CHEST

Original Research
PULMONARY FUNCTION TESTING

Discordance in Spirometric Interpretations

Using Three Commonly Used Reference
Equations vs National Health and
Nutrition Examination Study III*
Jacob Collen, CPT, MC, USA; David Greenburg, CPT, MC, USA;
Aaron Holley, MAJ, MC, USA; Christopher S. King, CPT, MC, USA;
and Oleh Hnatiuk, COL, MC, USA, FCCP

Background: Spirometry plays an essential role in the diagnosis and management of pulmonary
diseases. The accurate interpretation of spirometric data depends on comparison to a reference
population to identify abnormalities in ventilatory function. National guidelines recommended
the use of the National Health and Nutrition Examination Study (NHANES) III data set as the
preferred reference population for those persons 8 to 80 years of age in the United States.
Objectives: To determine the effect of using NHANES III reference equations, compared to those
of Crapo et al (Crapo), Knudson et al (Knudson), or Morris et al (Morris), on spirometric
interpretations in non-Hispanic white patients.
Methods: We conducted a cross-sectional study of all white patients undergoing spirometry testing at
our hospital from January 2000 through May 2007. Patients were classified as normal, restricted,
obstructed, or mixed, based on the American Thoracic Society (ATS)/European Respiratory Society
(ERS) guidelines, using the Crapo, Knudson, Morris, and NHANES III prediction equations.
Differences in the classifications based on the reference data set were evaluated.
Results: At total of 8,733 subjects (62.4% male subjects) were identified, with a mean age of 53 years.
Discordance was most common when the results from prediction equations by Knudson and Morris
were compared to those of NHANES III (45.5% and 35.3%, respectively). Diagnostic recategorizations occurred less frequently when the prediction equations by Crapo were compared with those of
NHANES III (15.9%). Relative to NHANES III, the prediction equations by Knudson, Crapo, and
Morris tend to overclassify obstruction and underclassify restriction.
Conclusions: There is significant discordance between the prediction equations put forth by Crapo,
Knudson, Morris, and the NHANES III. Our data suggest that the diagnostic reclassification of many
patients undergoing pulmonary function testing will occur when ATS/ERS guidelines are implemented. Pulmonologists and other physicians interpreting spirometry need to be aware of the
presence and nature of these changes.
(CHEST 2008; 134:1009 1016)
Key words: airway; pulmonary function test; spirometry
Abbreviations: ATS American Thoracic Society; BMI body mass index; CI confidence interval; ERS European
Respiratory Society; LLN lower limit of normal; NHANES National Health and Nutrition Examination Study;
OR odds ratio; PFT pulmonary function test

linicians, researchers, and patients all benefit

C from
the accurate interpretation of spirometry.
Accurate interpretation depends on the comparison
of acquired spirometric data with appropriate reference standards. Until recently, the recommendation
regarding choice of reference equations was left up
to individual pulmonary function test (PFT) laborawww.chestjournal.org

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

tories.1 Reference equations published by Crapo et al2

(Crapo), Knudson et al3 (Knudson), and Morris et al4
(Morris) were used commonly in past years.5 In 2005,
the American Thoracic Society (ATS) and the European Respiratory Society (ERS) published new combined guidelines recommending the use of reference
values from the National Health and Nutrition ExCHEST / 134 / 5 / NOVEMBER, 2008

1009

amination Study (NHANES) III population data set7

in the United States.6 At present, it is unclear how
many laboratories follow the ATS/ERS recommendations. However, this switch has the potential to
lead clinicians and researchers to change disease
classifications (eg, normal, obstruction, and restriction) or severities (eg, mild, moderate, or severe)
without changes in actual disease status.8 This may
subsequently result in changes in diagnoses, in alterations in diagnostic evaluations and therapies, as well
as in different populations being included or excluded from new research protocols.

