Author: Richard Johnston

I’ve been reading Miller et al’s Laboratory evaluation of Pulmonary Function which was published in 1987. That was an interesting time since PFT equipment manufacturers had mostly transitioned to computerized systems but there were still a lot of manual systems in the field. For this reason the book’s instructions are still oriented mostly around manual pulmonary function testing and there are numerous warnings about double-checking the results from automated systems.

The book includes extensive discussion on the calculations and formulas used for testing which makes it useful as a teaching resource. The authors were also very concerned about the correct way to run a PFT lab so there is a fair amount of discussion about staff requirements for education and training (including the medical director) and staff behavior and conduct. To this end each chapter includes extensive instructions on the proper way to perform tests and treat patients. Although the tone of this is somewhat dated and I’d like to say these kind of reminders shouldn’t be necessary, it doesn’t hurt to set a standard on the level of professionalism we should aspire to.

What caught my eye though, was a section in the chapter on Normal Values titled Interdependence of Normal Values which discussed of the value of deriving predicted TLC from predicted FVC. The authors were concerned that reference equations for different tests (and not just lung volumes) were being selected without concern for how well they fit together. I’ve previously written about the problems that results when inserting the reference equation for FVC into the reference equations for lung volumes. In one instance, the TLC was adjusted so that the final predicted TLC was equal to RV + VC, but this meant that TLC (and IC) were changed from the original reference equations. In another, the FVC was just substituted for SVC without adjustment which meant that RV + VC was not equal to TLC and IC + ERV was not equal to VC and this makes interpreting results problematic. What this means however, is that almost 30 years after this was published, this problem is still around.

As a solution, the authors point out that ratios, such as the FEV1/FVC ratio and the RV/TLC ratio tend to be relatively independent of height.

Since:

TLC = FVC + RV

This can be mathematically re-written as:

Which means that TLC can be derived from predicted FVC if the RV/TLC ratio is known.

So how accurate is the TLC derived this way? The first problem is that even in 1987 there were a variety of RV/TLC ratio reference equations.

Interestingly, males and females have slightly different patterns when their predicted RV/TLC ratios are compared. Males tend to have lower RV/TLC ratios overall; show the largest differences between predicted RV/TLC ratios at age 20, and these differences decrease with increasing age. Females tend to have higher RV/TLC ratios overall; show the smallest differences between predicted RV/TLC ratios at age 20, and these differences increase with increasing age.

At the time the book was written age was the only variable in all of the RV/TLC reference equations but since 1987 at least two lung volume studies have been published that do not have reference equation specifically for the RV/TLC ratio. Instead it was calculated from the reference equations for RV and TLC and these included height in one way or another as a variable. Although the differences are less dramatic for females than for males, both studies show that the RV/TLC ratio actually decreases with increasing height.

This indicates that the RV/TLC ratio may not be completely independent of height, but the differences are relatively small, and in fact are within the range of the RV/TLC ratios calculated from the reference equations that only had age as a variable.

How well does the TLC derived from predicted FVC compare to TLC calculated from the standard reference equations?

For males the majority of TLC reference equations height is the only variable and single equation that does include age (Crapo) shows an increasing (?!) TLC with age. When two of the more common FVC reference equations (NHANESIII and Morris) are used to derive TLC this resulted in patterns that were quite different from the standard TLC reference equations. For the NHANESIII FVC, the TLC had a curvilinear pattern which increased to a maximum at about age 40, and decreased thereafter. For the Morris FVC, TLC decreased fairly constantly with age. This alone must show that the FVC derived TLC is incorrect, doesn’t it?

Not necessarily. One study of lung volumes with a study group aged 65 to 85 showed a declining TLC with increasing age and the results from that reference equation is actually similar to the NHANESIII TLC over the same age range.

For females, about half of the standard TLC reference equations include age as a factor and about half don’t. For those that do include age TLC decreases with increasing age and the Morris TLC showed a similar decline. The NHANESIII TLC however, showed a marked curvilinear pattern that peaked at about age 50. The TLC from the 65-85 study group had approximately the same slope as the NHANESIII over the same age range but was lower.

When I started looking at this I expected that the FVC-derived TLC values would probably be similar to the standard TLC reference equations but it’s actually a bit of a mess, isn’t it?.

Here’s the problem though: the formula used by Miller et al to derive TLC is absolutely correct, particularly since it’s merely a mathematical restatement of TLC = FVC + RV. So why does it appear to fail? Partly because of the assumption that one RV/TLC ratio applies to all populations but also because of differences between study populations and different statistical analyses. Most particularly it fails because the decline in FVC and increase in RV/TLC ratio that occur with age have different slopes.

What this does is to throw a spotlight on the problem of mixing-and-matching different lung volume and spirometry reference equations. All of the lung volume subdivisions (and ratios) have to add up correctly and there is no way to insert a predicted FVC without the need to alter multiple values. No matter how this is done the choice of which values to alter and how to alter them is a completely arbitrary decision.

Note: I understand the temptation to replace the SVC reference equation with an FVC reference equation. It is a common perception that because the study populations for FVC studies are usually be larger and the quality of the statistical analysis tends to be better that this makes the FVC reference equations “better” than the SVC reference equations. Another reason is so that the reported predicted VC for spirometry and lung volumes match because its “obvious” that they should be the same. The problem is that there is no easy and reliable way to insert a predicted FVC into predicted lung volumes without distorting one or the other.

When the reference equations for spirometry and lung volumes come from the same study population (notably Gutierriez et al), the FVC-derived TLC matches the reference equation TLC for all ages, genders and heights. Unfortunately using a single study population is not a guarantee, since choices made when performing statistical analysis on different elements (Marsh et al) can cause the FVC-derived TLC to differ from the same study’s reference equation TLC.

I think that Miller et al’s original point about ratios being relatively independent of height actually has a fair amount of merit. When compared to the FVC and TLC volumes that occur across the range human heights the differences in the FEV1/FVC ratio and RV/TLC ratio are miniscule. This probably means that we should be placing more emphasis on ratios in general (IC/ERV? FEV3/FVC?) when interpreting test results.

An interesting question that the FVC-derived TLC raises is whether the fact that TLC changes little with age (mostly for males, partly for females) is actually correct. All FVC-derived TLCs decrease with increasing age (at least above age 50 they do). As noted, this may be a function of the different age slopes for FVC and RV/TLC, but if so why does this difference exist?

Finally, there are surprisingly few lung volume reference equations for non-Caucasian populations (and shame on all of us for that) but those that exist have RV/TLC ratios that are essentially the same as Caucasians. This at least suggests that the RV/TLC is well preserved across all ethnicities. Many ethnicities that do not have lung volume reference equations do have spirometry reference equations. In these populations the standard practice is to estimate predicted TLC by subtracting a specific percentage from a Caucasian reference equation TLC. Wherever this percentage is questionable it could be verified to some degree by deriving TLC from the predicted FVC and an average RV/TLC ratio.

Technical note: The average of the height-independent RV/TLC ratio reference equations for males is:

RV/TLC ratio = (Age x 0.29125) + 14.7

For females, it is:

RV/TLC ratio = (Age x 0.3424) + 16.4

References:

[A] Cordero PJ, Morales P, Benlloch E, Miravet L, Cebrian J. Static Lung Volume: Reference values from a Latin population of Spanish descent. Respiration 1999; 66: 242-250.

[B] Crapo RO, Morris AH, Clayton PD, Nixon CR. Lung volume in healthy nonsmoking adults. Bull Eur Physiopathol Respir 1983; 18: 419-425

[C] Garcia-Rio F, Dorgham A, Pino JM, Villasante C, Garcia-Quero C, Alvarez-Sala R. Lung volume reference values for women and Men 65-85 years of age. Am J Respir Crit Care Med 2009; 180: 1083-1091.

[D] Gutierrez C, et al. Reference values of pulmonary function tests for Canadian caucasians. Can Respir J 2004; 6: 414-424.

[E] Hankinson JL, Odencrantz JR, Fedan, KB. Spirometric reference values from a sample of the general U.S. Population. Amer J Resp Crit Care 1999; 159: 179-187

[F] Marsh S, Aldington S, Williams M, Weatherall M, Shirtcliffe P, McNaughton A, Pritchard A, Beaseley R. Complete reference ranges for pulmonary function tests from a single New Zealand population. New Zealand Med J 2006; 119: N1244.

[G] Miller WF, Scacci R, Gast LR. Laboratory evaluation of Pulmonary Function. J.B. Lippincott Co, 1987.

[H] Morris JF, Koski A, Johnson LC. Spirometric standards for healthy nonsmoking adults. Am Rev Resp Dis 1971; 103: 57-67.

[I] Neder JA, Andreoni S, Castelo-Filho A, Nery LE. Reference values for lung function tests. I. Static Volume. Braz J Med Biol Res 1999; 32: 703-717

[J] Roberts CM, MacRae KD, Winning AJ, Adams L, Seed WA. Reference values and prediction equations for normal lung function in a non-smoking white urban population. Thorax 1991; 46: 643-650.

[K] Stocks J, Quanjer PH. Reference values for residual volume, function residual capacity and total lung capacity. Eur Respir J 1995; 8: 492-506.

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

A friend is taking her father to a PFT lab (2500 miles away from where I am the moment so I couldn’t go along with them) because he has been short of breath for a couple of years, but oddly enough, only when lying on his side. I expect that despite these rather specific symptoms he will only get routine spirometry. I don’t necessarily fault the PFT Lab he’s going to for this, partly because physician orders often don’t include specifics, partly because they may not have the facilities to perform supine or lateral spirometry, and partly because its not clear lateral spirometry would show anything.

I don’t think that my lab is necessarily any better. We have only one room with an exam table that allows us to perform positional spirometry and that is largely because of the ALS patients we regularly see. Even so, unless we received specific orders to perform supine or lateral spirometry it’s unlikely that one of our technicians would think it was necessary and then take it on themselves to perform it. That itself is part of the problem not only for my lab but for the field of Pulmonary Function testing in general (but that’s another story).

The real problem however, is that the way in which spirometry is performed around the world is focused almost exclusively on detecting expiratory airway obstruction. It may be true that airway obstruction is primarily expiratory, but this ignores that fraction of individuals who have some degree of inspiratory obstruction. It also overlooks those individuals whose FVC is underestimated and FEV1/FVC ratio overestimated due to some degree of gas trapping. It also overlooks individuals that have positional airway obstruction that is not evident in the upright position.

We’ve fallen into the trap of thinking that there’s only one way to perform spirometry, and this is a mistake.

Spirometry is usually taken as being synonymous with the Forced Vital Capacity maneuver. That may be the way the word spirometry has come to be used but it really isn’t correct. Spirometry refers to any test that can be performed with a spirometer and (among other tests) includes the Slow Vital Capacity (SVC) and the Forced Inspiratory Vital Capacity (FIVC). This also means that there is much more that can be done with a spirometer than what it’s ordinarily used for.

A patient visit to a pulmonary clinic may appear simple but actually depends on the convergence of numerous high-value resources. Patient, physician, technician and clerical time. Waiting room, exam room, lab and office space. Spirometers, computers, an institution’s clinical database and the software that binds them together. I would like to suggest that the way in which spirometry is usually performed provides information that, although not necessarily incorrect, may instead be misleading for some fraction of patients. This fraction may or may not be small and there is no way to estimate this correctly since it will vary largely based on the patient population. Regardless, in order to maximize resource utilization, I think that the amount of information that can be obtained from spirometry needs to maximized as well.

For this reason I would like to suggest that a patient’s initial spirometry session should include Forced Expiratory Vital Capacity (FEVC) maneuvers, Forced Inspiratory Vital Capacity (FIVC) maneuvers, Slow Vital Capacity (SVC) maneuvers and when appropriate, positional FEVC maneuvers.

