Differences in Plethysmographic Lung Volumes

Differences in Plethysmographic Lung Volumes· Effects of Linked vs Unlinked Spirometry James H. Williams Jr., M.D., F.C.C.P.; and Harold Z. Bencowitz, M.D., F.C.C.~

Determination of absolute lung volumes in patients is most reliable when measured with body plethysmography. Many laboratories use data obtained with a spirometer not directly linked to the plethysmograph to calculate total lung capacity (TLC) and residual volume (RV) from thoracic gas volume (Vtg) measured at functional residual capacity (FRC) in the plethysmograph. The reliability of these calculations depends on the stability of FRC between these separate devices. We examined the differences in TLC and RV values calculated with linked and unlinked spirometers in 220 patients and found them statistically signi6cant (p<0.05).

Additionally, differences exceeding the 95 percent confidence intervals for repeated, linked determinations occurred in more than 5 percent of patients. The large-volume differences in TLC were often associated with differences in expiratory reserve volume (ERV) in the opposite direction, suggesting a shift in FRC. However, clinical diagnoses were infrequently (41220) altered by these differences, and recognition of the shift in FRC should further reduce this error. Therefore, the unlinked method appears acceptable. (Chest 1989; 95:117-23)

Measurements oflung volumes are frequently used in clinical medicine for evaluation of patients \\'ith lung diseases. Restrictive lung diseases may be associated with a reduction in total lung capacity (TLC), and the degree of restriction may be judged by the TLC.I-4 Although with usual criteria a relatively insensitive indicator of early interstitial lung disease, 1.2.~ the TLC may suggest the superimposition of interstitial diseases on chronic obstructive pulmonary disease (COPO), while either interstitial disease or advanced COPO may be associated with a decreased vital capacity (VC) and a decreased diffusion capacity, advanced COPD alone would lead to an increased rather than a decreased or normal TLC. 2.4.22 Although absolute lung volumes can be measured by either gas dilution techniques or body plethysmography, the former frequently underestimates lung volumes in patients with poorly communicating airspaces (COPO) when compared with body plethysmography. 2-4 Therefore, body plethysmography is generally accepted as the most reliable means of measuring of lung volumes. 2-4 Determination of lung volumes is usually based on measurement of thoracic gas volume (Vtg) at FRC, with TLC and residual volume (RV) then calculated from data obtained by spirometric study. This reflects difficulty in performing the panting maneuvers required for body plethysmography at TLC and R~4 As diagrammed in Figure 1, TLC may be calculated from FRC either by adding the inspiratory capacity (IC) or

by first subtracting the expiratory reserve volume (ERV) and then adding the inspiratory vital capacity (IVe). Similarly, RV may be derived from FRC by either subtracting ERV or by first adding IC and then subtracting the expiratory vital capacity (EVC). Either approach is believed acceptable,4 supported by demonstration that IVC and EVC do not significantly differ. 6 Theoretica1l~ the calculated values of TLC and RV would be most accurate if the spirometric values were obtained immediately after determination of FRC, with a spirometer directly linked to the exhalation port of the body plethysmograph. 9 •12 However, because FRC determined in the plethysmograph is believed to vary little, 2-4.8 many clinical laboratories utilize spirometric data obtained on a separate device to calculate TLC and R~12 This latter, "unlinked" approach has a number of technical advantages regarding ease of performance and costs, as will be discussed, but assumes that FRC is stable. Potential errors associated with this "unlinked" method, related to a shift in FRC, have been suggested by others. 7 •9 However, the differences in values obtained using a separate spirometer has not been examined in a large study, to our knowledge. We compared prospectively in 220 patients the determination of RV and TLC by these two approaches: spirometry was performed during plethysmography both with a "linked" spirometer and separately with an "unlinked" spirometer. Differences were compared with the variability of repeated, linked measures, and the effects of a shift in FRC were explored.