Materials and Methods

We conducted a cross-sectional study of all white patients who
were undergoing spirometry testing at our tertiary care center
from January 2000 through May 2007. All spirometry testing was
performed according to ATS standards (Pulmonary Function
Testing Equipment; SensorMedics; Yorba Linda, CA).9 Prior to
2005, spirometry end-of-test criteria were based on the 1994 ATS
standards.10 Measured spirometric indexes for individual patients
were interpreted in comparison to reference values that were
calculated based on demographic data using the reported equations for the respective reference equations. All patients were
reclassified using an algorithm that was similar to the ATS/ERS
recommended algorithm for spirometric interpretation (Fig 1).6
Patients whose measured FVC fell below the lower limit of the
normal range as determined using the equations by Crapo,
Knudson, Morris, or NHANES III were classified as having a
restrictive pattern. The severity of restriction was graded based
on the degree of decrement in FVC below the lower limit of
normal (LLN), in accordance with the 1986 ATS guidelines.11
Patients whose measured FEV1/FVC ratio fell below the LLN as
defined by Crapo, Knudson, or NHANES III were classified as
obstructed. The study by Morris did not include an equation for
the predicted FEV1/FVC. For Morris, we defined obstruction as
an FEV1/FVC of 0.70. Obstructive severity was based on the
degree of decrement in FEV1 determined according to the
ATS/ERS 2005 guidelines.6 Patients with both obstructive and
restrictive defects were categorized as having a mixed pattern.
This protocol was approved by the Department of Clinical
Investigation at our hospital.
*From the Departments of Internal Medicine (Drs. Collen
and Greenburg), and Pulmonary/Critical Care Medicine (Drs.
Holley, King, and Hnatiuk), Walter Reed Army Medical
Center, Washington, DC.
Presented at Combined Army-Air Force American College of
Physicians Conference November 18, 2007.
The views expressed in this article are those of the authors and do
not reflect the official policy of the Department of the Army,
Department of Defense, or the US Government.
The authors have reported to the ACCP that no significant
conflicts of interest exist with any companies/organizations whose
products or services may be discussed in this article.
Manuscript received March 6, 2008; revision accepted May 13,
2008.
Reproduction of this article is prohibited without written permission
from the American College of Chest Physicians (www.chestjournal.
org/misc/reprints.shtml).
Correspondence to: Christopher S. King, CPT, MC, USA, Walter Reed
Army Medical Center, Department of Medicine, 6900 Georgia Ave NW,
Washington, DC 20350; e-mail: christopher.king@na.amedd.army.mil
DOI: 10.1378/chest.08-0614
1010

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

FEV1/FVC% LLN
Yes

No

FVC% LLN

Yes

Normal

FVC% LLN

Yes

No

Restrictive Pattern

No

Obstructive Pattern

Mixed Defect

Figure 1. Spirometric interpretive strategy.

Due to a concern that discordance between spirometric

interpretations may be isolated to individuals who were near
thresholds for reclassification, we performed a post hoc analysis of discordance in which we calculated the number of
patients whose predicted FVC using the equations of Knudson, Crapo, and Morris differed from the predicted FVC using
the equations of the NHANES III. Significant differences in
predicted FVC between the prediction equations were defined as being greater than the ATS criteria for repeatability
(change in FVC 150 mL or 3% predicted). Significant
differences in the predicted FEV1 were also defined as a
change of 150 mL or 3% predicted.
Statistical Analysis
Relationships between continuous and categoric variables
were assessed using the Student t test and 2 test, respectively.
Agreement on the interpretation of spirometric patterns (ie,
normal, obstructive, restrictive, or mixed) between predictive
models was assessed using the statistic. Agreement on the
severity of obstruction and restriction was assessed using the
quadratic statistic. The reported p values are two sided.
Statistical significance was defined as p 0.05. All analyses were
performed using a statistical software package (STATA, version
9.2; StataCorp LP; College Station, TX). Multivariate logistic
regression was used to calculate the adjusted odds ratios for
discordance between interpretative strategies using equations by
Crapo, Knudson, and Morris in comparison against those from
the NHANES III. Covariates included in the logistic regression
models included sex, age ( 25, 25 to 34.9, 35 to 49.9, 50 to 64.9,
65 to 79.9, and 80 years), body mass index (BMI) [ 18.5, 18
to 24.9, 25 to 29.9, 30 to 40, and 40 kg/m2], being short (ie,
height in the lowest 2.5% for gender), or being tall (ie, height in
the highest 2.5% for gender).