Why?

One reason is to obtain a patient’s largest Vital Capacity (VC) for the FEV1/VC ratio. The FEVC, FIVC and SVC maneuvers are performed differently and a patient may reach a higher VC with a FIVC or SVC than they would with an FEVC. My experience is that perhaps as many as 20% of patients that perform a SVC reach a higher VC than they did during the FEVC maneuver. For at least a quarter of these patients, this increased VC reduces the FEV1/VC ratio enough that it is then consistent with airway obstruction when it wasn’t previously.

Another reason is to rule out inspiratory airway obstruction. The number of patients with inspiratory obstruction is probably small, but since the baseline assumption is that obstruction is primarily expiratory, this has not been studied to any degree. So the number may (or may not) be small but some fraction of patients have inspiratory obstruction and these individuals need to be identified. To some extent this would be satisfied by ending the FEVC maneuver with an FIVC maneuver (and this is part of the ATS/ERS spirometry guidelines) but those researchers who have studied inspiratory flow rates recommend performing the FIVC as a separate maneuver (which also maximizes the possibility of obtaining a larger FIVC than FEVC).

Another reason is to obtain the Inspiratory Capacity (IC) and the Expiratory Reserve Volume (ERV). A reduced IC, when seen along with moderate or severe expiratory airway obstruction, is an indication of gas trapping. In addition more than one researcher has shown that the ERV/IC ratio is the best indication of the affects of obesity on lung function.

Post-bronchodilator FEVC, FIVC and SVC maneuvers would also need to be performed to fully assess an individual’s response to bronchodilator. At the present time the ATS/ERS statement on interpretation only considers increases in FEV1 and FVC to be significant. Research has shown however, that significant increases in inspiratory flow rates or increases in IC can occur without a corresponding increase in FEV1 or FVC. The number of patients with these kind of increases is again probably small, but determining which patients exhibit them will certainly have a bearing on their clinical management.

Finally, whenever there are the appropriate patient symptoms, positional (supine or lateral) FEVC (and maybe FIVC?) maneuvers should also be performed. Changes in VC in the supine position when compared to upright are an indication of diaphragmatic weakness. Changes in flow-volume loop contours that occur in supine or lateral positions can indicate the presence of bodies pressing against the larger airways as well as other airway disorders. Again, the number of patients with these specific problems is probably small, but when a patient complains of a notable increase in dyspnea in a specific position, it should be investigated.

I will not disagree that the majority of patients are well served by routine FEVC testing. I will also not disagree that the majority of patients who have FIVC and SVC testing will not have any significant findings. The point is that there is a certain number of patients whose problems are not detectable without FIVC, SVC and positional testing. The fraction of the normal patient population with problems that can only be seen by these additional spirometry tests is not known because this has never been studied. At a guess it has to be at least 5% since approximately that fraction of patients that perform an SVC in my lab have an FEV1/SVC ratio that is significantly lower than the FEV1/FVC ratio.

To some extent I am advocating that all initial patient spirometry should include FIVC and SVC testing but that is mostly because the indications for whom this would be useful are unclear. This flow diagram is my best guess for a structured approach to this kind of testing:

I will acknowledge that there are a number of roadblocks towards regularly performing a full range of spirometry tests and the first of these is the software our test systems come with. At the present time I don’t know of any software that allows the FIVC to be performed and reported separately from an FEVC. At one time my lab’s software allowed us to mix-and-match the FEVC and FIVC components from different tests but this feature was lost several software upgrades ago.

Another problem is that comparing results to a baseline value is assumed to consist of only pre- and post-testing. So what happens when you perform baseline tests, post-bronchodilator tests and then positional tests? Not much since only one type of post-test can be reported unless you start a new patient visit and because in some labs billing, appointments and reporting are tied together this requires work-arounds both in reporting and billing.

What I’d like to see in a report would look something like this:

There is also the problem of compensation. There are no CPT codes for FIVC, SVC and positional spirometry which means its not possible to bill for them. This is a fault of our more than somewhat byzantine payment system (which still has no CPT code for MIPs and MEPs despite their having been included in the ATS statement on respiratory muscle testing in 2002). For this reason my lab has performed supine spirometry on ALS patients for over 15 years while only being able to charge for routine spirometry. I feel very strongly however, that a PFT Lab is a diagnostic service whose purpose is the accurate assessment of lung function and that means doing the tests that facilitate this. To do any less is to fail our patients and physicians.

One final thought is that performing FIVC, SVC and positional testing is also an opportunity to improve patient care without additional equipment. We all need new and updated equipment but let’s also work smarter and make the best use out of what we already have.

Spirometry is more than just the FVC. The FIVC, SVC and positional testing are also part of spirometry but are significantly underutilized and the information they can provide is often overlooked. It may not be necessary to routinely perform all of these tests (although I am obviously advocating that to some extent they should) but at the very least we should be willing to recognize that there are times when they should be performed.

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

Turbine spirometers have been around in one form or another for well over a hundred years. The accuracy of the early versions of this type of spirometer was poor, partly because of the turbine designs weren’t terribly efficient and partly because these devices were mechanical in nature and the gear trains or other mechanical linkages added a lot of friction and resistance.

The first electronic turbine spirometer was the Marion Labs Spirostat which came to market in the early 1970’s. It used a disposable in-line turbine where the turbine blades were directly in the stream of inspiratory and expiratory flow and rotated accordingly. There was an optical pickup (a light beam passing through a hole in the turbine) and the rotations were converted to inspiratory and expiratory volumes. The passageways through the Spirostat’s turbine sensor were quite narrow and the resistance to flow was high. The turbine also had a fair amount of mass, was perceptibly slow to start moving and slow to stop and would not have met the current ATS/ERS standards.

During the early 1980’s a turbine spirometer was developed and marketed that used swirl or deflector end-plates (the equivalent of fixed fan blades) to cause airflow to rotate helically through the body of the sensor.

This rotating airflow in turn causes a flat vane mounted vertically in the air stream to rotate. A pair of optical or Hall effect sensors count rotations and enable the direction of flow to be detected. This is the basis for most of the current turbine spirometers.

Rotating vane turbine sensors are primarily volumetric. This means is that across a relatively wide range of flow rates, each rotation of the vane is directly linked to the volume of gas flowing through the sensor. Even so, it is the differential pressure of the helical gas flow that causes a vane to rotate. The factors that affect the accuracy and sensitivity of any given turbine sensor are inertia, friction, the area of the vane and how effectively the deflector/swirl end-plates convert linear gas momentum into angular momentum.

The concerns I’ve heard most frequently voiced about turbine spirometers is their accuracy at low flow rates and how well they respond to rapid changes in flow rates.

For a variety of reasons turbine spirometers would be expected to be least sensitive to low flow rates. The ATS/ERS specifications on spirometry state that the end-of-test indication is a flow rate below 0.025 L/sec. The ability to meet this specification is not usually detailed for any spirometer regardless of which measurement technology it uses, so whether turbine spirometers are any worse (or any better) at measuring a flow rate this low is difficult to determine. I will note that the specifications found in the advertising brochures for several turbine spirometers indicate a flow accuracy that is well above this value (for example +/- 200 ml/sec) and that some of the cheapest turbine spirometers do not claim to measure FVC, but FEV6 instead (or in some cases only FEV1 and PEF). This is verified to some extent by one study which noted that a turbine spirometer significantly underestimated volume at low flow rates (< 0.5 L/sec). On the other hand another comparison between precision flow meters and a turbine sensor appears to show larger discrepancies at higher flow rates than at lower so this isn’t quite as clear as it should be.

There have been numerous comparison studies between turbine and more conventional spirometers. Although the majority of these studies showed that turbine spirometers met the ATS/ERS guidelines they also showed that FVC tended to be underestimated somewhat and that the FEV1/FVC ratio tended to be slightly overestimated. This appears to a problem specific to turbine spirometers since similar findings were shown across a half dozen studies for at least three different brands of turbine spirometers. It has been suggested that an insensitivity to low flow rates would cause spirometers to terminate spirometry efforts early and thereby underestimate FVC but since none of the comparative studies tracked expiratory times this is just speculation. There were also significant differences in methodologies between these studies and overall there is no overwhelming reason to believe that turbine spirometers are significantly less accurate than any other measurement technique.

An early study of a turbine flowmeter by Yeh et al indicated the presence of significant inertia which caused “lag-before-start” and “spin-after-stop” effects.

This finding has formed the basis of numerous criticisms of turbine spirometers but its validity is open to question, most particularly since the study did not describe the turbine flow meter’s mode of operation. At the time of the study there were many different turbine designs which included both the in-line as well as the deflection end-plate and vane types. The vane used in most present-day turbine spirometers has a very low mass and a relatively large surface area which means they are likely to respond quickly to any changes in flow rate. Quick does not mean instantaneous however, and although it may be small all turbines have some lag but it is unlikely to be anywhere near the degree noted in Yeh et al’s study.

One interesting point is that most turbine spirometers produce one “pulse” for every half-rotation and that each rotation is related to a specific volume of gas. At high flow rates, the pulse count rate is also high. One manufacturer indicated that their device had 144 pulses per liter (which works out to 6.9 ml/pulse). The current ATS/ERS recommendation is that spirometers should have a sampling rate of 100 hz. Even at modest flow rates (> 0.69 L/sec), the number of pulses/second is above 100 hz, but at lower flow rates the pulse count rate drops below 100 hz. At terminal flow rates (0.025 L/sec) the pulse count rate is only 3.6 hz. To some extent this means that turbine flow meters do not meet ATS/ERS standards but in reality it is a reflection of how they operate.

One of the primary selling points of turbine spirometers is that they do not require frequent calibrations. This appears to relatively true and was verified in at least one two-year longitudinal study. For this reason turbine spirometers are probably well suited to portable or field applications where calibrations are infrequent or non-existent and where accurate trends may be more important than absolute accuracy (personal use or clinical trials, for example). The same characteristics may well apply to office spirometry but since this seems to be an area that’s on the dividing line between using spirometry for monitoring and using it for diagnostics I am uncomfortable saying that regular calibration is not necessary. It should also be remembered that turbine sensors are mechanical in nature and can be irreversibly damaged by a drop or fall and this also makes me uncomfortable saying that regular calibration of some kind for any turbine spirometer is not necessary.

Other reasons to select turbine spirometers would be that they are relatively insensitive to gas composition, humidity and altitude. One study did find a sensitivity to temperature but that may be limited to the specific model being tested.

No flow sensor is perfect. Although it may be small, turbine lag, and the circumstances under which it occurs (low flow rates, high flow rates, changing flow rates) as well as the overall linearity of a turbine sensor requires software correction. How much and what kind of a correction is performed for a given spirometer is proprietary information however, and details are not shared with end-users. Strictly speaking, this is not particularly different for turbine spirometer manufacturers than it is for the manufacturers of pneumotach, mass flow, pitot, variable-orifice and ultrasonic spirometers and given the amount of resources each manufacturer has devoted to developing a particular flow sensor it’s also mostly understandable.

But even though turbine sensors appear to be relatively simple details like the aerodynamics of the helical air flow are probably quite complex. Despite an extensive search however, I’ve been unable to find any study that explored the physics of their operation (any doctoral candidates out there looking for a thesis topic?). The lack of information about turbine spirometer characteristics leaves me in something of a quandary and I would be more comfortable discussing them if they were better studied and understood.

Turbine spirometers come in a variety of configurations. Some use disposable sensors and some don’t. Some come with jeweled bearings and some don’t. Most come in a horizontal orientation but at least one doesn’t. Although there is some equivocal evidence they are insensitive to low flows all claim to meet the ATS/ERS standards and there is no overwhelming evidence that they don’t. The bottom line is that a spirometer should be chosen based not on whether it does (or doesn’t) use a turbine, but on whether the software used to perform, store, manage and report tests meets your needs.