*From the Departments of Medicine, University of California, Irvine and San Diego. ~1anuscript received Au~ust 10, 1987; revision accepted May 10, 19RR. Reprint requests: Dr. Williams, Pubnonary and Critical Care, DC In'ine .\ledical Center, OraT1f!.e, CA 92668

MATERIAL AND METHODS

Our subjects were 220 patients referred to the San Diego VA CHEST I 95 I 1 I JANUAR~

1989

117

S P I ROM E T E R

BOX

FRC

---s

A.

S P I ROM E T E R

BOX

FRC

B.

U N LIN KED

LIN KED

Hospital for routine pulmonary function testing. Subjects were excluded from this study if they were unable to perform the maneuvers with the spirometer or body plethysmograph. No subject was included more than once (if tests were repeated). The subjects' mean age was 56 (range 17 to 84) and 216 were males. Subjects were first asked to undergo spirometric evaluation using a 13.5-L wate~seal spirometer (WE Collins). The maneuver began with normal tidal breathing to identify a stable FRC, followed by slow exhalation to R~ inhalation to TLC, and finally forced exhalation (Fig lA). The ERV and IVC were measured according to standard techniques, with three acceptable trials of this maneuver used. 10 The final value of ERV used represented the mean of the three trials, and that oflVC represented the maximum of the three trials. 10 The subject was then placed in a constant-volume body plethysmograph (WE Collins). TIdal breathing was observed to identify a stable FRC, and Vtg was then determined with approximately 1 to 3 Hz (cyclesls) panting against a closed shutter as previously described. 5 •8 •11 Following each measurement, a pneumatic valve was opened to connect the subject directly to another 13.5-L wate~seal spirometer (WE Collins). The subject was instructed to inhale to

BOX

FIGURE 1. Approach. (Above): the maneuvers utilized to calculate RV and TLC, from plethysmographically-determined (box) FRC. (A) Unlinked method uses separate spirometer. (B) Linked method uses spirometer attached directly to exhalation valve of plethysmograph (box). Abbreviations and calculations in text. Of note, ERV and RV represent mean values in both methods. Maximal value of IVC is used to calculated unlinked TLC, while TLC determined by linked method was examined both as mean and maximal value.

TLC (thus measuring IC) and then exhale slowly to RV (measuring expiratory vital capacity; EVC) (Fig IB). Three trials were attempted in the study, with at least two acceptable efforts required for inclusion, based on patient ability to cooperate as subjectively assessed by the technician. TLC and RV were defined as "unlinked" (Fig lA) when they were calculated from the plethysmographic FRC (mean Vtg) using data obtained on the separate spirometer (mean ERV and largest IVC). This approach is normally used in our laboratories and has been supported by American and European standardization projects. 3.4 The TLC and RV were defined as "linked" (Fig IB) when calculated from direct determinations of TLC and R~ measured with the spirometer directly in-line with the plethysmograph. The initial maneuver following panting at FRC was an inspiratory capacity (IC) to measure TLC: An inspiratory maneuver was preferred by subjects in preliminary experiments, and is consistent with several previous reports of linked determinations of TLC. 7.9.12 The mean and largest value of TLC by the linked method were identified, the latter being potentially more analogous to the unlinked TLC as discussed below. The RV was then determined by

S P I ROM E T E R

BOX

S P I ROM E T E R

TLC

FIGURE 2. FRC shift. (Above): representation of the effect of shift in FRC from plethysmograph (box) to unlinked spirometer on calculations of ER~ R~ and TLC. l£ft, RV and TLC are held constant relative to position of FRC in plethysmograph (box), while FRC at separate spirometer is shifted up (.) or down (.). Right, the shifted FRC values at separate spirometer are aligned with box value, as would be done in calculating RV and TLC. Values ofTLC and ERV derived following this shift in FRC would vary from "real" values in opposite directions. 118