Results
During the study period, a total of 14,390 PFTs were
performed at our facility. The majority of subjects were
white (n 8,733; 60.7%) with the remainder categorized as African American (n 4,463; 31.0%), Hispanic (n 667; 4.6%), and Asian (n 527; 3.7%). This
study was limited to exploring spirometric discordance
among whites. Demographic information on the patients included in our study is listed in Table 1. A
histogram of patient ages is provided in Figure 2.
Original Research

Table 1Patient Demographics

Variables
Male subjects (n 5,446) 62.4%
Age, yr
Height, cm
Weight, kg
BMI, kg/m2
FEV1, L
FVC, L
FEV1/FVC
Female subjects (n 3,287) 37.6%
Age, yr
Height, cm
Weight, kg
BMI, kg/m2
FEV1, L
FVC, L
FEV1/FVC

Mean (SD)

Range

51.7 (19.6)
177.7 (7.3)
87.7 (16.5)
27.8 (4.8)
3.09 (1.13)
4.29 (1.23)
70.9 (12.8)

19101
132211
41199
1563.1
0.396.86
0.799.83
23100

55.9 (19.2)
162.8 (7.0)
72.7 (18.2)
27.5 (6.7)
2.13 (0.83)
2.90 (0.93)
72.6 (12.1)

1995
127191
31175
1362
0.274.92
0.326.49
26100

.005

Density

.01

.015

Comparing spirometric classifications identified

using equations by Crapo, Knudson, and Morris to
those identified by the NHANES III, we found
multiple discordant categories (Fig 3). Three discordance patterns were the most common and likely to
have a substantial impact on patient care and research. These include patients who were classified as
normal by the equations of Crapo, Knudson, or
Morris but were reclassified as restrictive by those of
the NHANES III; patients who were classified as
obstructive by the equations of Crapo, Knudson, and
Morris but were reclassified as normal by those of
the NHANES III; and patients who were classified
as obstructive by the equations of Crapo, Knudson,
and Morris but were reclassified as restrictive by
those of the NHANES III. Concordant classification
was observed in 76.3%, 68.5%, and 59.6% of patients, respectively, when comparing classifications
made by the equations of Crapo, Morris, and Knud-

40

20

60
Age

80

Figure 2. Histogram of patient age.

www.chestjournal.org

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

100

son to those by NHANES III. The Crapo predictive

equations classified 4,096 patients as normal. Of these
patients, 9.9% (n 406) were reclassified as restrictive
by the equations of the NHANES III. Of the 2,257
test results that were classified as obstructive, 23.5%
(n 530) and 2.2% (n 50), respectively, were
reclassified by the equations of the NHANES III as
normal and restrictive. The Morris predictive equations classified 5,625 test results as normal. Of these
results, 26.1% (n 1,467) were reclassified as restrictive by the equations of the NHANES III. Of
the 2,700 test results classified as being obstructive,
13.7% (n 371) and 7.2% (n 193), respectively,
were reclassified by the equations of the NHANES
III as normal and restrictive. The Knudson predictive
equations classified 4,294 test results as normal. Of
these, 30.0% (n 1,290) were reclassified as restrictive. Of the 4,249 patients who were classified as
obstructive, 30.1% (n 1,277) and 11.5% (n 488),
respectively, were reclassified by NHANES III as
normal and restrictive (Table 2). The severity of restriction was found to be mild for all patients who were
reclassified from normal by the equations of Crapo,
Knudson, or Morris to restrictive by NHANES III.
Of the patients who were reclassified from obstructed
by the equations of Crapo, Knudson, or Morris, the vast
majority fell into the mildly obstructed categories.
Evaluation of the diagnostic classification between
the equations of NHANES III and Crapo was
excellent ( 0.77), was fair with the equations of
Morris ( 0.43), and was poor with the equations
of Knudson ( 0.32). Comparisons of measures of
obstruction and restriction severity were excellent
between the equations of Crapo, Knudson, and
Morris and the NHANES III (Table 3).
The average predicted FVCs were slightly lower
for all prediction equations relative to the FVCs
predicted by the NHANES III equations (Table 4).
The mean difference in FVC compared to those
predicted by NHANES III equations ranged from
0.7% predicted (SD, 4.7% predicted) with the Crapo
equations to 6.2% predicted (SD, 4.7% predicted)
with the Knudson equations. Meaningful differences
in predicted FVCs were observed between most
individuals regardless of the choice of prediction
equations when compared against FVCs predicted
using equations from the NHANES III (Crapo,
6,003 L [69% predicted]; Knudson, 6,708 L [77%
predicted]; and Morris, 8,262 L [95% predicted]).
The vast majority of the discordance observed across
all comparison equations occurred in patients with
clinically important differences in predicted FVCs.
These results suggest that the large degree of discordance observed was not due to individuals being near
a diagnostic classification threshold.
CHEST / 134 / 5 / NOVEMBER, 2008