References:

Bongers T, O’Driscoll BR. Effects of equipment and technique on peak flow measurements. BMC Pulm Med 2006; 6: 14.

Caras WE, Winter MG, Dillard T, Reasor T. Performance comparison of the hand-held MicroPlus portable spirometer and the SensorMedics Vmax22 diagnostic spirometer. Respir Care 1999; 44(12): 1465-1473.

Chowienczyk PJ, Lawson CP. Pocket-sized device for measuring forced expiratory volume in one second and forced vital capacity. Brit Med J 1982; 285: 15-17.

Degryse J, Buffels J, Van Dijck Y, Decramer M, Nemery B. Accuracy of office spirometry performed by trained primary-care physicians using the MIR spirobank hand-held spirometer. Respiration 2012; 83: 543-552.

Dirksen A, Madsen F, Pedersen OF, Vedel AM, Kok-Jensen A. Long term performance of a hand held spirometer. Thorax 1996; 51: 973-976.

Godschalk I, Brackel HJL, Peters JCK, Bogaards JM. Assessment of accuracy of applicability of a portable electronic diary card spirometer for asthma treatment. Respir Med 1996; 90: 619-622.

Gunawardena KA, Houston K, Smith AP. Evaluation of the turbine pocket spirometer. Thorax 1987; 42: 689-693.

Jones KP, Mullee MA. Lung function measurement in general practice: a comparison of the Escort spirometer with the Micromed turbine spirometer and the mini-Wright peak flow meter. Respir Med 1995; 89: 657-663.

Liistro G, Vanwelde C, Vincken W, Vandevoorde J, Verleden G, Buffels J. Technical and functional assessment of 10 office spirometers. Chest 2006; 130: 657-665.

Paul KP, Schultz T. Evaluation of a pocket-sized turbine spirometer for clinical use with children. Respir Med 1997; 91: 369-372.

Pederson OF, Miller MR, Sigsgaard T, Tidley M, Harding RM. Portable peak flow meters: physical characteristics, influence of temperature, altitude and humidity. Eur Respir J 1994; 7: 991-997.

Pollard AJ, Mason NP, Barry PW, Pollard RC, Collier DJ, Fraser RS, Miller MR, Milledge JS. Effect of altitude on spirometric parameters and performance of peak flow meters. Thorax 1996; 51: 175-178.

Yeh MP, Adams TD, Gardner RM, Yanowitz FG. Turbine flowmeter vs. Fleisch pneumotachometer: a comparative study for exercise testing. J Appl Physiol 1987; 63(3): 1289-1295.

Wild LB, Dias AS, Fischer GB, Rech DR. Pulmonary function tests in asthmatic children and adolescents: Comparison between a microspirometer and a conventional spirometer. J Bras Pneumol 2005; 31(2): 97-102.

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

I’ve been thinking a bit about the shape of flow-volume loops lately. In part this has been about ways to accurately describe them in reports; in part speculation about the information that may be embedded in them that isn’t in any of the routinely reported spirometry values; and in part about how the human eye perceives and categorizes them in a way that is difficult to simplify and put into a computer algorithm. A couple days ago I found a recent article where a geometrical analysis was applied to flow-volume loops in individuals with COPD and this got me curious about what other graphical techniques have been used to analyze flow-volume loops.

Given how long flow-volume loops have been around (over 50 years) the graphical analysis of flow-volume loops has been attempted remarkably few times. Excluding a handful of strictly numerical approaches (based primarily on MEF and MIF ratios) I was only able to find three graphical analysis techniques. I think this small number says volumes about the difficulty of analyzing flow-volume loop shapes meaningfully. Despite different degrees of sophistication the reality is that none of these techniques has ever seen any kind of common usage. Even so these attempts are both interesting and instructive.

The most recent technique is a fairly straightforward geometric approach from Lee et al and its use appears to be limited primarily to individuals with airway obstruction.

The flow-volume loop is analyzed primarily to determine what the authors call the Area of Obstruction (Ao). To do this, a diagonal line is drawn from peak flow to the end of exhalation. The area that exists between the actual flow-volume loop contour and this diagonal line is defined as the area under the diagonal (Au). The area of Au is then compared to the area of a triangle (At) defined by the peak flow, the exhaled volume at the time of the peak flow, and the end of exhalation. The area of obstruction, which is actually a ratio, is then calculated as:

The more the flow-volume loop curve bows inwards from the diagonal (often described as coving, scooping, or being concave) the closer Ao comes towards 1. When the flow-volume curve is flatter, Ao comes closer to 0. In a sense, Ao describes how concave a flow-volume loop is without actually defining the shape of the contour.

Lee et al also determined what they called the area of the rectangle (Ar, which is basically defined as the percent predicted Peak Flow times the percent predicted FVC), and then other additional parameters, which included Ao/Ar, Ao/PEF and Ao/FVC. From their study population they compared these parameters as well as more routine PFT values such as FEV1, FEV1/FVC ratio and PEF with the RV/TLC (which they considered a marker of hyperinflation) and 6-minute walk distance (6MWD). Statistical analysis showed that all Ao values (Ao, Ao/Ar, Ao/PEF, Ao/FVC) correlated highly with RV/TLC and that Ao/Ar and Ao/PEF correlated better with 6MWD than did FEV1 or the FEV1/FVC ratio.

This technique is highly limited both by the observed PEF and by the need for the expiratory curve to be concave as well as the assumption that a diagonal drawn between the PEF and end-of-exhalation describes the ideal expiratory flow rate. Although Ao does to some extent describe the concavity of a loop, this does not necessarily translate well into the degree of obstruction. As an example the following loop has an Ao somewhere around 0.75:

But the results for this loop are quit normal:

	Observed:	%Predicted:	Predicted:
FVC:	4.98	106%	4.69
FEV1:	3.74	105%	3.57
FEV1/FVC:	75	99%	76
PEF:	12.44	136%	9.17

This loop has an Ao around 0.65:

but is from an individual with very severe airway obstruction:

	Observed:	%Predicted:	Predicted:
FVC:	1.81	57%	3.21
FEV1:	0.61	25%	2.43
FEV1/FVC:	33	44%	76
PEF:	1.09	19%	5.88

And this approach does not work well with any loops that are convex rather than concave:

For these reasons I suspect that if Lee et al’s study population was broader and contained more subjects with both normal and restrictive lung function rather than just those with COPD, the correlation between Ao, RV/TLC and 6MWD would likely be a lot less statistically significant.

One final issue with Lee et al’s approach is that I never understood the clinical relevance of Ar (the area of the rectangle). Since this area the product of height (peak flow) and width (FVC volume), the area of two very different flow volume loops (think restriction and obstruction) could be very similar. The purpose of Ar and how it was supposed to improve the relevance of Ao (if it did) were not explained.

Kapp et al took a somewhat simpler approach and measured the angle between lines drawn from the MEF@50% to an imaginary point described by the TLC and PEF, and to the end-of-exhalation. This angle (B) can be derived by straightforward geometrical calculations using PEF, MEF@50% and FVC. One advantage of Kapp et al’s approach over Lee et al’s is that it can describe both concave and convex contours (although interestingly Kapp et al used the terms concave and convex oppositely to the current use of these terms which in turn means that not only have the words that have been used to describe flow-volume loops evolved, they have never really been standarized).

An angle less than 180 degrees indicates a concave expiratory contour and one that is greater than 180 degrees a convex contour. Kapp et all showed that the angle was lower in males than females, and that it decreased with age and with an increase in smoking pack-years. They also categorized the angles associated with different lung diseases, however the overlap between different classifications was significant and so the usefulness of angle B as a discriminating factor is limited.

The weakness of Kapp et al’s approach is that it solely based on an angle that occurs at MEF@50%. In particular, as airway obstruction becomes more severe there is a point at which an inflection point (an abrupt change in expiratory flow rate) appears in the flow-volume loop.

Kapp et al’s Angle B is not able differentiate between the presence or absence of this inflection point. I can also see that Angle B would likely be inaccurate when peak flow is low or delayed, or when expiration terminates early.

Mead’s approach is probably the most sophisticated since it measures what he termed the slope ratio at intervals throughout exhalation. The slope ratios consist of a comparison of the slope of a line immediately tangent to the flow-volume loop with a chord line drawn from the end-of-exhalation.

The measurement is then performed at 10% intervals from 80% and 10% of the FVC.

A slope ratio of 1 would indicate a that expiratory flow is straight line between peak flow and the end-of-exhalation. Slope ratios above 1 indicate a concave contour and the degree of concavity increases as the slope ratio grows higher. Slope ratios below 1 indicate a convex contour and the degree of convexity increases as the slope ratio gets closer to 0.

A flow-volume loop likes this:

would generate a slope ratio graph something like this:

One like this:

would generate a graph something like this:

Mead was able to show that there were relatively distinct patterns for asthma, chronic bronchitis and emphysema. He was also able to show the distinct changes that occur in asthmatics during exercise-induced bronchoconstriction and post-bronchodilator.

Although Mead’s approach provides more information than the other techniques, I am hard pressed to say that it actually provides more information than can be obtained from the original flow-volume loop. Similarly, although Lee et al and Kapp et al developed methods for quantifying the degree of concavity (and for Kapp et al the degree of convexity as well) of a flow-volume loop, a significant failing (at least from my point of view) in all three techniques is their inability to detect the presence (or absence) of the inflection point that is often seen in severe airways obstruction.

Although I found these different analytical techniques to be interesting, and they have certainly given me something to think about when categorizing flow-volume loop shapes, it is not surprising that graphical analysis of flow-volume loops has shown little or no practical value in the routine interpretation of spirometry results. All of these techniques are entirely dependent on high quality and repeatable spirometry efforts. Any attempt to analyze spirometry efforts with short exhalation times, suboptimal patient effort or poor reproducibility will produce inconsistent and possibly misleading results. In addition, some of the distinctly diagnostic flow-volume loop contours (in particular those with expiratory or inspiratory plateaus) are not recognizable by any of these techniques.

Since Mead’s slope ratio is generated by comparison to a chord line originating from the end-of-exhalation its results are not particularly dependent on peak flow. Both Lee et al and Kapp et al’s approaches however, are highly dependent on peak flow. To some extent this is a problem because ATS/ERS guidelines on the selection of spirometry efforts does not consider PEF at all but are instead only concerned with the highest combined FVC and FEV1.

The most significant problem with any graphical analysis of a flow-volume loop is that all techniques to one degree or another are focused on the concavity (or convexity) the expiratory curve and these factors by themselves do not necessarily differentiate between the presence, absence or degree of airway obstruction. Context is also important. A flow-volume loop that is normal for a 70-year old would likely be considered abnormal for a 20-year old.

I would be interested in a computer algorithm that was able to categorize the shapes of flow-volume loops because this could potentially aid the interpretation of spirometry results in clinics and physician offices. At the moment however, not only is there no standardized terminology, even human categorization is inconsistent. This is partly because there are no really agreed-upon conventions for describing flow-volume loops but it’s also because experience and the ability to judge the quality of a spirometry effort are critical to knowing whether the loop contour is actually diagnostic or just an artifact of a suboptimal test effort, and it is these aspects of the assessment process that are the most difficult to quantify (and to teach).

References:

Kapp MC, Schachter EN, Beck GJ, Maunder LR, Witek TJ. The shape of maximum expiratory flow volume curve. Chest 1988; 94: 799-806.

Lee J, Lee C-T, Lee JH, Cho YJ, Park JS, Oh Y-M, Lee S-D, Yoo HI. Graphic analysis of flow-volume curves: a pilot study. BMC Pulmonary Medicine 2016 16: 18.

Mead J. Analysis of the configuration of maximum expiratory flow-volume curves. J Appl Physiol 1978; 44(2): 156-165.