FRC

--i

•

F RC

S H 1FT

FRC~

CAL C U L A T ION S Differences in Plethysmographic Lung Volumes (Williams, Bencowitz)

a slow expiratory vital capacity (EVC) from TLC. \\;th RV representing the mean of three efforts. The reproducibility of the linked volume measurements \\'as examined in two ways, First, the 95 percent confidence interval of repeated measurements was approximated by the SO (x 1.96) of the distribution of differences between mean and individual determinations for each subject. 11 Second, a volume-nomlalized variance was calculated (the coefficient of variation, or CV), represented hy the square root ofthe individual variances ofrepeated measurements from their mean, multiplied by 100 (for percent). and divided by the mean value for each individual. This relationship is as follo\\'s: CV = j[(Xl-xl~(lOO)/(n)(x.)] In this population, n represents the total number of measurements (n = 639), ~ represents one determination in a Jtiven individual. and i, represents the mean value for that individual. This commonly of variahility is employed convention assumes that the ma~itllde related to the magnitude ofthe volume bein~ tested, 11 an assumption which is examined be10\\'. Each patient was evaluated by a sin~le technician. hut values were not calculated until after testin~ \\'as completed. Suhjects were classified as having obstructive airways disease if the FEV/FVC \\'as below the 95 percent confidence interval of a normal population ,.and the criteria for restriction were not met. They \\'ere classified as having restriction if the TLC was less than 80 percent of predicted" s If neither criterion \\'as met. the subjects \\'ere considered nonnal, If both were met, they \\'ere classified as heinJ! both obstructed and restricted, deemed "mixed disease'" Differences between the values obtained via the t\\·o methods were evaluated by paired t tests, acceptin~ statistical significance where p<0.05. The three values of TT.C (includin~ the maximal linked value) were first analyzed for variance and covariance of repeated measures. To normalize for volume, differences \\'ere similarly compared as a percentage of the value ohtained \"ith the unlinked method, analogous to the CV described above. The values obtained were also grouped by disease category, as defined by the unlinked method, and woup differences compared hy analysis of variance. While this approach most accurately identifies systenlatk differences bern'een two methods,1'3 it does not reflect the clinical significance of these differences. lfi Therefore. we also exanlined the frequency with which differences ex~eded the 95 percent confidence intervals of repeated mea~urements \\'ith the linked method. The frequency with which the values obtained by the linked method would have altered the cla-;sification of disease defined hy the unlinked method is also reported. As an indication that the FRC represented during unlinked spirometry was not reliably the same as that measured in the box. we examined the relationship between differences (unlinked-linked) in ERV and TLC, We reasoned that the pattern of the relationship

would su~est whether differences reflected an altered effort or a shift in FRC. This may be more clearly understood hy examinin~ Figure 2, and it \\'ill be discussed a~ain be I0\\'. With the linked method, ERV represented the difference bern'een the mean Vt~ at FRC and the mean R~ \\,hich should he comparahle to the mean ERV determined at unlinked spirometry. Unless otheT\\;se stated. all reported values represent means (± SD). The statistical analyses \\'ere performed \\'ith the assistance of a statistical soft\\'are package (R~f DP) t1tilizinJ,! several pro~ams (lD. 3D, 5D. 60. 70. IR. 2V). It should he recalled that, when multiple analyses art~ made. the relative statistical siJtnificance of "p<0.05" for each individual analysis becomes pro~ressi\'ely Io\\'er. Therefore. t values are ¢ven in nlost circunlstances for direct evaluation. RESULTS