1011

14,390 Subjects
4463 AfricanAmerican

Crapo

530
406
170
124
64
50
23
11
8
3

1194 Other
8,733 White

Knudson

NHANES

Obstructive Normal
Normal Restrictive
Mixed Restrictive
Obstructive Mixed
Restrictive Normal
Obstructive Restrictive
Mixed Obstructive
Normal Obstructive
Mixed Normal
Restrictive Mixed

NHANES

Morris

1467
772
371
223
193
25
19
5

1290 Normal Restrictive

1277 Obstructive Normal
897 Obstructive Mixed
488 Obstructive Restrictive
21
Mixed Obstructive

NHANES

Normal Restrictive
Obstructive Mixed
Obstructive Normal
Normal Obstructive
Obstructive Restrictive
Normal Mixed
Mixed Restrictive
Restrictive Mixed

Figure 3. Discordant classifications.

The mean difference in FEV1 (in liters) was lower

for Crapo equations than for the NHANES III
equations (0.04), while it was greater for both the
Knudson equations (0.08) and the Morris equations
(0.27) [Table 5]. A difference of 150 mL or 3%
predicted was observed in a large percentage of
patients for all studies, ranging from 49% for Crapo
to 99% for Morris. Depending on the change in
FVC, these differences may not have resulted in
discordant PFT interpretation; however, they could
potentially impact the interpretation of severity of
obstruction.
An assessment of the predictors of discordance
was undertaken (Table 6). Relative to men, women
experienced less discordance between all three prediction equation and those of the NHANES III.

Compared to those persons aged 35 to 49.9 years,

less discordance was observed in those persons 25
years old when using equations by Crapo (odds ratio
[OR], 0.65; 95% confidence interval [CI], 0.51 to
0.83) or Knudson (OR, 0.56; 95% CI, 0.47 to 0.66),
but not those by Morris. An age of 65 to 74.9 years
predicted less discordance between the equations of
Crapo and the NHANES III (OR, 0.64; 95% CI,
0.54 to 0.76), whereas increased discordance was
observed among those 65 years of age with the
equations by Knudson and Morris. Being underweight (BMI, 18.5 kg/m2) or overweight (BMI,
40 kg/m2) increased the odds of discordance
relative to being of normal weight for all prediction
equations. Short stature increased the odds of discordance with all of the prediction equations. Tall

Table 2Patterns of Discordance*

NHANES III
Study
Crapo

Morris

Knudson

Diagnosis

Normal

Restrictive

Normal
Restrictive
Obstructive
Mixed
Normal
Restrictive
Obstructive
Mixed
Normal
Restrictive
Obstructive
Mixed
Total

3,679 (89.8)
64 (4.8)
530 (23.5)
8 (0.8)
3,910 (69.5)

406 (9.9)
1,262 (95.0)
50 (2.2)
170 (16.2)
1,467 (26.1)
209 (97.7)
193 (7.2)
19 (9.8)
1,290 (30.0)
89 (100)
488 (11.5)
21 (20.8)
1,888 (21.6)

371 (13.7)
3,004 (70.0)
1,277 (30.1)
4,281 (49.0)

