation, phase I is where nitrogen-free gas is exhaled from the airways. Phase II is the S-shaped transition between pure deadpace gas and alveolar gas. Phase III is alveolar gas. Fowler back-extrapolated the phase III slope and using a planimeter (a mechanical device that measures area) he found the mid-point of Phase II that was defined by equal areas (A1, A2). Numerous gases have been studied for this purpose (helium, oxygen, nitrogen, argon, nitrous oxide, carbon dioxide) and they all produce similar curves and with the possible exception of helium they all produce reasonably similar values for dead space. Theoretically, the Fowler dead space measurement could be performed by any system capable of measuring single-breath nitrogen washouts but it is actually performed most frequently by volumetric capnography systems used to monitor ICU patients and patients undergoing anesthesia. For this and other reasons anatomic dead space is now most frequently measured using CO2.

The most frequent criticism of the Fowler technique has been the way the phase III regression line is back-extrapolated and that is because it is “selected by drawing a straight line along the alveolar plateau and extending it to the left.” The alveolar plateau is not necessarily even or flat so the accuracy of the regression line is up to the observer’s eye.

For this reason other approaches to measuring dead space have been developed. Because volumetric capnographs are usually automated a lot of attention has been towards approaches suitable for computer algorithms. A particularly simple example of this is the threshold technique. A specific exhaled CO2 threshold (usually just above the noise or zero level) is selected and the exhaled volume up to the point the threshold is exceeded is considered to be dead space. This approach likely underestimates dead space because the measurement primarily consists of the volume in phase I. It is however, very easy to implement and has been occasionally used on ventilators and anesthesia equipment.

Another approach is to look at the rate of change in CO2 during exhalation. When this is plotted, the peak value is equivalent to the highest slope of the phase II part of exhalation which usually occurs in the middle of the phase II slope, which is also presumed to be equivalent to the dead space volume.

This technique is sensitive to both analyzer noise and the data sampling interval and for this reason the results are often poorly reproducible.

The Pre-Interface Expirate (PIE) algorithm uses multiple steps. The first is to divide the maximum exhaled CO2 (usually the PETCO2) by two and use this to select a location within phase II of the exhalation. The volume at this location is then used to isolate a larger section of the phase II (and possibly part of phase III).

The ΔPeCO2/ΔVolume is then calculated at intervals (usually the sampling frequency or some fraction thereof) for this section of the exhalation and finally, the average of the interval’s ΔPeCO2/ΔVolumes is obtained. The exhaled volume corresponding to the average value is then considered to be the dead space.

The dead space calculated by the PIE technique is sensitive to the maximum exhaled PECO2 (because that determines how much, if any, of phase III is included in the calculations) and alternate versions of this approach have been proposed that “adjust” the maximum exhaled PECO2 in one way or another.

Fowler’s method plotted the exhaled CO2 concentration against the exhaled volume but a number of the alternate methods plots the exhaled CO2 volume against the exhaled volume and uses linear line-fitting or linear regression analysis with either second-order or third-order polynomials to back-extrapolate.

To some extent the approaches using exhaled CO2 volume rather than CO2 concentration show less variance and higher repeatability than the Fowler technique. Researchers have shown that the Fowler technique is sensitive to CO2 analyzer noise and this noise is reduced during the integration process for exhaled CO2 volume. The alternate techniques appear to have their own sensitivities to such things as the phase III slope however, which is usually significantly increased in patients with COPD and asthma. Although some of the alternate techniques produce relatively reproducible results there is no clear consensus as to which of these approaches is most accurate or whether they are accurate at all.

When interpreting the results of dead space measurements it must be remembered that airway volume and therefore anatomical dead space is not a fixed quantity. Airway volume depends on posture and is largest when an individual is in the upright position and lowest in the supine or prone positions, with the recumbent position being in-between. Airway volume also depends on lung volume and is largest at TLC and smallest at RV. It is usually assumed that when it is reported that unless otherwise specified airway dead space is measured from an inhalation that started at FRC.

Anatomical dead space measurements are also sensitive to breath holding time and for this reason Cotes recommends a breath holding time of 0.4 to 0.6 seconds. Prolonged breath holding time, whether deliberate or accidental, tend to decrease the measured volume. This is presumed to be due to cardiogenic mixing in the airways and not to diffusion. In addition at least one study showed that measured dead space increases as respiratory rate increases although the reason this occurs is not clear. Finally, results from another study indicated that measured dead space may also increase as tidal volume increases.