Linked spirometry was more difficult to perform, requiring more time to explain to subjects, and associated with less consistent ahility or willin~ess of subjects to perform, as informally assessed by the technicians. Nevertheless, 199 of the 220 patients included in the study \\'ere able to complete all three trials adequately. Of21 \\'ho reportedly would or could not complete a third linked maneuver, reportedly because of fatigue or anxiety, all were classified as havin~ obstructive airn'ays disease by the unlinked method, and their linked data are hased on t\\,O trials. Based on unlinked data, the 220 patients studied included 61 normal, 135 obstructed, 17 restricted, and seven mixed-disease patients. The mean value of Vtg measured at FRC was 4.402 I~ (± 1.491, ran~e 1.980 to 10.780), \\,hile the mean, unlinked determination of TLC \\'as 7.114 L (± 1,.517, range 3,514 to 11.607) and of RV \\'as 3.128 L ( ± 1.307, range 0.775 to R.072). The 95 percent confidence inten'a) for FRC determinations \vas 402 mL \vith a coefficient of variation (CV) of 4.5 percent. Similar analysis of linked deternlinations of RV and TI."C revealed the 95 percent confidence intervals to he 629 ml and 517 ml, corresponding to CVs of 11.0 percent and 4.0 percent, respectively. When we examined the relationship beh\'een deviation from the mean of individual values (x1-xJ and the mean values (xJ, the correlations were

Table 1- Differences in Lung Volumes· R~

Unlinked-Linked - J ' - _ _ _ ----.

TLC, Unlinked-Linked m .... n

TLC:. Unlinked-Linked m ..,

___- - - J '

Classification*

No.

ml

% Linked

III I

% Linked

nll

% Linked

Normal Obstructed Restricted Both Total

61 135 17 7 220

155( ±392) 62( ± 391) 76( ±366) 40( ± 276) 87(±386)t

4.49(:t 17.0) 1. RO( :t II. 1) 1.71(:t 33.4) 1.42(:t 12.2) 2.5.1(:t 15.9)+

262(7404) 209(:t 312) 230(:t 210) 2.34( :t 175) 275{ :t 329)t

3.57("!: :5.40) 3.7l("!: 4.01) 4.54(:t 3.(0) 4.91(:t 3.02) 3. 7R( ~ 4.44)t

72(:t 374) fil( 7367) 5S(:t 25.5) 44(:t 310) 6.'3( :t 3.50)§

0.00(±5.21) 0.76( ± 4.03) O.O6( ± 4.00) 0.72( ± 7.12) 0.77( ± 5.fX»+

*Means (± SO) of differences between calculated values of RV and TLC usinJt the linked and unlinked spirometers. The ahsolute volume differences are reported in milliliters and the volume-normalized values as perc-ent of the unlinked value. Disease cla~sjfications are hased on unlinked data, as described in the text. tp
119

Table 2 - Frequency of lArge Volume Differences· Classification

---Nonnal ()hstructed Restricted Both llJtal

R~

No.

V nlinked-Linked

TLC. Unlinked- Linked melln

TLC, Unlinked-Linked mu

fil

6 (10%) 9(7%) 1 (6%) 0 16(7%)

12 (20%) 31 (23%) 1 (6%) 0 44 (20%)

9(15%) 16 (12%) 1 (6%)

1.'3" 17 7

220

-------------

0 26 (12%)

group' art" the sanH' as dpS{Tibed in l:'lble I. The frequency of differences which exceeded the 95% confidence intervals of repeated IneaSllres of tht·s{· VohtnlPs hy th~ link(~d method are displayed. Numbers of patients with restricted and mixed disease are too snlall for ('oJnparison.

·Th~

poor for all values, includin~ FRC (r = - 0.005, p>0.80), TI.JC (r== -O.OOR, p>0.80), and RV (r = - 0.029, p>0.4.5). The mean difference (unlinked-linked ± SO) in RV hetween the h\'o nlethods was only 89 (± 387) (ml) (l:'lhle I), hut this diff{~rt~n('e was statistically si~ificant (t=.1.41, pO.70). IJifferences in values of TI.JC (Tahle 1) tended to he somewhat larger, with a mpan difference 275 ml ( ± 329), which \\'as also statistically si~nificant (t = 12.40, pO.70) vary by clinical dia~nosis. The maxinlal Tl~C value deternlined by the linked method also differed from the unlinked value (Table 1). The variance and covariance ofthese three repeated measures ofTI..JC in each individual, were Significantly different (pO.90). Differences between linked and unlinked values exceeded the 9.5 percent interval (629 ml) in only 7.3 Table 3 - Functional Classification of Subjects· Classification Normal Obstructed Restricted Both