Obstructive

Mixed

11 (0.3)
1,553 (68.8)
23 (2.2)
223 (4.0)
1,364 (50.5)

1,587 (37.4)
1,587 (18.2)

3 (0.2)
124 (5.5)
850 (80.9)
25 (0.4)
5 (2.3)
772 (28.6)
175 (90.2)

897 (21.1)
80 (79.2)
977 (11.2)

Total
4,096 (46.9)
1,329 (15.2)
2,257 (25.8)
1,051 (12.0)
5,625 (64.4)
214 (2.5)
2,700 (30.9)
194 (2.2)
4,294 (49.2)
89 (1.0)
4,249 (48.9)
101 (1.2)
8,733

*Values are given as No. (%).

1012

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

Original Research

Table 3Agreement Among Crapo, Morris, and

Knudson, and the Severity of Obstructive and
Restrictive Defects
Study

Obstructive Severity

Restrictive Severity

Crapo
Morris
Knudson

0.92
0.91
0.86

0.93
0.93
0.88

Table 5Differences in Predicted FEV1 Between

Crapo, Morris, and Knudson, and NHANES III*
Differences

Crapo

Knudson

Morris

Predicted FEV1
L
% predicted
Observed

0.04 (0.16)
1.7 (6.1)
4,315 (49)

0.08 (0.13)
8.4% (3.0)
6,324 (72)

0.27 (0.11)
2.5% (4.6)
8,676 (99)

*Values are given as mean (SD), unless otherwise indicated.

Compared to NHANES III.
Values are given as No. (%).

stature was protective from discordance when spirometry interpretations from the equations of Crapo
were compared to those from the equations of the
NHANES III (OR, 0.37; 95% CI, 0.24 to 0.59).
Discussion
Our primary finding was the presence of significant discordance between standardized spirometric
interpretations using NHANES III data compared to
older reference equations in non-Hispanic white
patients. This particular population was chosen because the older studies predominantly enrolled
whites. Pulmonary function laboratories currently
using Knudson may experience reclassification of
nearly one half of their spirometric interpretations
when implementing the new ATS/ERS guidelines,
depending on the amount and severity of disease in
their population. Laboratories currently using Crapo
or Morris will experience reclassifications to a lesser
degree. The most common patterns of reclassification are normal to restrictive, obstructive to normal,
and obstructive to restrictive, using Knudson, Morris, or Crapo vs NHANES III, respectively.
A recent article by Sood and colleagues8 compared
prior reference spirometry sets (Kory et al,11 Crapo
et al,3 Morris et al,4 Knudson et al (1976),12 and

Table 4 Differences in Predicted FVC Between

Crapo, Morris, and Knudson, and NHANES III*
Differences
Predicted FVC
L
% predicted
Observed,
Discordant PFT
interpretations
Significant differences
in predicted FVC
No significant
differences in
predicted FVC

Crapo

Knudson

Morris

0.04 (0.16)
0.7 (4.7)
6,003 (69)

0.26 (0.15)
6.2 (4.7)
6,708 (77)

0.09 (0.14)
2.0 (3.8)
8,262 (95)

963 (69)

3,800 (96)

2,321 (75)

426 (31)

173 (4)

754 (25)

*Values are given as the mean (SD), unless otherwise indicated.

Compared to NHANES III ( 150 mL or 3% predicted).
Values are given as No. (%).
www.chestjournal.org