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

A report came across my desk today and at first glance it looked fairly straightforward. There was a mildly reduced TLC and FVC, and although the SVC was slightly lower than the FVC it looked like this patient had mild restriction.

	Observed:	%Predicted:	Predicted:
FVC:	1.73	68%	2.56
FEV1:	1.23	65%	1.89
FEV1/FVC:	71	97%	73
TLC:	3.58	73%	4.89
FRC:	2.07	75%	2.78
RV:	1.94	83%	2.33
RV/TLC:	54	114%	48
SVC:	1.69	66%	2.56
IC:	1.51	72%	2.11
ERV:	0.13	30%	0.45

In addition, the flow-volume loop looked fairly typical for restriction, with a normal peak flow and a reduced volume.

When I looked at the DLCO results however, I suddenly got a different picture. Specifically, the VA from the DLCO was larger than the TLC and the inspired volume (Vinsp) was significantly larger than both the FVC and the SVC.

	Observed:	%Predicted:	Predicted:
DLCO:	13.51	83%	16.23
VA:	3.87	82%	4.73
Vinsp:	2.26

The ATS/ERS recommends using the largest VC, whatever the source, when calculating the FEV1/VC ratio. When I substituted the Vinsp for the FVC in the FEV1/VC ratio I got was 54.4 (75% of predicted) and that immediately told me that the patient, despite the apparently normal FEV1/FVC ratio, actually had moderate airway obstruction.

As importantly, the VA and Vinsp made the reported TLC more than a bit questionable. Before thinking about making any corrections however, it’s important to determine whether the VA is believable. VA is calculated as a function of the inspired volume and the exhaled tracer gas concentration. The alveolar sample the exhaled gas concentration is measured from is optimized for DLCO measurements however, not lung volume measurements.

In this case, the tracer gas (CH4, green trace) is reasonably flat which indicates good gas mixing and suggests the VA is probably a good indication of the TLC.

Given all this, is it possible to correct the reported TLC? In the past I’ve had reports where the FVC was substantially larger than the SVC and by analyzing the flow-volume loop I was able to derive IC and ERV.

In these cases I was able to correct the TLC by assuming the FRC was correct and then re-calculating the TLC and RV using the new IC and ERV. This time however, all I had was the volume curve from the DLCO test.

A simple approach would be to take the existing RV and then add the Vinsp:

1.94 L + 2.26 L = 4.20 L (86% of predicted).

but that assumes the existing RV was correct. The problem with this is that the existing SVC maneuver is already known to be suboptimal and that means that the ERV and therefore the RV are also questionable. If the patient’s ERV was instead assumed to be normal (0.45 L) then subtracting the predicted ERV from the FRC would give an RV of:

2.7 L – 0.45 L = 1.62 L (70% of predicted)

and TLC would be:

1.62 L + 2.26 L = 3.88 L (79% of predicted)

But a reduced RV in the presence of moderate airway obstruction just doesn’t seem quite right (although certainly not impossible if it is a mixed defect) and the TLC derived this way has a percent predicted that is a bit lower than the VA (predicted TLC and predicted VA differ by the amount of deadspace, which in this case is assumed to 0.16 L) so this probably isn’t correct.

So is the ERV actually correct? Reviewing the SVC effort showed that the ERV portion looked fairly good.

The SVC effort lasted for about 10 seconds and half that time was below FRC. This means that there is a good chance that the RV is reasonably correct and that a TLC derived from the current RV and the Vinsp would also be reasonably correct.

As far as I’m concerned, the evidence shows the patient most likely has a TLC that is above our LLN (80% of predicted). The problem with all this and the reason that I am not actually able to correct the reported TLC is that without a “real” ERV or IC any method I use is just a guess and I can’t report numbers I can’t justify. In the end, the best I could do was to point out the moderate airway obstruction and say that the TLC was likely underestimated.

Despite the fact that the FVC, SVC and Vinsp are all vital capacities, our lab software does not alert a technician to any discrepancies between these measurements during testing. It’s not until the final report is generated that any differences between these values becomes evident. I would like to see the lab’s technicians take a harder look at the reported results and to be quicker to perform additional SVC maneuvers but I also realize that patients often balk at additional testing and that there can also be time constraints.

The SVC is an often overlooked part of lung volume measurement and it’s not clear to me that the link between the lung volume subdivisions measured during an SVC and the reported TLC and RV are as well understood as they should be. At one time or another we all learned that:

SVC = IC + ERV

FRC + IC = TLC

FRC – ERV = RV

Not exactly rocket science, but how often is this simple set of relationships kept in mind when assessing test results?

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

I was reading an article recently that made an off-hand reference to the 100% oxygen shunt fraction test. Results from the test were included in the data analysis but the equations the researchers used were not presented nor were they referenced, nor was the procedure described. This is probably because the shunt fraction test and its equations are very much old-school pulmonary physiology but even if the subject is probably covered at one time or another in physiology classes I suspect that some of the issues involved in the calculation are not as well understood as they should be.

There are some similarities between the deadspace-tidal volume ratio (Vd/Vt) and the shunt fraction but even though they are both are involved in gas exchange (and to some extent they also correlate with each other) they are measuring different things. When blood flows through the lung some blood passes through well ventilated alveoli and becomes fully saturated; some blood passes through poorly ventilated alveoli and is only partially saturated; and some bypasses the alveoli entirely. The resulting arterial oxygen content is the summed average of all of these compartments.

There are two different ways that shunt fraction can be measured and calculated; physiological and anatomical. The physiological shunt equation can be performed at any FiO2 (but usually around the FiO2 of room air) and requires that arterial and mixed venous blood samples be taken more or less simultaneously and then analyzed for PO2 and SaO2. The basic formula is:

Where:

Qs = blood flow through shunt

Qt = total blood flow

Cc’O2 = pulmonary capillary O2 content

CaO2 = arterial O2 content

CvO2 = mixed-venous O2 content

Oxygen content is the milliliters of oxygen per liter of blood and is calculated from:

Where:

Hb = hemoglobin (grams/decaliter)

SO2 = oxygen saturation (%)

PO2 = oxygen partial pressure

The pulmonary capillary O2 content cannot be measured directly (and strictly speaking it is more of a conceptual value than a real one) and is usually estimated from the alveolar air equation (although “ideal” pulmonary capillary blood has a PO2 gradient of about 1 mm Hg from alveolar air this is insignificant enough that it is usually ignored).

Where:

Pb = barometric pressure in mm Hg

FiO2 = fractional concentration of inspired oxygen

PaCO2 = arterial partial pressure of CO2

RER = respiratory exchange ratio

The oxygen content of the pulmonary capillaries is determined by first estimating the oxygen saturation from the PAO2 and this can be done either visually from the oxygen dissociation curve:

or from Severinghaus’s formula:

and then calculating Cc’O2 accordingly.

Note: Interestingly neither the oxygen dissociation curve nor Severinghaus’s formula take carboxyhemoglobin (or methemoglobin) into account. For that matter, this issue has not been included in any of the textbook discussions of shunt fraction I’ve read. COHb skews the relationship between PO2 and SO2 (downwards if you’re working from PO2 to SO2, upwards if you’re working from SO2 to PO2). Normal COHb levels in non-smokers are 1-2 and this amount of COHb is unlikely to make a significant difference in the shunt fraction calculations. In the absence of any firm guidelines however, when higher levels of COHb are present they should probably be used to adjust Cc’O2 accordingly.

Taking normal values and working backwards, PAO2 is:

The pulmonary capillary oxygen saturation is therefore:

And the pulmonary capillary oxygen content is:

Mixed venous blood nominally has a PO2 of 40 and an oxygen saturation of 75%, so:

CaO2 will then be calculated from an individual’s actual PaO2 and SaO2. Depending on the specific results the shunt fraction will be:

 

The physiological shunt fraction can only be calculated when both the arterial and the mixed-venous PO2 and SO2 are known. For this reason it is most often performed in a cardiac cath lab, operating room or intensive care unit where indwelling arterial and central venous lines are relatively common. The physiological shunt calculation cannot differentiate between the shunting caused by poorly ventilated alveolar units and that from an anatomical shunt, however. The anatomical shunt fraction can be calculated by a separate procedure however, and this is where the 100% O2 test comes into play.

By having a patient breath 100% O2 until the nitrogen has been washed out of their lung (nominally 20 minutes), the oxygen concentration in even poorly ventilated units will approach 100%. This means that the partial pressure and saturation of blood leaving both the poorly and the well ventilated alveolar units will be the same. For this reason, any decrease in the arterial oxygen content will be due solely to an anatomical shunt.

If a patient has an indwelling central venous catheter, the calculation of anatomical shunt can proceed the same way as already detailed. If only an arterial sample can be obtained (which is usually the case in a PFT Lab) an arterial-venous O2 content difference of between 4.4 and 5.0 can be assumed and the shunt fraction calculated accordingly.

The limitations of the shunt fraction calculations have to do in part with some of the assumptions about normal values and in part with the accuracy of blood gas measurements. The alveolar air equation, for example, assumes that the respiratory exchange ration (RER) is 0.8 but the only way to be sure is by actually measuring VO2 and VCO2. Strictly speaking, an RER that is different than 0.8 will probably not make a significant difference in calculated PAO2, Sc’O2 and Cc’O2, but it is still an assumption. Using an a-v O2 content difference of 4.4 to 5.0 on the other hand, is a much larger assumption. It is justified to some extent by the fact that the 100% O2 test is usually made at rest and these are reasonable values for an individual at rest but again, it is an assumption.

Far more concerning are the limitations in accurately measuring PaO2 and SaO2, particularly at higher FiO2’s. Two different studies have shown that the type of syringe used to obtain an ABG (glass versus plastic) and how it was stored (on ice or at room temperature) made a significant difference in the calculated shunt fraction even when the ABG samples were analyzed quickly. When there was a longer wait before analysis the error in PaO2 could cause the calculated shunt fraction to be twice as large as it really was. The reason this happens is partly due to diffusion through the plastic syringes and partily to continued metabolism within a blood sample when kept at room temperature. The least amount of change was seen when glass syringes kept on ice.

Interestingly, a similar study with ABG samples taken at normal FiO2 (PO2 ≈ 100) showed the opposite effect. The measured PO2 tended to rise, again more in plastic syringes than in glass, and again this likely due to diffusion. Interestingly, PO2 fell in glass syringes kept on ice and the authors, Knowles et al, point out that the solubility of O2 rises as temperature falls and that with more O2 in solution PO2 may decrease.

Finally, blood gas analyzers are usually calibrated using gas concentrations in the normal physiological range. Any arterial blood sample with PO2 above 200 mm Hg is well outside this range and I am concerned about what kind of error bar there is for PO2’s that are even higher. Pretto et al used blood tonometered with 95% O2 and 5% CO2 but interestingly they did not report the measured PO2 but only the change in PO2 over time. Smeenk et al obtained blood samples from individuals undergoing the 100% oxygen test as a pre-op assessment for coronary bypass surgery and the average PO2 of their gold standard samples (glass syringe, iced, 5 minute delay) was 590 mm Hg. This is an A-a gradient of around 80 mm Hg and may well be appropriate, but it also means that the average anatomical shunt fraction was 10% and Cotes et al indicates that the normal anatomical shunt for individuals in the same age range is around 4%.

The shunt fraction test is not commonly performed in pulmonary function labs. True anatomic shunts are relatively rare and the most appropriate patient for the 100% O2 shunt fraction test would be one with a reduced SaO2 at rest that does not significantly improve with supplemental O2.