TLC, Unlinked

TLC. Linked m..an

TLC. Linked mu

61

61 132 17 10

133 16

135 17 7

61

10

*The effects of differences in values determined by unlinked and linked methods of calculatin~ TLC are displayed. Spirometry was used to identify obstructed physiology. and values of TLC as deternlined by each method were used to determine restriction. 120

percent of RV determinations (Table 2). However, differences exceeded the respective 95 percent confidence interval in 20.0 percent of mean TLC determinations and 11.8 percent of maximal TLC determinations by the linked method (Table 2). There was a tendency for these large differences to be less cOlumon among patients with restrictive lung diseases, although the nunlhers of these patients were relatively small for reliable conlparison. The frequency with which the diagnosis would be altered by these differences in measured TLC is demonstrated in Table 3. Three patients classified as simply obstructed by the unlinked method had nlixed disease by the linked method, using either mean or maximal values. One additional patient classified as restricted by the unlinked method was reclassified as obstructed by the maximal, linked value ofTLC. None of the three patients reclassified as mixed disease was obese, while the patient shifted from restricted to obstructed by a maximal, linked TLC \\'as obese (103.5 kg and 174 cm tall). We also examined the differences in vital capacity (VC) measured with the unlinked and linked spirometer. The uunlinked" VC represented a maximal inspiratory effort (which had been used to calculate TLC), while "linked" VC represented the maximal expiratory maneuver (measured immediately following a direct determination of TLC). The mean difference in VC \\'as only 46 ml ( ± 325), which just reached statistical significance (t = 2.10, p<0.05). When differences were normalized to unlinked VC, the mean difference was again snlall (0.8 percent), with a large SD (± 5.0), but differences which were again statistically significant (t = 2.28, p<0.05). As with other volume measurements, differences did not vary significantly by disease classification (p>O.40). When the differences in VC were plotted against the differences in TLC (using the maximal linked value), a relatively weak positive correlation was found (r = 0.373), which was statistically significant (p
1 1

111 1 1 21 1 1 - - - - - - - - -1221121121 1 1 2 11 1 1 2312111 13 224845 1222 ~3 1 J 54",..

11 ---------------

1- - - - - - - - - - - 1 11 21 11 .~ .. ~~ 1

- - - -- - - -

A"' • •

2 21 12321

1 1 1211 1 112 41 1 1·---1---------------

3223~1414

2

1+

1

1

-1

'--_....a...._ _-'--_...-.lL-.-_~ -2 -1

_ _......_

ERV

_ _ _ I L . . . . _ _......._ _

0

D IFF ERE NeE

1

(unlinked - linked)

linked TLC and unlinked TLC (r = - 0.550, p
The data demonstrate that mean differences in plethysmographic lung volunles determined with linked vs unlinked spirometry are relatively snlall' but statistically significant. The ranges of these differences were relatively large, exceeding the 95 percent confidence intervals of repeated, linked, measurements in more than 5 percent of patients. However, clinical diagnoses were altered by these differences in only four of 220 patients studied. As noted above, the practice of using a spirometer which is not directly linked to the plethysnlograph to determine TLC and RV has been supported by standardization reports. 2 -4 This approach presumes that FRC is relatively stable, as reported by DuBois et alB when initially describing the constant volume plethysmograph. They examined three normal subjects repetitivel}; and found that 2 SD from the nlean for each patient represented 181, 161, and 159 ml, corresponding to 4.8, 5.6, and 3.7 percent of the mean values, respectively. Viljanen et aP7 more recently reported