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

Knudson et al3) against NHANES III data in 1,106

patients who had been referred for spirometry at their
institution in central Illinois. Sood et al8 evaluated
patients for the presence or absence of obstructive and
restrictive abnormalities, and measured agreement
among prediction equations. They found that agreement was poor between Crapo and NHANES III at
identifying the presence of these defects (weighted ,
0.13). Agreement between Knudson and Morris and
data from NHANES III was similar (weighted , 0.46
and 0.44, respectively). Similar to our study, Sood et al8
reported good to excellent agreement on the severity of
obstructive and restrictive defects between Crapo,
Morris, and Knudson and NHANES III. The differences in results between their study and ours likely
represent differences in the methodologies and samples between the studies. In the present study, we
offer the benefit of looking at a significantly larger
study population with a wider geographic distribution. In addition, instead of categorizing patients in
two domains on the presence or absence of abnormalities, we classified patients as having a normal,
obstructive, restrictive, or mixed pattern. We feel
that this distinction is of more utility for clinicians
who are interpreting these data. However, the overall conclusion of their study and ours is that changes
in the interpretations of PFTs after converting from
older reference equations to NHANES III will be
common.
Both the number of reclassifications and the
change in interpretation patterns are clinically important. Although it is difficult to predict the exact
number of patients affected, it is undoubtedly large.
In patients who are already being followed up with
serial spirometry, original diagnoses may be reconsidered, leading to additional diagnostic testing and,
potentially, to altered courses of therapy. Unless all
laboratories are currently using NHANES III prediction equations, individuals undergoing initial spirometric evaluations may experience unnecessary
referrals or will not be referred when needed.
Screening spirometry has been advocated for
smokers over the age of 45 years and patients with
respiratory symptoms.13 In addition, data on the
CHEST / 134 / 5 / NOVEMBER, 2008

1013

Table 6 Predictors of Discordance*

Crapo

Knudson

Morris

Predictors

OR

95% CI

OR

95% CI

OR

95% CI

Female gender
Age, yr
25
2534.9
3549.9
5064.9
6579.9
80
BMI, kg/m2
18.5
18.524.9
2529.9
3039.9
40
Short
Tall
Area ROC

0.72

(0.630.81)

0.49

(0.450.54)

0.88

(0.800.97)

0.65
0.88
1.00
0.99
0.64
0.98

(0.510.83)
(0.711.08)

0.56
0.80
1.00
1.49
1.78
1.24

(0.470.66)
(0.680.94)

1.04
1.15
1.00
1.60
2.67
2.26

(0.871.26)
(0.971.37)

0.73
1.00
0.94
1.13
1.57
1.67
0.37
0.60

(0.441.23)

1.84
1.00
0.96
1.39
3.51
1.32
1.10
0.65

(1.332.57)

1.94
1.00
0.92
1.46
2.81
1.32
1.17
0.63

(1.402.70)

(0.841.16)
(0.540.76)
(0.791.22)

(0.811.08)
(0.981.31)
(1.152.13)
(1.262.20)
(0.240.59)

(1.311.69)
(1.572.03)
(1.041.47)

(0.861.06)
(1.251.56)
(2.664.62)
(1.041.68)
(0.881.40)

(1.391.83)
(2.333.05)
(1.892.69)

(0.821.03)
(1.301.63)
(2.183.64)
(1.041.67)
(0.921.59)

*Short lowest 2.5% in height for gender; Tall top 2.5% in height for gender; ROC receiver operating characteristic.

utility of office-based testing in the primary care

setting continues to accumulate.14,15 With increases
in use by providers who are not specifically trained in
interpretation, unexpected changes in reference standards could have a large impact on missed diagnoses,
and inappropriate referrals, tests, and treatments.
From the researchers perspective, the widespread
implementation of the ATS/ERS guidelines using
percent predicted as a selection criterion will result
in different patient populations being studied now
compared with those in previous studies. The results
from populations defined by the older criteria may
not be applicable to current patient populations.
Additionally, changing the guidelines may have a
broader public impact. Because spirometric measures presented as percent predicted are used in a
variety of settings, changes will affect disability ratings,16 lung transplantation referrals,17 and preoperative risk assessments,18 as well as many other areas.
While the reasons for the discordance between
reference equations cannot be defined with complete certainty, we propose several potential reasons.
One potential reason may be differences between
the reference populations studied. The populations
in Crapo and Morris consisted primarily of members
of The Church of Jesus Christ of Latter Day Saints
(Mormons). These individuals may have had less
exposure to air pollution and cigarette smoke than
the population studied in NHANES III. Additionally, this white population predominately consisted
of individuals of Northern and Middle European
descent but may differ biologically from white Americans as a whole. If individuals studied by NHANES
III had more smoke and pollution exposure resulting
1014