The physiological shunt fraction could be considered the reverse side of Vd/Vt. Perfusion inhomogeneities exist just as much as ventilation inhomogeneities but this may be overlooked because pulmonary function testing is oriented far more around the ventilation side of respiration than the perfusion side. Ventilation and perfusion inhomogeneities are core features of many pulmonary diseases. For this reason the shunt fraction and the differences between its physiological and anatomical components need to be part of the education of all pulmonary technologists. Like Vd/Vt however, there are also limitations to the accuracy of the shunt fraction calculation both from assumptions that may or may not be reasonable, and from the measurement accuracy of PO2 and SO2.

References:

Aboab J, Louis B, Jonson B, Brochard L. Relation between PaO2/FiO2 and FiO2: a mathematical description. Chapter in Applied Physiology in Intensive Care Medicine, Pinsky MR, Brochard L, Mancebo JM editors. Springer-Verlag Heidelberg, 2006.

Conrad SA, Kinasewitz GT, George RB. Pulmonary Function Testing. Principles and Practice. Churchill Livingston Publishing, 1984.

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

Knowles TP, Mullin RA, Hunter JA, Douce FH. Effects of syringe material, sample storage time and temperature on blood gases and oxygen saturation in arterialized human blood samples. Respir Care 2006; 51(7): 732-736.

Pretto JJ, Rochford PD. Effects of sample storage time, temperature and syringe type on blood gas tensions in samples with high oxygen pressures. Thorax 1994; 49: 610-612.

Smeenk FWJM, Janssen JDJ, Arends BJ, Harff GA, van den Bosch JA, Schonberger JPAM, Postmus PE. Effects of four different methods on sampling arterial blood and storage time on gas tensions and shunt calculation in the 100% oxygen test. Eur Respir J 1997; 10: 910-913.

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

I often find topics for this blog in a sideways fashion. Recently while searching for something else I ran across an article about the minimum clinically important difference (MCID) of the Residual Volume (RV) in patients with emphysema. I’ve come across the MCID concept before but I had never really followed up on it. This time I started researching MCID and immediately ran across a number of articles about the MCID of the 6-minute walk test (6MWT). This got me to review the articles I have on hand and I found that since I last wrote about the 6MWT I’ve accumulated quite a few new (or at least new to me) reference equations as well as a number of articles about performance issues. Given all this how could I not re-visit the 6MWT?

In addition to the 6 reference equations I had previously I’ve found another 13 female and 14 male reference equations for the 6MWT (total 19 female, 20 male) which is an opportunity to re-visit the selection process. This immediately raises the question about what factors should be used to calculate the predicted 6-minute walk distance (6MWD). Because the 6MWT is essentially an exercise test age has an obvious effect on exercise capacity so it is no surprise that with the exception of one set all of the reference equations consider age to be a factor. It should be noted however, that many of the reference equations are intended to be only applied over a limited range of ages and this may limit their utility.

Given the fact that stride length and therefore walking speed are directly related to height it is somewhat surprising to find that only twelve of the male and eleven of the female reference equations consider height to be a factor. When height is a factor, the predicted 6MWD is usually affected something like this:

Weight also affects exercise capacity but an interesting question is whether the observed 6MWD should be compared to a predicted 6MWD based on a “normal” weight or whether the 6MWD should be adjusted to the individual’s actual weight and assessed accordingly. To some extent this is already an issue in current PFT predicted equations. For example, weight is not a factor in any of the FVC or TLC reference equations and when lung volumes are decreased in the presence of obesity they are considered to be abnormal. On the other hand, the reference equations I use for maximum oxygen consumption during a CPET include weight as a factor and for a number of reasons this is likely the correct approach. For this last reason I would think that weight should be a factor and ten of the reference equation sets consider weight (or BMI) to be a factor. When weight is a factor, the predicted 6MWD is usually affected like this:

Interestingly two sets of equations consider the maximum heart rate attained during the 6MWT (as a percent of predicted) to be a factor. When a closer look is taken at these equations what can be seen is that the higher the maximum heart rate is during the 6MWT, the higher the 6MWD is expected to be. This could be because a higher heart rate could be seen as the result of a greater patient effort but a known effect of deconditioning is a higher heart rate for a given work load and conversely, fit individuals have a lower heart rate for the same workload. There seems to be a bit of disconnect between the heart rate and the 6MWD and for this reason I have trouble seeing how including heart rate as a factor improves the accuracy of a predicted 6MWD.

Finally, one set of equations uses an individual’s observed FEV1 as a factor. This is the same set of equations that does not use age as a factor and is therefore using FEV1 as substitute. This is an interesting notion since a reduced FEV1 occurs with both age and lung disease and both will have an effect on exercise capacity but it also means the formula is equating age and lung disease and this may be a bit too simplistic. It also means that the predicted 6MWD is being adjusted for an individual’s lung disease and I don’t think this is appropriate.

As usual I’ve graphed the predicted values for individuals with an average height and weight (and where needed a normal max HR and FEV1). Also as usual, there is a large scatter among the predicted 6-minute walk distances which makes deciding which reference equation to use more than a little difficult.

To some extent I am bothered more by the higher predicted 6-minute walk distances than I am by the low ones. My lab has used Enright et al’s 1998 reference equations ([I] on both graphs) for over 15 years and the results seem to correlate with the functional capacity of our patients quite well. In particular however, it is very rare for us to get anybody with a 6MWD that is above 100% of predicted. As a comparison we get more CPET patients with a max VO2 above 100% of predicted than patients with a 6MWD above 100% of predicted and every year we perform 3-4 times as many 6MWTs as we do CPETs. Despite this the predicted 6MWD’s from Enright’s 1998 reference equations are mostly below the average of the predicted 6MWD’s and around 20% below the highest predicted 6MWD’s.

Because these high predicted 6MWD’s bothered me I did a quick search and found that when walking speeds reach 1.9 to 2.1 meters/second most individuals transition to a jog or a slow run. I’m not a kinesthesiologist but it seems to me that for an individual to be able to walk faster than about 2 meters/sec (7.2 km/hour, 4.4 MPH) they would have to adopt an exaggerated walking pattern which may not be a jog but is also probably not a “normal” walk either. A number of the reference equations would require an individual to walk faster than the usual walk-to-jog transition speed (although this applies more to males than to females) in order to attain a “normal” 6-minute walk distance and this makes me a bit skeptical of these specific reference equations.

Having said all that, more than one investigator has noted significant ethnic differences in 6MWD’s and that reference equations should be selected accordingly. It may well be that Enright’s 1998 equations have hit a “sweet spot” for my lab (even though we service an ethnically diverse population). It may also be that we aren’t expecting enough from our patients and they should be doing better than we think they should.

The physical layout of the 6-minute walk test will also have an effect on the 6MWD. The ATS/ERS standards specify a 30 meter (or 100 foot) track length (distance between turnaround points) but this distance may not always be available. At least one study has shown that a shorter track distance (10 meters) led to a lower 6-minute walk distance (not really surprising given that the time and distance spent during turnaround don’t count towards the 6MWD). Conversely, another study showed that a completely circular track (no turnaround points) led to slightly longer 6MWD’s. This may be important because a number of the studies that generated reference equations used track lengths different than the ATS/ERS standards (10-20 and 40-50 meters). It’s worth mentioning that several (not all) of the equations that predict the highest 6-minute distance were performed on tracks longer than what the ATS/ERS specifies. For these reasons whatever 6MWT track layout is available to a lab should have some affect on either the selection or the interpretation of the 6MWD.

When selecting a 6MWD reference equation it seems to me that the most relevant ones are those that include height and weight (or BMI) as factors. Ethnicity (or at least a similarity in location) of the study population may also be important and should be considered as well. An ideal situation would be to have a record of previous 6MWD’s and the relative disability of the patient’s performing them, and to use these when comparing different reference equation. Regardless of how a 6MWD reference equation is selected, once put to use it should be monitored for a while. If too many patients have either an elevated or overly reduced 6MWD (factoring in relative disability, of course), then it may be necessary to look at a different reference equation.

Even when the correct reference equation has been selected one of the primary uses of the 6MWT is to assess changes in functional capacity after an intervention of one kind or another (rehabilitation, drug changes, surgery, therapy). This is where the minimum clinically important difference (MCID) comes into play.

MCID has been defined as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate, in the absence of troublesome side effects and excessive cost, a change in the patients management”. The MCID is not the same as a statistically significant change since what is being measured is clinical and not statistical relevance. There appear to be two different approaches towards measuring the MCID (Anchor and SEM) but it’s not clear that either one generates values that are significantly different from each other. One important point however, is that the MCID is a change in 6MWD distance and not a percent change.

The MCID for the 6MWD has been determined for a number of conditions:

MCID (meters):	Condition:	Reference:
25	CAD	Gremeaux et al.
25-33	Chronic Resp. Dis.	Singh et al
30	COPD	Polkey et al
31	COPD	Puhan et al
54-80	COPD	Wise et al
54	COPD	Laviolette et al
42	COPD	Hernandes et al
24-45	IPF	du Bois et al
28	IPF	Swigris et al
30.5	IPF	Holland et al
33	PAH	Mathai et al

It would be nice if there was more of a consensus on the 6MWD MCID, but given the diversity of the underlying conditions and of their severity among the study populations this isn’t surprising. In addition the MCID values do not really take the baseline condition of an individual patient into consideration and are applied to their study populations as a whole. It’s also important to note that Hernandes et al showed there is a learning effect and that on average patients with COPD improved by 27 meters from their first to their second 6MWT test.

MCID analysis should therefore not be performed when comparing a patient’s first and second tests. In addition when a post-intervention MCID comparison is made, some allowance should be made for the patient’s baseline condition, with a lower number used for more severe conditions and a higher for less severe. Even so, the most recent ERS/ATS statement on the 6MWT (Singh et al) indicates that the MCID is 25-33 meters for patients with chronic respiratory diseases and this is likely to be what most labs use as a standard.

The 6-minute walk test is widely used as an assessment of functional status. It requires a minimum amount of equipment but an attention to detail is still necessary and the ERS/ATS standards should be closely followed whenever possible. The results of a 6MWT are non-specific but they relate well to a patient’s exercise capacity. The 6MWD reference equations are useful as a yardstick when assessing of the severity of any functional limitations but there are limitations to their accuracy and it is the changes that occur in an individual’s 6MWD that are more significant. Although there is a lack of consensus on the values for the minimally clinically important difference, when assessing increases or decreases in the 6MWD the MCID remains a more appropriate tool than a percent change.