A._..._~

_ _

~

FIGL'RE 3. Relationship ofdiflerences in EMV and 1l1axilnal TLe. (Abute): scatter-plot of data ubtained fnllll 220 patients. N uluerals represent nUIl.ber of values in that re~ion. Dotted lines, 95 percent <..'unfidence bands for repeated Ineasures of TLC bv linked rnethud. A statistit'allv significant (p
the coefficient of variation of 50 duplicate Ineasurenlents of FRC to be 6.7 percent. The National Institutes of Health Standardization Project' exalnined duplicate nleasurenlents in 100 patients, with a distribution ofdiseases sinlilar to our patients; they reported a coefficient of variation of 4.4 percent, The variability in our 220 patients was siIuilar: the coefficient of variation was 4.5 percent, and the 95 percent confidence interval (1.96 SO) represented 402 lUI. In support of the unlinked approach, the variability of FRC was less than that of R\' and TLC, \\'hich denlonstrated 95 percent confidence intervals of ± 629 and 517 fill, respectively. The coefficients of variation for RV and TLC were 11.0 percent and 4.0 percent, respectively, which Blight argue that TLC was Blost reproducible. However, this cunvention of nornlalizing the variability of lung volulne Ineasurenlents to absolute volurne appears unfounded. When \\'l) exaluilled the relationships bet\\'eell nlean values and the deviation fronl nlean values in individuals, \\'e found statistically insignificant correlations, Additionally, the distribution of individuals' differences frolll their individual Ineans took a two-tailed (two-sided) rather than a one-tailed (one-sided) distribution for all three volunles. A one-tailed distribution would be anticipated when approxinlating a nlaxinlal (TLC) or nlinirnal (RV) value, while a two-tailed distribution would be expected for FRC. One explanation is that the inherent errors in determining FRC frorn the slope on the oscilloscope nlay generally exceed the variability of patient efforts. These data support utilizing absolute volurnes rather than percent predicted to deternline differences in lung volunle measurements, an approach suggested by other authors. 21 CHEST I 95 I 1 I JANUAR'Y, 1989

121

The relative reproducibility of .FRC determinations within the plethysnlograph \vould not preclude a shift in FRC outside the box at a separate spironleter. The coefficients of variation of RV and TLC obtained in the current study agree \vith previous reports,:3.91i.H~ of snlaller salnple size, supporting their accuracy. Therefore, our delnonstration that differences between linked and unlinked values exceeded the 95 percent confidence lilnits for repeated nleasures nlore often than expected (>5 percent) suggests that factors other than those norlnally encountered with repetitive lueasurelnents \\'ere involved. Differences in effort could affect the deterlnination of RV and TLC. Along this line, the mean difference \vas reduced \vhen the unlinked TLC \vas compared with the Inaxinlal, rather than nlean, linked TLC. This likely reflects the use of the (nlaximal) vital capacity on the separate spironleter for calculation of the unlinked TLC. The luaxiInal ves offer a crude indicator of effort at each apparatus, and the mean value ofVC \\'as greater \\,hen deternlined \vith the unlinked spirometer. However, the respective VC values do not contribute directly to differences in calculated RV or TLC, as can be appreciated in Figure 1. Additionally, \\'hile a greater effort could explain a greater TLC \\'ith the unlinked nlethod, it \\'ould not explain the higher RV Nevertheless, the \\'eak (r2:=: 0,139) but statistically significant positive correlation beh\'een differences in VC and differences in TLC suggests that differences in eHurt Inay have played sOlne role in the differences in TLC detertuinations. In contrast, differenccs in ER\i lueasured by thc two rnethods correlated negath'ely and 1l10rC strongly (r2= 0.303) \\'ith diflerell<.:es in TLC. This negative correlation cannot be readily explained on the basis of differences ill effort. Increased patient effort \\'ould be associated \\'ith a positi\'e correlation ifboth inspiratory and expiratory efl()rts \vere affected silnilarly, or \\'ith near-zero correlation if eflurt affected inspiratory and expiratory eHorts independcntly. Instead, this suggests that the FRC Ineasured in the plethyslllograph differed (shifted) frolll FRC at the separate spirollleter. The potential effect of a shift in ~~RC can be appreciated in Figure 2. A factitiously snlall ERV and large TLC \\'ould be calculated if FRC were shifted "down" on the separate spirolneter, and the opposite would occur with a shift "up" in FRC. The majority (86 percent) of diflerences in TLC exceeding the 95 percent confidence interval of repeated, linked measures, ,,'ere associated \\'ith differences in ERV in the opposite direction. These findings suggest tl.at systetnatic errors in TLC measurelnents in an unlinked systenl nlay be suspected when the deviation of TLC frolll predicted is associated with a deviation of ERV in the opposite direction. Unfortunately, a siInilar analysis of R\,' 122