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

in mild obstructive lung disease, this could potentially explain the large number of individuals reclassified from obstructive to normal by NHANES III in
comparison to all three older reference equations.
Technical differences may have contributed to discordance among equations as well. Morris utilized a
volume threshold rather than the back extrapolation
that is currently recommended by the ATS in determining the start of a test.9 This may result in a falsely
low FEV1 and an increased number of patients being
reclassified from normal to obstructive. Knudson utilized a pneumotachometer, which may have terminated maneuvers early, resulting in a decreased FVC.8
The Crapo reference equation utilized the largest FVC
and FEV1 sum rather than the largest value from
separate efforts, as recommended by the ATS. This
may result in a 50-mL reduction in the FVC.8 Finally,
NHANES III required a five-maneuver minimum and
provided computer-generated feedback to the technician that a 1-s plateau had been reached.7 All of the
above reasons may explain the large number of patients
reclassified from normal to restrictive who were observed in our study.
We found several demographic factors that make
discordance more likely. Age 50 years was associated with an increased likelihood of discordance
between the Knudson and Morris equations and
NHANES III data. Obesity was associated with an
increased likelihood of discordance in all three older
reference equations when compared to NHANES III.
Jones and Nzekwu19 demonstrated a linear decrease in
vital capacity with increasing BMI. It is likely that
obese patients who are classified as normal by older
reference equations for the reasons listed above fell
Original Research

below the LLN when compared to those classified by

the NHANES III reference equation. Finally, patients
in the lowest 2.5% of height had an increased risk of
discordance for all reference populations compared to
NHANES III. Female sex was protective against discordance for all reference equations.
Although our study was not designed to identify
the optimal reference equation for this population,
we believe the NHANES III equations are the best
available to date. The NHANES III equations are
based on a sample size that was 10-fold greater than
those in any of the prior studies and included a
heterogeneous population that was enrolled from
1988 to 1994. These equations are less subject to
cohort effect in comparison to those derived by the
Crapo, Morris, and Knudson studies, which enrolled
patients in earlier decades (ie, the 1960s and 1970s).
Our study has several limitations. We purposefully
limited our study to non-Hispanic white patients. The
affect of race on the discordance between various
reference equations will require further study. Additionally, our study was performed at a military institution. The patient population included active-duty soldiers, their dependents, and retirees. It is possible that
this population may not be generalizable to the population of a civilian hospital. Another limitation of our
study was the use of a fixed FEV1/FVC ratio of 0.7
for the determination of obstruction for the Morris
equation. We elected to use this technique as no
formula for LLN was listed in the original Morris
article. We also felt that this was consistent with the
interpretive strategy utilized by the majority of PFT
laboratories utilizing the Morris reference equation.
Studies20 have shown a tendency to overcall obstruction in extremes of age and height with this technique,
so it is likely that using a fixed ratio overestimated the
number of patients who were classified with obstruction by the Morris equation. Only 17.9% of the discordance observed between NHANES III and Morris is
explained by reclassification from obstructive to normal
or restrictive; so, while the use of a fixed FEV1/FVC
ratio may have exaggerated the rate of discordance, it
was not the primary cause of discordance among these
reference equations. It should also be noted that the
thresholds utilized for the interpretation of ventilatory
pattern are based on sharp, predetermined LLNs. If
patients fall just below the LLN, it is possible that
test-retest variability may result in interpretive changes
in these borderline patients. This may eliminate
discordance in an individual patient; however, it does
not invalidate the overall finding of discordance among
reference equations on a population level. Poor test
quality may result in the misclassification of ventilatory
patterns but should not affect a population-level comparison of reference equations. It may lead to a skewed
subgroup analysis of demographic predictors of discorwww.chestjournal.org