Vd/Vt is frequently measured in a variety of situations using the modified Bohr equation with arterial PCO2 and mixed-expired CO2. The Bohr (respiratory) dead space is a combination of anatomical and physiological dead space. Anatomical dead space however, is most often estimated rather than measured and when it is estimated the simple formula of 1 ml per pound of body weight is most often used. This relationship originated from a study performed in 1955 and clinicians have used both actual and estimated body weight (although they have often failed to specify which) for this purpose. A recent study has indicated that neither actual or predicted body weight is a reliable indicator of dead space. This is not surprising since other studies have shown that dead space volume in fact correlates best with FRC or TLC.

There are a small handful of reference equations for airway dead space and the majority of them are based on values other than weight

From [1]:

Vd (ml) = 4.9 + (0.0113 x TLC) + (0.062 x tidal volume (ml))

Where:

Male TLC = ((7.61 x ht (M)) – (0.019 x wt (kg)) – 4.73) x 1000

Female TLC = ((7.38 x ht (M)) – (0.016 x age) – 6.35) x 1000

From [2]:

Male:

Vd (ml) = 3.846 x ht(cm)^2.499 x 10^-4

Vd (ml) = (1.765 x wt (kg)) + 32.16

Vd (ml) = Antilog((0.393 x BSA(M²)) + 1.468)

Vd (ml) = (36.703 x FRC (L, measured)) + 49.381

Female:

Vd (ml) = 23.823 x ht(cm)^2.131 x 10^-4

Vd (ml) = (1.913 x wt (kg)) + 21.267

Vd (ml) = Antilog ((1.913 x wt(kg)) + 1.407)

Vd (ml) = (38.175 x FRC (L, measured)) + 36.593

From [3], male and female:

Vd (ml) = (194 x ht (M)) – 73

Volumetric capnograph monitors are frequently used to monitor ventilated patients during anesthesia or in the intensive care unit and many of these devices are capable of automatically calculating airway dead space. This may be a useful measurement but users of these devices should be aware of which method is being used to measure dead space volume and the factors that can affect their results.

Anatomical dead space is often perceived as the volume of the conducting airways. To some extent this is true but during an inhalation some alveoli receive fresh gas well before the presumed anatomical dead space has cleared the airways and some alveoli receive fresh gas only well after it has cleared. All techniques that measure anatomical dead space use an averaging process of one kind or another and depending on the technique the measured dead space can be biased by the phase I, phase II or phase III parts of exhalation. There is no particular consensus on dead space measurement techniques and for this reason, the original Fowler technique, despite its known limitations, continues to be the most commonly performed technique, even when only used as a comparison.

Researchers and clinicians often use an individual’s body weight as a way to estimate anatomical dead space in order to determine physiological dead space but this is an oversimplification and likely to be inaccurate. In addition dead space is not a fixed value and depends on posture and lung volume but these facts are often overlooked. Despite all these issues anatomical dead space remains an important concept and is an important factor in the ventilation of the lung.

References:

[1] Astrom E, Niklason L, Drefeldt B, Bajc M, Jonson B. Partitioning of dead space – a method and reference values in the awake human. Eur Respir J 2000; 16: 659-664.

Bartels J, Severinghaus JW, Forster RE, Briscoe WA, Bates DV. The respiratory dead space measured by single breath analysis of oxygen, carbon dioxide, nitrogen or helium. J Clin Invest 1954; 33(1): 41-48.

Brewer LM, Orr JA, Pace NL. Anatomic dead space cannot be predicted by body weight. Resp Care 2008; 53(7): 885-891.

Cotes JE, Chinn DJ, Miller MR. Lung Function, 6^th Edition. Blackwell Publishing, 2006.

Fowler WS. Lung function studies. II. The respiratory dead space. Amer J Physiol 1948; 154: 405-416.

Fowler WS. Lung function studies. IV. Postural changes in respiratory dead space and functional residual capacity. J Clin Invest 1950; 29(11): 1437-1438.

[2] Hart MC, Orzalesi MM, Cook CD. Relation between anatomic respiratory dead space and body size and lung volume. J Appl Physiol 1963; 18(3): 519-522.