cannot be made, because ERV directly determines RV (Fig 1). Therefore, these values would be expected to vary inversely, ,,'hether due to a reduced expiratory effort or to a shift in FRC. However, RV and TLC both tended to be higher in the unlinked system. This parallel trend \\'ould be expected with a shift in FRC. These data tend to support in concept the report by Reinert and Trendelenburg in 1971. 7 They examined two patients, one normal and the other with COPD. Using an integrated pneumotachograph linked to the exhalation port of the body plethysmograph, they found that FRC shifted by 300 and 500 ml in these two subjects, respectively. They suggested that accurate deterlllination ofTLC and RV would require use of a linked systelll. However, they did not demonstrate the effect of this shift in FRC on TLC and RV directly, and their limited data did not explore the clinical impact of these differences. Previous investigators have used an electronically integrated, pneulTIotachographic signal to determine spirolnetric volullles directly in the plethysmograph.7.9.12.1i.19 We initially examined an integrated flo\\' signal as \\'ell but found electronic drift at TLC and RV halllpered the accuracy ofour results. Additionally, the response characteristics of flow devices are not al\\'ays linear,20 potentially linliting the reliability of volullles deternlined by such systems. Therefore, \\'e used a standard, \\'ater-seal spirometer. This device, and the plleulllatic control valve systeln, were relatively inexpensive and reliable adaptions to existing equiplnent. More sophisticated, cOlllputer-assisted analysis of on-line data permit correction for drift and nonlinearity of signals, although at greater expense. Denlonstration of the accuracy and reliability of such devices \\'ould be iInportant, ho\\'ever. We identified statistically significant differences in values of TLC and RV deternlined with a separate spironleter when cOlllpared with a spirometer directly linked to the plethysmograph. However, the mean differences \\'ere slnall, and clinical diagnoses were changed only infrequently. Whether use ofless limited criteria for "mixed" disease (TLC less than 80 percent predicted) would have increased the frequency of altered diagnoses is not known. Regardless, large differences in deterlllinations of TLC appeared often related to a shfit in FRC during performance on the separate spirolueter. This shift might be recog~ized by the tendency for ERV to vary fron1 the real (or expected) value in the opposite direction to that which TLC varies. This error could then be confirmed by repeat testing. Recognition of this potential shift may inlprove the reliability of data determined with a spirolneter which is not linked to the plethysmograph. We believe that these data support the use of an unlinked spironleter for determination ofTLC and RV frolll FRC nleasured in a body plethysmograph. Differences in Plethysmographic Lung Volumes (Williams, Bencowltz)

ACKNOWLEDGMENTS: The authors thank Christine Anthony, Korry Jacobs, and Jill Minch for performing pulmonary function tests, and Hedda Schnur for preparation of this manuscript.

