Document gE3JkD8DyLrJ144VJNLxNb5X3

Reference Spirometric Values Using Techniques and Equipment that Meet ATS Recommendations1-3 i *ROBERT 0. CRAPO, ALAN H. MORRIS, and REED M. GARDNER Introduction T h e American Thoracic Society (ATS) and The American College of Chest Physicians have both recommended that spirometry be part of the routine evaluation of patients with respiratory diseases and those at risk of eloping respiratory diseases (2). pts have been made and are made to standardize the methods asuring, and inter(2-4). A statement has been issued by merous reports of ence values with prediction equaave been published, but few have used the equipment and ues suggested by the ATS ( 5 , 6 ) equations are linear height and age as in- s, although nonequations have been proposed is a common practice, for several spirometric parameters, to cale a predicted value from a regres- ation using f 20 Yo as the mal, despite evidence that this is not a valid normal limits for pre- ues (14). There is also disagreement on whether or not smokers should be included in reference studies. w e believe that studies using lifetime nsmoking populations provide the t predictors for reference values beCause of the high association of smokk g with pulmonary disease. Four $dies that reported normal values for hfetime nonsmokers (1,7-9), which we believe to be the most appropriate ref- erenCe population, did not consistently Use either equipment or techniques that meet ATS standards. The present study was performed to Provide prediction equations and more Precise lower limits of normal for Qhometric parameters in healthy, -11-screened normal subjects using ATS recommended techniques and SUMMARY F o r d expiratory volumes and flows were measured in 251 healthy nonsmoking men and women using techniques and equipment that meet American Thoracic Society (ATS) ncommendations. Linear regressionequations using height and age alone predict spirometric parametersas well as mom complex equations usingaddltlonai variables. Singlevalues for 95 YO confidence intervals am acceptable and should replace the commonly uaed methodof subtract- ing 20 K to determine the lower limit of normal for a predicted value. Our study produced predicted valuers for forced vital capacity and forced expiratoryvolume In one second that were almost identkal to thoro predicted by Morris and associates (1) when the data from their study were modified to be compatible with tho back extrapolation technique recommended by the ATS. The study of Morris and cdieaguw was petformed at sea ievd in rural subjects,w h m m w n was performed lit an altitude of 1,400 m in urban rubjecta. Either the pres. ent study or the study of Morris and ceworkem, modified to back extrapolation, could k mendedfor predictingnormalvalues. AM REV RESPIR OIS 1961; 129:659-664 equipment demonstrated to satisfy ATS criteria (5). Methods More than 90 070 of the subjects were volunteers from the Church of Jesus Christ of Latter-day Saints (Mormon). Each volunteer filled out a modified British Research Council questionnaire (IS), was examined by a pulmonary physician, and received a chest radiograph. "Normal"subjects were those who met all of the following criteria: ( I ) a lifetime nonsmoker (total smoking of less than 0.5 pack-yr and no smoking within 6 months of the study); (2) no symptoms of lung, heart, or chest wall disease; (3)a normal chest radiograph; (4) a normal physical examination of the heart, lungs, and chest wall. All tests were performed between 4 and 9 P.M. in the LDS Hospital Pulmonary Function Laboratory at Salt Lake City, Utah (altitude, 1.400 m). Subjects were weighed and measured in indoor clothing without shoes. Age was recorded to the nearest birthday. Spirometrywas performed with a water-seal metal bell spirometer (P1300, Warren E. Collins, Braintree, Mass.) connected to a potentiometer (10, 16). This spirometer meets ATS requirements (6) (courtesy of Dr. John L. Hankinson, National Institute of Occupational Safety and Health Appalachian Laboratory, Morgantown, W.Va., who tested the same model 13.5-L Collins, metal bell spirometer). The spirometer was calibrated daily (16). Spirometric analysis was performed with a previously described computerized system (10, 16) modified to provide tm,forced expiratory volumes and flows usmg the back extrapolationtechnique recommended by the ATS (5). Analog data from the potentiometer were sampled 100 timesh with a 10-bit analog-to-digital signal converter (resblution, f 15ml). Samplingwas initiated during the inspiration that preceded the forced expiratory maneuver. A three-point running average of the volume differenceswas continuously computed until th6 maximal volume difference occurred (peak flow). The waveform was then back extrapolated from the point of peak flow to determinetime zero (3,5). Theextrapolated volume was always less than 10 070 of the forced vital capacity(FVC). The waveforms with zero time, 1-s and 3-s time marks were displayed to the technician, who verified them. The data were stored on magnetic disks and were processed by a data analysis program called STRATO(17). All displaceable volumes were reported in liters BTPS, and the forced expiratory flow during the middle half of the forced vital capacity (Received in original form July 3, 1980 and in revisedform January 20, 1981) I From the Departments of Medicine and Biophysics, LDS Hospital and University of Utah, Salt Lake City, Utah. This study was supported by a grant from the Utah Heart Association. Requests for reprints should be addressed to Robert 0. Crapo, M.D., Pulmonary Division, LDS Hospital, 325 8th Ave., Salt Lake City, UT 84143. 