Hatch T, Cook KM, Palm PE. Respiratory dead space. J Appl Physiol 1953; 5(7): 341-347.

Heller H, Konen-Bergman M, Schuster K-D. An algebraic solution to dead space determination according to Fowler’s graphical method. Comp Biomed Research 1999; 32: 161-167.

[3] Kars AH, Bogaard JM, Stijnen T, de Vries J, Verbraak AFM, Hilvering C. Dead space and slope indices from the expiratory carbon dioxide tension-volume curve. Eur Respir J 1997; 10: 1829-1836.

Radford EP. Ventilation standards for use in artificial respiration. J Appl Physiol 1955; 7(4): 451-460.

Tang Y, Turner MJ, Baker AB. Systematic errors and susceptibility form noist of four methods for calculating anatomical dead space from the CO2 expirogram. Brit J Anaesth 2007; 98: 828-834.

Wolff G, Brunner JX, Weiberl W, Bowes CL, Muchenberger R, Bertschmann. Anatomical and series dead space volume: concept and measurement in clinical praxis. Applied Cardiopulmonary Pathophysiology 1989; 2: 299-307.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

I’ve been planning on putting together a tutorial on characterizing and interpreting the contours of flow-volume loops so I’ve been accumulating flow-volume loops that are examples of different conditions. Lately I was reviewing some of them and noticed that when I tried to compare loops from different individuals with similar baseline conditions that the different sizes of the flow-volume made this difficult. For example, these two loops are both from individuals with normal spirometry.

One is from short, elderly female and one is from a tall, young male. If all you had to look at was the flow-volume loops, you might think that the smaller loop was abnormal, but the larger loop actually comes from a spirometry effort with an FVC that was 92% of predicted while the smaller loop’s FVC was 113% of predicted. The difference in sizes of these loops is of course due to the difference in age, gender and height between these individuals but also because of settings we’ve made in our lab software and because of the ATS/ERS spirometry standards.

Our lab software will allow us to display and print flow volume loops in only one of four sizes:

• 2 L x 4 L/S
• 4 L x 8 L/S
• 6 L x 12 L/S
• 8 L x 16 L/S.

There is an option that lets the software automatically select the scale but we decided to mandate the 8×16 format for all flow-volume loops. One reason we did this is that most of the pulmonary physicians were used to working with just a single scale and found it difficult to adjust when the scale kept changing. Another reason however, was that we found that the software made the decision about which scale to use based on the FVC, not on the peak flow. This caused some flow-volume loops to get chopped off:

Admittedly our decision to go with an 8×16 format was made at least 15 years ago but rightly or wrongly the one thing we’ve never gotten a complaint or question about was the scale used to report flow-volume loops.

So why is the flow-volume loop displayed the way it is? As so many things are it is actually the result of somewhat arbitrary decisions. When I first acquired a spirometry system in the mid-1970’s that was capable of performing flow-volume loops there were no particular standards for the way they were supposed to be reported and the X-Y recorder that came with the system allowed us to select which axis was which, which direction was which and the amplitude in both the X and Y directions. The most common convention at that time was to have volume on the X-axis with exhalation on the top and inspiration on the bottom but in some research papers the loop was on its side with flow as the X-axis. Having TLC (maximum inhalation) on the left side and RV (maximum exhalation) on the right side of the loop was also the most common convention but again a fair number of research papers showed it the other way around. We ended up reporting flow-volume loops oriented the same they are now, but we chose a 1:1 scale for flow and volume (hey, it made sense to us at the time) so for a number of years our flow-volume loops were very, very tall.

By the mid-1980’s however, standards had been set not only for the exhalation-inhalation and TLC-RV directions, but for a ratio of 2:1 for flow and volume and this is what has led to the limited number of scales that our lab software has for printing flow-volume loops.

Note: I’ve tried to track down the origins of the 2:1 ratio between flow and volume and couldn’t find anything specific. The 1983 ACCP spirometry recommendations are mute on any technical specifications whatsoever. The 1987, 1995 and 2005 ATS and ATS/ERS standards all mandate the 2:1 ratio, but this is not attributed to any source. Playing around with different scales I’ve found that 2:1 tends to optimize the ability to recognize contour. A 1:1 ratio smears contour in the vertical direction and 4:1 smears it in the horizontal direction and in both of these directions contour quickly become unrecognizable.