6MWD Reference Equation Demographics:

Reference:	Age Range:	Number:	Distance (meters)	Ethnicity:
[A]	18-50	127	30	Arab
[B]	40-90	181	10	Caucasian
[C]	40+	229	40	North African
[D]	29-67	617	30	Brazilian
[E]	55-75	70	45	Australian
[F]	20-60	323	30	Caucasian, Obese
[G]	40-80	238	30	Caucasian
[H]	20-50	48	30	Caucasian
[I]	43-77	117	30.5	Caucasian
[J]	77 +/- 4	2117	30.5	Caucasian
[K]	20-80	79	20	Canadian
[L]	45-85	70	30	Canadian
[M]	14-84	61	30	Brazilian
[N]	45-85	109	45	Australian
[O]	40-80	155	30	Tunisian
[P]	20-80	175	30	Chilian
[Q]	45-85	35	45	Chinese
[R]	20-80	132	30	Brazilian
[S]	50-85	29	50	Caucasian
[T]	40-60	101	30	North India

Male 6MWD Reference Equations:

Reference:
[A]	(2.81 x Ht(cm)) – (age x 0.79) – 28.5
[B]	1266 − (7.8 x age) − (5.92 x BMI)
[C]	720.5 – (5.14 x age) – (2.23 x Wt(kg)) + (2.7198 x Ht(cm))
[D]	890.46 – (6.11 × age) + (0.0345 × age^2) – (4.87 × BMI) + 48.87
[E]	64.69 + (3.12 x Ht(cm)) + (23.29 x FEV1 (L))
[F]	894.2177 − (2.07 x age) – (5.51663 x BMI)
[G]	361 + (2 x Ht(cm)) + ((max hr/pred max hr x 100) x 3) – (4 x age) – (1.5 x Wt(kg))
[H]	518.853 + (1.25 x Ht(cm)) – (age x 2.816)
[I]	(7.57 x Ht(cm)) – (5.02 x age) – (1.76 x Wt(kg)) – 309
[J]	493 + (2.2 * Ht(cm)) – (0.93 x Wt(kg)) – (5.3*age) + 17
[K]	868.8 – (2.99 x age)
[L]	970.7 – (5.5 x age) + 56.3
[M]	622.461 – (age x 1.846)
[N]	867- (5.71 x age) + (1.03 x Ht(cm))
[O]	299.8 – (4.34 x age) + (3.426 x Ht(cm)) – (1.46 x Wt(kg)) + 62.5
[P]	530 – (3.31 x age) + (2.36 x Ht(cm)) – (1.49 x Wt(kg))
[Q]	(5.50 x ((HR max/pred HR max)*100)) + (6.94 x Ht(cm)) – (4.49 x age) – (3.51 x Wt(kg)) – 473.27
[R]	390 + (2.14 x Ht(cm)) – (2.37 x age) – (2.34 x BMI)
[S]	269.31 + (5.14 x Ht(cm)) – (5.78 x age) – (2.29 x Wt(kg))
[T]	127.121 + (3.654 × Ht(cm) − (4.139 × age)

Female 6MWD Reference Equations:

Reference:
[A]	(2.81 x Ht(cm)) – (age x 0.79) – 28.5
[B]	1064 − (5.28 x age) − (6.55 x BMI)
[C]	560.5 – (5.14 x age) – (2.23 x Wt(kg)) + (2.7198 x Ht(cm))
[D]	890.46 – (6.11 × age) + (0.0345 × age^2) – (4.87 × BMI)
[E]	64.69 + (3.12 x Ht(cm)) + (23.29 x FEV1 (L))
[F]	894.2177 − (2.07 x age) – (5.51663 x BMI) – 51.4489
[G]	331 + (2 x Ht(cm)) + ((max hr/pred max hr)*100) x 3) – (4 x age) – (1.5 x Wt (kg))
[H]	479.783 + (1.25 x Ht(cm)) – (age x 2.816)
[I]	(2.11 x Ht(cm)) – (2.29 x Wt(kg)) – (age x 5.78) +611
[J]	493 + (2.2 * Ht(cm)) – (0.93 x Wt(kg)) – (5.3*age)
[K]	794.1 – (2.99 x age)
[L]	970.7 – (5.5* age)
[M]	560.961 – (age x 1.846)
[N]	525 – (2.86 x age) + (2.71 x HT(cm)) – (6.22 x BMI)
[O]	299.8 – (4.34 x age) + (3.426 x Ht(cm)) – (1.46 x Wt(kg))
[P]	457 – (3.46 x age) + (2.61 x Ht(cm)) – (1.57 x Wt(kg)),
[Q]	(5.50 x ((HR max/pred HR max)*100)) + (6.94 x Ht(cm)) – (4.49 x age) – (3.51 x Wt(kg)) – 473.27
[R]	683 + (0.91 x Ht(cm)) – (3.94 x age) – (3.57 x BMI)
[S]	269.31 + (5.14 x Ht(cm)) – (5.78 x age) – (2.29 x Wt(kg)) – 51.31

References:

[A] Alameri H, Al-Majed S, Al-Howaikan. Six-min walk test in a health adult Arab population. Resp Med 2009; 103: 1041-1046.

ATS Statement: Guidelines for the six-minute walk test. Amer J Respir Crit Care Med 2002; 166: 111-117.

Bansal V, Hill K, Delmage TE, Brooks D, Woon LJ, Goldstein RS. Modifying track layout from straight to circular has a modest effect on the 6-min walk distance. Chest 2008; 133: 1155-1160.

Beekman E, Mesters I, Hendriks EJM, Klaassen MPM, Gosselink R, van Schayck OCP, de Bie RA. Course length of 30 metres versus 10 metres has a significant influence on six-minute walk distance in patients with COPD: an experimental crossover study. J Physiotherapy 2013; 59: 169-176.

[B] Beekman E, Mesters I, Gosselink R, Klaassen MPM, Hendriks EJM, Van Schayck OCP, de Bie RA. The first reference equations for the 6-minute walk distance over a 10 m course. Thorax 2014; 69(9): 867-868

[C] Ben Saad H, Prefaut C, Tabka Z, Hadj Mtir A, Chemit M, Hassasoune R, Ben Abid T, Zara K, Mercier G, Zbidi A, Hayot M. 6-minute walk distance in health North Africans older than 40 years: Influence of parity. Resp Med 2009; 103: 74-84.

[D] Britto RR, Probst VS, de Andrade AFD, Samora GAR, Hernandes NA, Marinho PEM, Karsten M, Pitta F, Parreira VF. Reference equations for the six-minute walk distance based on a Brazilian multicenter study. Brazilian Journal of Physical Therapy 2013; 17(6): 556-563.

[E] Camarri B, Eastwood PR, Cecins NM, Thompson PJ, Jenkins S (2006) Six minute walk distance in healthy subjects aged 55–75 years. Resp Med 2006; 100: 658–665.

[F] Capodaglio P, De Souza SA, Parisio C, Precilios H, Vismara L, Cimolin V, Brunani A. Reference values for the 6-min walk test in obese subjects. Disability & Rehabilitation 2013; 35(14): 1199-1203.

[G] Casanova C, et al. The 6-min walk distance in health subjects: reference equations from seven countries. Eur Respir J 2011; 37: 150-156.

Carter R, Holiday DB, Nwasuraba C, Stocks J, Grothus C, Tiep B. 6-minute walk work for assessment of functional capacity in patients with COPD. Chest 2003; 123: 1408-1415.

[H] Chetta A, et al. Reference values for the 6-min walk test in healthy subjects 20-50 years old. Respir Med 2006; 100: 1573-1578.

Cote CG, et al. Validation and comparison of reference equations for the 6-min walk distance test. Eur Respir J 2008; 31: 571-578.

Du bois, RM, Weycker D, Albera C, Bradford WZ, Costabel U, Kartashov A, Lancaster L, Noble PW, Sahn SA, Szwarcberg J, Thomeer M, Valeryre D, King TE, Six-minute walk test in idiopathic pulmonary fibrosis. Test validation and minimally clinically important difference. Am J Respir Crit Care Med 2011; 183(9): 1231-1237.

[I] Enright PL, Sherril DL. Reference Equations for the six-minute walk in healthy adults. Amer J Respir Crit Care Med 1998; 158: 1384-1387.

[J] Enright PL, McBurnie MA, Bittner V, Tracy RP, McNamera R, Arnold A, Newman AB. The 6-min walk test: A quick measure of functional status in elderly adults. Chest 2003; 123: 387-398.

[K] Gibbons WJ, Fruchter N, Sloan S, Levy RD. Reference values for a multiple repetition 6-minute walk test in healthy adults older than 20 years. J Cardiopulmonary Rehab and Prevention 2001; 21: 87–93

Gremeaux V, Troisgros O, Benaim S, Hannequin A, Laurent Y, Casillas JM, Benaim C. Determining the minimal clinically important dierence for the six-minute walk test and the 200-meter fast-walk test during cardiac rehabilitation program in coronary artery disease patients after acute coronary syndrome.. Archives of Physical Medicine and Rehabilitation, 2011; 92 (4): pp.611-619.

Hernandes NA, Wouters EFM, Meijer K, Annegarn J, Pitta F, Spruit MA. Reproducibility of 6-minute walking test in patients with COPD. Eur Respir J 2011; 38: 261-267.

[L] Hill K, Wickerson LM, Woon LJ, Abady AH, Overend TJ, Goldstein RS, et al (2011) The 6-min walk test: responses in healthy Canadians aged 45 to 85 years. Applied Physiology, Nutrition, and Metabolism 2011; 36: 643–649.

Holland AE, Hill JC, Conron M, Munro P, McDonals CF. Small changes in six-minute walk distance are important in diffuse parenchymal lung disease. Resp Med 2009; 103: 1430-1435.

Holland AE et al. An official European Respiratory Society/American Thoracic Society technical standard: field walking tests in chronic respiratory disease. Eur Respir J 2014; 44: 1428-1446.

[M] Iwama AM, Andrade GN, Shima P, Tanni SE, Godoy I, Dourado VZ. The six-minute walk test and body weight x walk distance product in healthy Brazilian subjects. Braz J Med Biol Res 2009; 42: 1080-1085.

[N] Jenkins S, Cecins N, Camarri B, Williams C, Thompson P, Eastwood P. Regression equations to predict 6-minute walk distance in middle-aged and elderly adults. Physiotherapy Theory and Practice 2009; 25(7): 516-522.

Larsson UE, Reynisdottir S. The six-minute walk test in outpatients with obesity: reproducibility and known group validity. Physiother Res Int 2008; 13(2): 84-93.

Laviolette L, Bourbeau J, Bernard S, Lacasse Y, Pepin V, Preton M-J, Baltzan M, Rouleau M, Maltais F. Assessing the impact of pulmonary rehabilitation on functional status in COPD. Thorax 2008; 63: 115-121.

[O] Masmoudi K, Aouicha MS, Fki H, Dammak J, Zouari N. The six minute walk test: which predictive values to apply for Tunisian subjects aged between 40 and 80 years? Tunis Med 2008; 86(1): 20–26.

Mathai SC, Puhan MA, Lam D, Wise RA. The minimal important difference in the 6-minute walk test for patients with pulmonary arterial hypertension. Am J Respir Crit Care Med 2012; 186(5): 428-433.

[P] Osses AR, Yanez VJ, Barria PP, Palacios MS, Dreyse DJ, Diaz PO, et al. Reference values for the 6-minutes walking test in healthy subjects 20–80 years old. Revista Medica de Chile 2010; 138: 1124–1130.

[Q] Poh H, Eastwood PR, Cecins NM, Ho KT, Jenkins SC. Six-minute walk distance in healthy Singaporean adults cannot be predicted using reference equations derived from Caucasian populations. Respirology 2006; 11: 211–216.

Polkey MI, et al. Six-minute-walk test in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013; 187(4): 382-386.

Puhan MA, Chandra D, Mosenifir Z, Ries A, Make B, Hansel NN, Wise RA, Sciurba F. The minimal important difference of exercise tests in severe COPD. Eur Respir J 2011; 37: 784-790.

Singh SJ et al. An official systematic review of the European Respiratory Society/American Thoracic Society: measurement properties of field walking tests in chronic respiratory diseases. Eur Respir J 2014; 44: 1447-1478

[R] Soares MR, Pereira CA. Six-minute walk test: reference values for healthy adults in Brazil. The Jornal Brasileiro de Pneumologia 2011; 37: 576–583.

Swigris JJ, Wamboldt FS, Behr J, du Bois RM, King TE, Raghu G, Brown KK. The 6 minute walk in idiopathic pulmonary fibrosis: longitudinal changes and minimum important difference. Thorax 2010; 65: 173-177.

[S] Troosters T, Gosselink R, Decramer M. Six minute walking distance in healthy elderly subjects. Eur Respir J 1999; 14: 270-274.

[T] Vaish J, Ahmed F, Sinigla R, Shukla DK. Reference equation for the 6-minute walk test in healthy North Indian males. Int J Tuberc Lung Dis 2013; 17(4): 698-703.

Wise RA, Brown CD. Minimally clinically important differences in the six-minute walk test and the incremental shuttle walking test. J COPD 2005; 2(1): 125-129.