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

dance if a disproportionate number of tests of poor

quality are performed for one demographic group. We
did not exclude patients based on the quality of spirometry, which is a limitation of our study.
In summary, physicians interpreting spirometry
findings need to be familiar with the changes that
will occur in interpretation, as outlined in our study,
when adopting the new ATS/ERS guidelines. Physicians observing patients longitudinally should be
aware that changes in interpretation may be due to
changes in the reference standard even in the setting
of equivalent spirometric measures. In this setting,
the absolute values should be compared across tests
instead of relying solely on the percent predicted
values.21 Additionally, spirometric equipment manufacturers need to ensure that NHANES III reference equations and LLNs are part of the software
package provided with their products. Last, pulmonary function laboratories and their medical directors need to ensure that all physicians relying on
their services are aware of the potential impact of
these changes.
References
1 American Thoracic Society. Lung function testing: selection
of reference values and interpretative strategies. Am Rev
Respir Dis 1991; 144:12021218
2 Crapo RO, Morris AH, Gardner RM. Reference spirometric
values using techniques and equipment that meet ATS
recommendations. Am Rev Respir Dis 1981; 123:659 664
3 Knudson RJ, Lebowitz MD, Holberg CJ, et al. Changes in the
normal maximal expiratory flow-volume curve with growth
and aging. Am Rev Respir Dis 1983; 127:725734
4 Morris JF, Koski A, Johnson LC. Spirometric standards for
healthy nonsmoking adults. Am Rev Respir Dis 1971; 103:
57 67
5 Ghio AJ, Crapo RO, Elliott CG. Reference equations used to
predict pulmonary function: survey at institutions with respiratory disease training programs in the United States and
Canada. Chest 1990; 97:400 403
6 Pellegrino R, Viegi G, Brusasco V, et al. Interpretative
strategies for lung function tests. Eur Respir J 2005;
26:948 968
7 Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general US population. Am J
Respir Crit Care Med 1999; 159:179 187
8 Sood A, Dawson BK, Henkle JQ, et al. Effect of change of
reference standard to NHANES III on interpretation of
spirometric abnormality. Int J Chron Obstruct Pulmon Dis
2007; 2:361367
9 Miller MR, Hankinson J, Brusasco V, et al. Standardisation of
spirometry. Eur Respir J 2005; 26:319 338
10 American Thoracic Society. Standardization of spirometry, 1994
update. Am J Respir Crit Care Med 1995; 152:11071136
11 Kory RC, Callahan R, Boren HG, et al. The Veterans
Administration-Army cooperative study of pulmonary function: 1. Clinical spirometry in normal men. Am J Med 1961;
30:243258
12 Knudson RJ, Slatin RC, Lebowitz MD, et al. The maximal
expiratory flow-volume curve: normal standards, variability,
and effects of age. Am Rev Respir Dis 1976; 113:587 600
CHEST / 134 / 5 / NOVEMBER, 2008

1015

13 Ferguson GT, Enright PL, Buist AS, et al. Office spirometry

for lung health assessment in adults: a consensus statement
from the National Lung Health Education Program. Chest
2000; 117:1146 1161
14 Dales RE, Vandemheen DL, Clinch J, et al. Spirometry in the
primary care setting: influence on clinical diagnosis and management of airflow obstruction. Chest 2005; 128:24432447
15 Yawn BP, Enright PL, Lemanske RF Jr, et al. Spirometry can
be done in family physicians offices and alters clinical
decisions in management of asthma and COPD. Chest 2007;
132:11621168
16 American Thoracic Society. Evaluation of impairment/disability secondary to respiratory disorders. Am Rev Respir Dis
1986; 133:12051209

1016

Downloaded From: http://journal.publications.chestnet.org/ on 01/11/2015

17 Egan TM, Murray S, Bustami RT, et al. Development of the

new lung allocation system in the United States. Am J
Transplant 2006; 6:12121227
18 Beckles MA, Spiro SG, Colice GL, et al. The physiologic
evaluation of patients with lung cancer being considered for
resectional surgery. Chest 2003; 123(suppl):105S114S
19 Jones RL, Nzekwu MM. The effects of body mass index on
lung volumes. Chest 2006; 130:827 833
20 Roberts SD, Farber MO, Knox KS, et al. FEV1/FVC ratio of
70% misclassifies patients with obstruction at the extremes of
age. Chest 2006; 130:200 206
21 Cooper BG. Reference values in lung function testing: all for
one and one for all? Int J Chron Obstruct Pulmon Dis 2007;
2:189 190

Original Research