So what’s wrong with reporting flow-volume loops this way? Well, standardization is great and I am a big fan of it but in this case I was trying to find a way to normalize flow-volume loop contours across a wide range of heights, ages and genders. After thinking about this for a while it occurred to me that the best way to do this was to get rid of liters and liters/second and instead use percent of predicted for volume and flow.

My lab’s software lets me download the flow and volume data for specific spirometry efforts and by uploading this data into a spreadsheet and then fiddling with it a bit I found I was able to re-scale flow-volume loops using percent of predicted. When I did this, I found the loops to actually be a lot more informative than when they are reported as liters and liters/second. Here are the “normal” loops from above re-scaled as percent of predicted:

And here’s the other spirometry effort:

And here’s another spirometry effort with moderately severe airway obstruction (FEV1 56% of predicted):

And finally one more with very severe airway obstruction (FEV1 28% of predicted):

Re-scaling flow-volume loops using percent of predicted makes it immediately evident whether the FVC and PEF are normal or not while at the same time bringing the contour of the loop to the forefront. Yes, all loops are reported in the “same” size, but I found it to be an easy conceptual shift to think about flow and volumes as a percent of predicted rather than as liters and liters/second and didn’t find this to be a problem at all.

And yes, the same percent predicted information could be added to conventional flow-volume loops, but I didn’t find this nearly as informative.

Does re-scaling flow-volume loops preserve the 2:1 PEF/FVC ratio mandated by the ATS/ERS standards? Close, but not exactly.

Note: I found the curves in these graphs to be very interesting. They are in part an artifact of the units involved (liters and liters/second) but I still wonder what they may be saying about spirometry and age and height.

But a good question would be whether the 2:1 ratio is actually optimal. 2:1 was probably chosen because its ability to emphasize the flow-volume loop contour was reasonably correct but as importantly, it was easy to implement. At the moment at least, I would argue that re-scaling flow-volume loops using the percent of predicted does at least a good a job at emphasizing the flow-volume loop contour as the 2:1 ratio.  I would also say there’s nothing magical about the 2:1 ratio nor do all  loops need to be reported with the same ratio as long as the contour is preserved.

The downside of reporting flow-volume loops this way is that they are then tied to a specific set of reference equations. Clinically however, I don’t have a problem with this at all. Flow-volume loops would be reported using the same reference equations that are used to report the spirometry results numerically, so what’s the difference?

I will admit this idea needs more study. I am interested to see what flow-volume loop contour characteristics are preserved (or aren’t) across different heights and ages when the same percent of predicted numerical values are present. At some point I’ll probably find out since I am going to continue collecting different flow-volume loops and when I create the tutorial on flow-volume loop contours I will most probably be using percent of predicted scaling.

So my proposal is that flow-volume loops should be reported using percent of predicted rather than liters and liters/second. The value of this is that it increases the information density of the reported flow-volume loop and it does this by emphasizing the contour of the flow-volume and makes its “normalacy” (or lack thereof) more apparent without decreasing its readability in any way.

I’ve had some discussions lately with people from different areas of the PFT field about reports and one of the issues has been the need to improve their readability. Flow-volume loops are informative to us but this is mostly due to familiarity and not due to their inherent readability. The standards for reporting flow-volume loops have not changed for well over 30 years and they originated at a time when the fidelity of a represented signal was of primary importance. Data fidelity is still important but at the same time it’s become significantly more important for the information embedded in the data to be meaningfully transmitted to a much larger audience. I wasn’t thinking about these things when I re-scaled flow-volume loops using percent of predicted; I was just trying to normalize them across different ages, heights and genders for teaching purposes. It may well be that re-scaling them in this way is still not the best way to improve readability but I hope that it at least stirs up some interest in this topic.

So, are there any policy makers listening? Any equipment manufacturers? Just curious.

References:

ACCP Scientific Section. Statement on spirometry. Chest 1983; 83(3): 547-550

Brusasco V, Crapo R, Viegi G et al. ATS/ERS Task Force: Standardisation of lung function testing. Standardisation of spirometry. Eur Respir J 2005; 26: 319-338

Standardization of spirometry – 1987 Update. Am Rev Respir Dis 1987; 136: 1285-1298.

Standardization of spirometry – 1994 Update. Am J Respir Crit Care Med 1995; 152: 1107-1136.

PFT Blog by Richard Johnston is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License