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

For the first dozen or so year that I worked in a pulmonary function lab it was with counter-weighted, volume-displacement water-seal spirometers more or less like this:

Patients would do a series of tests and I’d end up with a bunch of pen traces on kymograph paper that I’d have to measure with a ruler and use a desktop calculator (it was about a foot square, weighed a couple of pounds and had a nixie tube digital display) to create a hand-written report. I’m not going to suggest that these spirometers were in any way better than what we’re using now but I have to say that I would have seen the following problems more or less immediately.

Recently I was reviewing a report from a patient with very severe obstruction and noticed something a bit off about the flow-volume loop. Specifically, the end-exhalation of the tidal loop looked like it was at a significantly higher volume than the end of the FVC effort.

Because the high-frequency sawtooth pattern (from the patient, not the equipment) makes it a little hard to see if this is what was really happening, I downloaded the raw data and re-graphed the volume-time curve with a spreadsheet.

When I did this it was first evident that the patient started the inhalation for the FVC maneuver before they’d finished exhaling. Next, despite exhaling for 13.8 seconds (the entire trace spans about 44 seconds) the patient did not exhale all the way back to the same end-exhalation level they’d had while breathing tidally. This is actually a sign of gas trapping and relatively common with severe obstruction. It also means that the inhaled volume from the lowest end-exhalation to the maximum inhalation was about a half a liter larger than the reported FVC.

The reason that I say that this is something I would have noticed on older equipment is that with our lab software the volume-time curve looks like this:

which only shows one second of test data before the start of the FVC’s maximal exhalation. This means the baseline shift that occurred during the FVC maneuver does not show up at all.

Strictly speaking I doubt that the FIVC volume derived from the more complete volume-time tracing meets the official criteria for an inspiratory vital capacity, particularly since it occurs over more than one breath. Even so, it is “real” and the gas trapping it shows is an important clinical finding. Even more importantly I don’t know of any test systems where this would have been evident.

So, if that flow-volume loop showed gas trapping, what about this one?

Again, the end-exhalation of the tidal breath is at a higher volume than the one at the end of the FVC maneuver. This happened for a completely different reason than the first example however, and an important clue is that it was performed on one of our test systems that has volume displacement spirometer.

When I downloaded the raw data for this test and graphed it using a spreadsheet, what I got was this:

The real difference is that the end of exhalation for all of the tidal breaths showed a consistent drift and what was most likely happening was that the patient was leaking around the mouthpiece. The weight of the spirometer bell is enough to maintain a small amount of positive pressure inside the spirometer and if there is a leak then the bell will drop during testing.

When I analyzed the trace more carefully what I found was:

There was a more-or-less constant drift of about 0.085 L/sec. That means that during the ~10 seconds the patient exhaled, the FVC was probably underestimated by about 0.85 L. This has to be a guesstimate however, since there is no way to be sure the drift was constant.

Again, this is a problem that would have showed up immediately on the kymograph paper of an older spirometer but wasn’t evident at all in my lab’s testing systems. To some extent the problem is related to the use of a test system with a volume-displacement spirometer but patient or equipment leaks can occur on any test system and their effect on the results are probably going to be less evident than they were here.

I’m not advocating that we return to the “good old days” of counter-weighted, volume-displacement water-seal spirometers and kymograph paper. What I do feel however, is that all too often our test systems and the choices made by engineers, programmers and manufacturers end up limiting instead of enhancing what we can see. There were clues in the flow-volume loops that something wasn’t right but I had to do some serious digging in order to find out what was actually happening. I don’t expect our test systems to be able to correct the results for gas trapping or for a leak but I also don’t expect for them to make it more difficult than it should be to detect that these things were happening.

I’d like to suggest that all test systems come with a “raw data” or “diagnostic” mode where all of the unprocessed information collected during a test is quickly and easily available. To people like myself and anybody who needs to troubleshoot problems, this is a complete no-brainer, but how useful would it be to regular staff? I’ve visited a lot of PFT labs and met a lot of technicians and many of them seem uninterested in how test results are derived. The possible reasons for this could fill a couple blogs but I have to wonder if at least part of the reason is that for last several decades all our test systems have been a “black box”. Patient test data goes in one side, numerical and graphical results come out the other side and it isn’t possible to see what happens in-between. How can staff ever be interested and knowledgeable when the processes they work with every day are opaque? Quick access to “raw” data would be a great teaching tool but if not that then at the very least it would allow staff to easily get a better look at questionable results.

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

The patients whose reports I review have always been very accommodating. An issue of one kind or another catches my attention and before I know it I find several more reports that are similarly involved. Thanks to our patients I’ve had a number of reports come across my desk recently that showed a combination of restrictive and obstructive defects. This particular one may not be the best possible example but it seems to illustrate several points fairly well.

	Observed:	%Predicted:	Predicted:
FVC (L):	1.12	40%	2.80
FEV1 (L):	0.75	35%	2.16
FEV1/FVC (%):	67	86%	78
TLC (L):	1.92	42%	4.54
FRC (L):	1.18	48%	2.47
RV (L):	0.76	44%	1.73
RV/TLC (%):	40	104%	38

Interpreting results like this as combined (or mixed) defects using the ATS/ERS algorithm seems relatively straightforward.

The algorithm starts by using the FEV1/FVC ratio to determine whether obstruction is present and only then considers whether or not the FVC and TLC are normal. It occurred to me however, that this assumes that the normal range of the FEV1/FVC ratio is preserved when TLC decreases below normal. Given the markedly different causes of restrictive lung disease it would seem that saying that the FEV1/FVC ratio should remain within the normal range over a relatively broad range of lung capacities (and without necessarily knowing the cause for any reduction) seems a bit far-fetched. Interestingly enough however, it actually turns out to be reasonably true.

In general, there are three categories of restrictive diseases; interstitial, chest wall and neuromuscular. I’ve been unable to find any studies that specifically studied the relationship between the FEV1/FVC ratio and TLC but there are numerous studies of these disorders where spirometry and lung volumes were performed.

In regards to interstitial disease, a study of over 200 patients with idiopathic pulmonary fibrosis and a reduced TLC showed that the FVC was decreased more than the FEV1 (61% vs 71%). Similarly, a study with a small number of patients with Sjogren’s disease (an autoimmune disease that can cause interstitial lung disease) showed that those individuals with a reduced TLC had an FVC that was decreased slightly more than the FEV1.

In regards to chest wall diseases, several studies on patients with ankylosing spondylitis also showed that in those individuals with a reduced TLC that the FVC decreased more than the FEV1. Finally, in regards to neuromuscular disease a study of 25 patients with ALS, the FVC was decreased slightly more than the FEV1.

This means that at least for these examples of restrictive disease the FEV1/FVC ratio is preserved and maybe even slightly elevated. Given this, it does appear that a reduced FEV1/FVC ratio in combination with a reduced TLC is a probably a reliable indicator of a mixed defect. It could even be argued that because in all of these cases the FVC decreased somewhat more than the FEV1 the lower limit of normal (LLN) for the FEV1/FVC ratio should increase when restriction is present. Since the relationship between FVC and FEV1 differs somewhat from one category of restriction to another and because the cause of restriction is often unknown making an accurate adjustment of this kind does not seem to be realistic.

Once a set of results have been interpreted as having a mixed defect however, it turns out that assigning a level of severity to the restrictive and obstructive components is more than somewhat problematic. Part of the reason for this is that the ATS/ERS standards do not include an index of severity for a reduced TLC and instead just say that a TLC below the LLN indicates restriction.

Note: The recommendation that the LLN be used to determine restriction is somewhat problematic due to the fact that most lung volume reference equations (which often come from several decade-old studies) do not include an LLN. LLN therefore must be calculated using the statistics for the entire study population which means the LLN becomes a fixed value that is subtracted from the predicted TLC regardless of an individual’s height or age. Moreover, the TLC itself is often adjusted using correction factors for ethnicity and there are no recommendations on how to adjust the LLN for this. Finally, the predicted lung volume values are also often modified by the insertion of an FVC from a different set of reference equations and this can skew the TLC away from its original reference value without any modification of the LLN. For these (and other) reasons many labs just use 80% of predicted as a cutoff instead of the LLN.

The severity index my lab uses for TLC has been handed down from generation to generation and originated at the NIH in the 1970’s. I can’t say that it is 100% correct or that it has been clinically verified but it is in common use in many PFT labs:

TLC ≥ 80%	Normal
TLC ≥ 60% and <80%	Mild
TLC ≥ 40% and <60%	Moderate
TLC < 40%	Severe

The other problem is that the ATS/ERS recommendations for the severity of airway obstruction are based solely on the FEV1 percent predicted.

ATS:	Mild	FEV1 ≥70% predicted
	Moderate	FEV1 ≥60% and ≤70% predicted
	Moderately Severe	FEV1 ≥50% and ≤60% predicted
	Severe	FEV1 ≥35% and ≤50% predicted
	Very Severe	FEV1 <35% predicted

In a mixed defect however, the FEV1 is reduced not just because of obstruction but because of restriction as well. At least two different approaches towards correcting this have been suggested.

Balfe et al proposed using the FEV1/FVC ratio to assess severity. This approach was originally a recommendation from the Intermountain Thoracic Society’s 1984 manual. Specifically the FEV1/FVC ratio’s confidence interval (1.65 times the standard error of the estimate) is used to derive an index of obstruction severity.

ITS:	Mild	FEV1/FVC reduced by ≥1 and <2 CI
	Moderate	FEV1/FVC reduced by ≥2 and <4 CI
	Severe	FEV1/FVC reduced by >4 CI

Balfe et al showed that when the FEV1 percent predicted was used, over 90% of the patients with mixed defects were interpreted as having severe obstruction. When the FEV1/FVC ratio was used over 90% of the same patients were instead interpreted has having mild or moderate obstruction. Using the FEV1/FVC ratio from the mixed report above, the level of obstruction would be changed from severe to mild. To some extent this seems to be a reasonable approach, but is limited by the dependence of the FEV1/FVC ratio on the FVC and the quality of the expiratory effort.

Gardner et al on the other hand, proposed adjusting the FEV1 percent predicted using the TLC percent predicted. Specifically, the percent predicted FEV1 is divided by the TLC percent predicted and then the ATS/ERs criteria are used. For example, using the results from the mixed report above:

35% predicted/42% predicted = 0.83 or 83% predicted

which also changes the severity from severe to mild. Similar to Balfe et al, this approach reduced the assessed severity of 83% of the patients in a study group of 199 patients This approach also seems to be reasonable and has the advantage that it is independent from the FVC and FEV1/FVC ratio. The one criticism I would make of Gardner et al’s proposal is that the reference equations for FEV1 and TLC are derived from different populations and the effect that different reference equations than those selected for the study would have on the correction factor was not studied.

An observation made by several different investigators is that when restriction and obstruction are both present, the RV should be elevated and in fact the absence of an elevated RV was used to exclude patients from a study population in one or two instances. The point is that an elevated RV is a sign of obstructive hyperinflation and should be present in a mixed defect. It’s not clear to me however, that an elevated RV should always be present in a mixed defect, particularly given the different causes of restriction and particularly when the obstruction is “mild”. There is also the fact that an elevated RV can be due in part or in whole to a suboptimal SVC maneuver. For these reasons it would seem that an elevated RV shouldn’t be a required feature in the interpretation of a mixed defect.

One final point is that when restriction is relatively severe the amount by which a decrease in FEV1 will cause a coexisting obstructive defect to be observed becomes quite small. In the report presented above, if the FVC was actually normal the FEV1 would need to decrease by 0.30 L from its predicted value for the FEV1/FVC ratio to be 86% of predicted. In this case however, when the FVC is 40% of predicted the FEV1 only needs to be 0.11 L less than it “should” be for the FEV1/FVC ratio to be 86% of predicted. This volume is less than the required repeatability for the FVC and FEV1 (0.15 L). Since individuals with this degree of restriction are often debilitated the quality of the FEV1 (and FVC) should be carefully considered before the interpretation of a mixed defect is made.