REFERENCES 1 Crystal RG, Bitterman PB, Rennard SI, Hance AJ, Keogh BA. Interstitial lung disease of unknown cause. Disorders characterized by chronic inflammation of the lower respiratory tract. N Engl J Med 1984; 310:235-44 2 Ferris BG, et al. Recommendation standardization procedures for pulmonary function testing. Am Rev Respir Dis 1978; 118:5588 3 Division of Lung Diseases, National Heart, Lung, Blood Institute. Epidemiology Standardization Project, Appendix 6. Am Rev Respir Dis 1978; 118:92-11 4 Working party, European Community for Coal and Steel. Standardized lung function testing: static lung volumes and capacities. Bull Eur Physiopathol Respir 1983;19(suppl 5):11-21 5 Walters LC, King TE, Schwarz MI, Waldron JA, Stanford RE, Cherniack RM. A clinical, radiographic, and physiological scoring system for the longitudinal assessment of patients with idiopathic-pulmonary fibrosis. Am Rev Respir Dis 1986; 133:97-

103

6 Bencowitz HZ. Inspiratory and expiratory vital capacity. Chest 1984; 85:834-35 7 Reinert M, Trendelenburg F. Advantage of simultaneous body plethysmographic and spirographic recording. Pneumonologie 1971; 146:79-82 8 Dubois AB, Bothelho SY, Bedell GN, Marshall R, Comroe JH. A rapid plethysmographic method for measuring thoracic gas volume: a comparison with nitrogen washout method for measuring functional residual capacity in normal subjects. J Clin Invest 1956; 35:322-26 9 Bohadana AB, Teculescu D, Peslin R, Da Silva J, Pina J. Comparison of four methods for calculating the total lung capacity measured by body plethysmography. Bull Eur Physiopathol Respir 1980; 16:769-76 10 ATS Statement. Snowbird workshop on standardization of spirometry. Am Rev Respir Dis 1979; 119:831-38

11 Leith DE, Mead J. Principles of body plethysmography. Bethesda, Md: National Heart, Lung, Blood Institute publication; November 1974 12 Heyer RA, Fogarty eM, Metzger LF, Murphy DMF. Standardization of total lung capacity [Abstract]. Am Rev Respir Dis 1978;117:126 13 Armitage ~ Berry G. Statistical methods in medical research, 2nd ed. Oxford: Blackwell Scientific Publications, 1987 14 Crapo RO, Morris AH, Gardner RM. Reference spirometric values using techniques and equipment that meets ATS recommendations. Am Rev Respir Dis 1981; 123:659-64 15 Goldman L, Becklake M. Respiratory function tests: normal values at median altitudes and the prediction of normal results. Am Rev Tuberc 1959; 79:457-67 16 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-10 17 Viljanen AA, Viljanen BC, Halttunen PK, Kreus KE. Body plethysmographic studies on non-smoking healthy adults. Scand J Clin Lab Invest 1981; 159 (suppl): 35-50 18 Schaanning CG, Gulsvik A. Accuracy and precision of helium dilution technique and body plethysmography in measuring lung volumes. Scand J Clin Lab Invest 1973; 32:271-77 19 Zapletal A, Motoyama EK, Van de Woestijne K~ Hunt VR, Bouhuys A. Maximum expiratory flow-volume curves and airways conductance in children and adolescents. J Appl Physiol 1969; 26:308-16 20 Dawson A. Pneumotachography. In: Clausen JL, Zarins L~ eds. Pulmonary function testing guidelines and controversies: equipment, methods, and normal values. London: Academic Press, 1982:91-97 21 Morris AH, Kanner RE, Crapo RO, Gardner RM. Clinical pulmonary function testing: a manual of uniform laboratory procedures. Salt Lake City: Intermountain Thoracic Society, 1984 22 Barnhart S, Hudson LD, Mason SE, Pierson DJ, Rosenstock L. Total lung capacity: an insensitive measure of impairment in patients with asbestosis and chronic obstructive pulmonary diseases. Chest 1988; 93:299-302

CHEST I 95 I 1 I JANUAR'f, 1989

123