Mixed defects, where both restriction and obstruction are present at the same time, are relatively rare. In one study with a general population of 7506 patients that had both lung volumes and spirometry performed only 151 patients (2%) satisfied the criteria for a mixed defect. The FEV1/FVC ratio is a critical component in diagnosing mixed defects and appears to remain within normal limits both within the different categories of restrictive lung disease and over a broad range of lung volumes.

Assigning severity to the restrictive and obstructive components of a mixed defect remains problematic due to limitations in the current ATS/ERS standards. Using the ATS/ERS standard for obstructive severity based on the FEV1 percent predicted appears to significantly overestimate the actual level of severity in mixed defects. At least two different proposals have been made to correct this problem and both have similar effects on the assigned severity. As always, test quality is important and the diagnosis of a mixed defect should neither be made nor should it be excluded when test quality is lacking.

References:

Balfe DL, Lewis M, Mohsenifar Z. Grading the severity of obstruction in the presence of a restrictive ventilatory defect. Chest 2002; 122: 1365-1369.

Berdal G, Halvorsen S, van der Heijde D, Mowe M, Dagfinrud H. Restrictive pulmonary function is more prevalent in patients with ankylosing spondylitis than in matched population controls and is associated with impaired spinal mobility: a comparative study. Arthritis Research & Therapy 2012; 14: R19

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

Culver BH. Obstructive? Restrictive? Or ventilatory impairment? Chest 2011; 140: 568-569

Gardner ZS, Ruppel GL, Kaminsky DA. Grading the severity of obstruction in mixed obstructive-restrictive lung disease. Chest 2011; 140: 598-603.

Grimby G, Fugl-Meyer AR, Blomstrand A. Partitioning of the contributions of rib cage and abdomen to ventilation in ankylosing spondylitis. Thorax 1974; 29: 179-184.

Diaz-Guzman E, McCarthy K, Stoller JK. Frequency and causes of combined obstruction and restriction identified in pulmonary function tests in adults. Respir Care 2010; 55(3): 310-316.

King TE, Tooze JA, Schwarz MI, Brown KR, Cherniak RM. Predicting survival in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2001; 164: 1171-1181.

Lechtzin N, Wiener CM, Shade DM, Clawson L, Diette GB. Spirometry in the supine position i

Lanier RC, Olsen GN. Can concomitant restriction be detected in adult men with airflow obstruction? Chest 1991; 99: 826-830

Segal I, Fink G, Machtey I, Gura V, Spitzer SA. Pulmonary function abnormalaties in Sjogren’s syndrome and the sicca complex. Thorax 1981; 36: 286-289.

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

My hospital has an active liver transplantation program and all transplant candidates get a full panel of PFTs in my lab. The number of liver transplant candidates we get varies from week to week but probably averages between 150 and 200 a year. As with any population a certain number of them have COPD or other lung diseases but there are some that have normal spirometry and lung volumes but a reduced DLCO. This latter group of patients likely have hepatopulmonary syndrome (HPS).

There are three hallmarks of hepatopulmonary syndrome. First is the presence of a liver disease (most commonly cirrhosis and hepatitis although liver cancer can also be a cause). Second are intrapulmonary vascular dilations (usually determined by transthoracic contrast-enhanced echocardiography). The third are gas exchange abnormalities, which include hypoxia and a reduced DLCO. The more severe cases of HPS may also have some additional (and somewhat unusual) symptoms: platypnea (dyspnea induced by the upright position) and orthodeoxia (a decrease in PaO2 and SaO2 when changing from the supine to upright positions).

For reasons that aren’t completely clear liver disease can cause chronic vasodilation of the systemic and pulmonary vasculature. The normal diameter of the pulmonary capillaries is in the range of 8-15 microns. When dilated due to liver disease they can be as large as 100 or even 500 microns in diameter. This allows mixed-venous blood to pass through the pulmonary capillaries very quickly or even directly into the pulmonary veins, and this in turn causes arterial hypoxia.

HPS severity is usually graded according to the level of hypoxia. First, for HPS to be considered at all, an individual’s alveolar-arterial oxygen gradient (PAaO2) needs to be greater than 15 mm hg. After that HPS is graded using PaO2 (room air, sea level) as:

	PaO2 (mm Hg)
Mild	≥80
Moderate	<80, ≥60
Severe	<60, ≥50
Very Severe	<50

Interestingly CO2 retention is never seen in hepatopulmonary syndrome, and in fact since these individuals usually chronically hyperventilate, hypocapnia (PaCO2 < 35) and respiratory alkalosis are often present.

The reasons that DLCO is reduced in HPS are complex and one study or another has indicated that it is due in part to:

• Increased V/Q mismatching
• Increased intrapulmonary shunting
• Increased thickness of the alveolar-capillary membrane
• An increased distance between the alveolar membrane and the central stream of the blood flow through the dilated capillaries doesn’t allow for adequate diffusion of carbon monoxide
• A decreased pulmonary capillary transit time

Although the pulmonary capillary dilation is usually seen diffusely throughout the lung, on a local level it is often heterogeneous. This causes increased V/Q mismatching and is often exacerbated because the ability of the pulmonary vasculature to vasoconstrict due to hypoxia is blunted or even absent. This causes poorly ventilated lung units to continue to be perfused and an increase in intrapulmonary shunting.

In addition the relationship between ventilation and perfusion within the lung is affected by gravity. In the early 1960’s J. B. West proposed a three zone lung model.

In this model the upper zone of the lung has the lowest pulmonary arterial pressure and therefore the lowest perfusion. At rest it therefore has the highest ventilation/perfusion ratio. The lower zone has the highest pulmonary arterial pressure, the highest perfusion and at rest the lowest V/Q ratio. Ventilation and perfusion tend to be relatively well matched in the middle zone.

The more severe forms of HPS have a distinct positional component and this has been shown to be caused by a redistribution of blood flow towards the lower lung zones. Platypnea usually occurs in conjunction with orthodeoxia which is defined as a decrease in PaO2 by ≥5% or ≥4 mm Hg when changing from the supine to the upright position. Interestingly, although there is a significant change in PaO2 in orthodeoxia at least one study showed no significant change in DLCO from supine to upright in the same patients. Given the small number of patients (n=5) with orthodeoxia whose DLCO was studied in both the positions it’s unclear whether this applies to all patients with orthodeoxia.

A small number of studies have shown that both the membrane component of the diffusing capacity (DMCO) and the pulmonary capillary blood volume (Vcap) are reduced in HPS. Although the reduced DMCO could be taken as indication of a thickened alveolar-capillary membrane, a finding which is seconded to some extent by the fact that smooth muscle hypertrophy and fibrosis in the small pulmonary arteries have been found in individuals with HPS, a careful analysis in one study indicated that it is more likely an artifact of a reduced pulmonary capillary transit time.

Individuals with HPS usually have a significantly elevated cardiac output. This is due in part to a reduced systemic vascular resistance (SVR) and pulmonary vascular resistance (PVR) but may also in part be a response to chronic hypoxia. The elevated cardiac output however, decreases the pulmonary capillary transit time and contributes to hypoxia.

Interestingly, at least one study showed that some individuals with orthodeoxia have an elevated cardiac output while supine and a normal cardiac output while upright. The reason for this finding and whether it is a contributing factor to the orthodeoxia are unclear.

HPS primarily affects the circulatory system so unless a coexisting lung disease is present spirometry and lung volumes are usually normal. Even so a certain number of patients with liver disease will show a restrictive pattern due to ascites, pleural effusion, respiratory muscle weakness and an enlarged liver.

Hypoxemia is usually progressive in hepatopulmonary syndrome and long-term survival rates are poor. HPS used to be considered a contraindication towards liver transplantation but the finding that hypoxia often improves post-transplant has caused this to be re-thought. One study found a 23% 5-year survival rate in patients with HPS that did not receive a liver transplant and a 63% 5-year survival rate among those that did. It is interesting to note that although oxygenation often improves post-transplant, in general DLCO does not. This finding may be related to the length of time that DLCO is studied post-transplant however, since one case study showed that a 7-year post-transplant survivor had an increase in DLCO that was not present 16 months post-transplant.

In many different lung diseases a reduced DLCO is due to a ventilation-perfusion mismatch. This mismatch is often due to a maldistribution of ventilation but hepatopulmonary syndrome is an example of the maldistribution of perfusion. A reduced DLCO seen in the presence of liver disease is frequently a sign of hepatopulmonary syndrome. Transplant centers worldwide have reported that between 5% and 32% of the patients referred for liver transplantation have HPS but the actual frequency among all patients with liver disease is unknown.

Hepatopulmonary syndrome causes systemic and pulmonary vasodilation and through a variety of mechanisms, some of which remain poorly understood, cause hypoxia and a reduced DLCO. The level of hypoxia (and to some extent the reduction in DLCO) are directly related to the severity of HPS. Individuals with severe HPS also tend to have a positional component and show an increase in dyspnea and a decrease in PaO2 when moving from supine to upright and for this reason may need to be evaluated in both positions.

References:

Collisson EA, Nourmand H, Fraiman MH, Cooper CB, Bellamy PE, Farmer DG, Vierling JM, Ghobrial RM, Busuttil RW. Restrospective analysis of the results of liver transplantation for adults with severe hepatopulmonary syndrome. Liver Transpl 2002; 8: 925-931.

Degano B, Mittaine M, Guenard H, Rami J, Garcia G, Kamar N, Bureau C, Peron JM, Rostaing L, Riviere D. Nitric oxide and carbon monoxide transfer in patients with advanced liver cirrhosis. J Appl Physiol 2009; 107: 139-143.

Fallon MB, Krowka MJ, Brown RS, Trotter JF, Zacks S, Robert KE, Shah VH, Kaplowitz N, Forman L, Wille K, Kawut SM. Impact of hepatopulmonary syndrome on quality of life and survival in liver transplant candidates. Gastroentrology 135(4): 1168-1175.

Gomez FP, Martinez-Palli G, Barbera JA, Roca J, Navasa M, Rodriguez-Roisin R. Gas exchange mechanism of orthodeoxia in hepatopulmonary syndrome. Hepatology 2004; 40: 600-666.

Lima BLG, Franca AVC, Pazin-Filho A, Araujo WM, Martinez JAB, Maciel BC, Simoes MV, Terra-Filho J, Martinelli ALC. Frequence, clinical characteristics and respiratory parameters of hepatopulmonary syndrome. Mayo Clin Proc 2004; 79: 42-48.

Rodriguez-Roisin R, Krowka MJ. Is severe arterial hypoxaemia due to hepatic disease an indication for liver transplantation? A new therapeutic approach. Eur Respir J 1994; 7: 839-842.

Rodriguez-Roisin R, Krowka MJ, Herve P, Fallon MB. ERS Task Force: Pulmonary-hepatic vascular disorders. Eur Respir J 2004; 861-880.

Rodriguez-Roisin R, Krowka MJ. Hepatopulmonary syndrome – a liver-induced lung vascular disorder. N Engl J Med 2008: 358: 2378-2387.

Scarlata S, Conte ME, Cesari M, Gentilucci UV, Miglioresi L, Pedone C, Picardi A, Ricci GL, Incalzi RA. Gas exchanges and pulmonary vascular abnormalities at different stages in chronic liver disease. Liver International 2011; 31(4): 525-533.

Swanson KL, Wiesner RH, Krowka MJ. Natural history of hepatopulmonary syndrome: Impact of liver transplantation. Hepatology 2005; 41: 1122-1129.

West JB. Regional differences in gas exchange in the lung of erect man. J Appl Physiol 1962; 17(6): 893-898.

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