659 I ' 660 -mCRAPO. MORRIS, AND TABLE 1 COMPARISON OF TECHNIQUES AND INSTRUMENTS IN SPIROMETRY REFERENCE STUDIES Instrument Studv Time Zero Techniaue TvDe Meets ATS Specitlcations Altitude (MI Locale Smokers Present study Morris, et a/. (1) Schoenberg, et a/. (7) Knudson, et a/. (8) Cherniack and Raber (9) Schmidt, et a/. (10) Kory, et a/. (11) Cotes, et a/. (12) Ferris, et a/. (13) BE K ? BE F K K K' BE Collins 13.5-L metal bell Stead-Wells Pneumotachygraph Pneumotachygraph Wedge@ Collins 13.5-L metal bell Collins 13.5-L metal bell McKerrow water-seal Collins 6 L Yes Yes ? ? Yes YeS Y8s No No 1,400 < 150 SL 730 230 1,400 Mixed SL 310 Urban Rural Mixed Urban Mixed Urban Urban 7 Rural No No No No No Yes Yes Yes Yest Deflnlrlons of abbrevierlons: ATS, Amerlcan Thoraclc Society, BE = back extrapolationtechnlque (5); K = Kory technique (11); F = mlnlmal flow threshold; SL = at or near sea level. * Modifled to Kory technique (100 ml rather than 200 ml used). t Separate smokers and nonsmokers. (FEFzs-~s)was reported in liters BTPS/S. Spirogramswere performed on subjectsin a sitting position until three acceptable trac- ings were obtained (5). The best test was defined by the Intermountain Thoracic Society (ITS)as *-test with the highest sum of FVC and the forced expiratory volume in one second (FEVI) (3). Although the ATS recommends computing the FEFZWSfrom the best test, it suggests reporting the largest FVC and FEVl regardless of the tracing from which they were obtained (5). The best test values for FVC and FEVI were therefore compared with the largest values of FVC and FEV, recorded by eacbs@ject. Multiple linear regressions were performed with standard statistical programs using STRATO (17-19). The spirometric values were regressed against the independent variables of height, age, weight, and body surface area. Transformations (X2,In X,fi,XI *Xz,X3)of FVC and FEVI and all independent variables were tested in an attempt to improve the predictability of the equations. The standard error of the estimate (SEE)and the coefficient of determination (Rz) were calculated for each equation. Residuals (measured value - predicted value) were graphically examined by comparing them with each independent variable and with the predicted dependent variable. The standard errors (Sf)for predicted single new dependent values (0were calculated from equations presented by Zar (18) (see Appendix). For all spirometric values and ratios. 95 Qo confidence limits were calculated by one-tailed f tests, because clinical interest is usually focused on whether or not the patient's measured value is less than the lower limit of normal. Separate 95 Vo confidence limits for each dependent variable were calculated as a function of height and age, at 10-cm and 10-yr increments, respectively, over the entire range of height and age of the sample population. Results Three hundred and eleven subjects were screened; 251 (126 women and 125 men) were designated "normal" and were included in the study (table 2). Each age decade, from 15 through 84 yr, contained an equal number of subjects (table 3). The mean differences between the largest and the best test values were 8.0 f 24.6 ml for FVC and 7.1 f 28.6 ml for =VI. The minimal spirometer accuracy recommended by the ATS is f 50 ml, or f 3 70of the reading (5, 6). Only two of the 251 subjects had a difference (maximum-best test) that exceeded these recommendations. For the regressions of FVC and FEVl against height and age, there was no significant difference between equations using maximal or best test values. Withinsubject coefficients of variation were measured for the three repeated tests of FVC and FEVI in 50 subjects. The co- Age, yr Height, cm Weight, kg TABLE 2 PHYSICAL CHARACTERISTICS OF SUBJECTS Range Men Mean i SD Women Range *Mean SD 15-91 157-194 60-111 49 f 20 175 f 7 78 f 13 17-84 146-178 44-105 49 f 20 161 f 7 68 f 13 efficient of variation f SD for pvc was 2.01 f 1.21 %; for FEV,, 1.96 1.41 %. The regression equations for m a and women are shown, respectively, in tables 4 and 5 . Tables of normal value by height and age are contained in Qbles B and C in the Appendix. We did not find that the addition of weight, body surface area, or transformations significantly improved the predictabil- ity of the regression equations ushg height and age alone. Analysis of re&uals showed that homoscedasticity was present for all equations. For t h e who wish to calculate precise confi- dence limits for each patient, the for- mulas and methods for measuring the 95 Yo confidence interval for a single new predicted value are shown in the Appendix. For any new predicted value, the confidence interval and, there fore, the lower limit of n o d inmaes as height and age vary from the mean height and age of the sample population. We found, however, that the range of confidence interval sizes was small and that, for clinical purposes, one could substitute a single mid-range value for all equations (tables 4 and 5). Such a substitution made a maximalerror in the 95 Vo confidence interval of less than 1 Yo for FVC, FEVI, and FEV3, and of only 1.6 Yo for FEFZ~-,~. A comparison of several simple regres- sion equations with the complex non- linear equations recommended by Schoenberg and colleagues (7) is summarized in table 6. Morris and associ- ates (1) calculated their forced expira- tory volumes and flows using the Kory technique rather than the back extra- polation technique now recommended bytheATS(1,5, 11). TheaverageFEVl calculated with the back extrapolation technique exceeds that calculated with the Kory technique by 179 ml(20). (In 20 of our normal subjects, this mean difference was 1% ml). By adding 179 ml to FEV,, we adjusted the predicted TABLE 3 AGE DISTRIBUTION OF SUBJECTS Men WOmen 15-24 25-34 35-44 45-54 55-64 6574 75-84 85-91 Total 17 18 18 18 19 19 19 18 18 19 17 17 15 17 2 125 126 REFERENCE VALUES FOR SPIROMETRY 661 TABLE 4 PREDiCTlON EQUATIONS FOR SPIROMETRIC PARAMETERS IN MEN Test, Units ~~ 1 FVC, L BTPS FEV,., L BTPS FEV,, L BTPS ! FEV,, L BTPS ,FE,F, L BTPS/S FEV,IFVC, Yo FRIJFVC,% 0.0800H O.0327H O.0414H 0.WH 0.0204H -0.1300H - 0.0827H Equations 95 % Confidence R' SEE Interval' - O.0214A -- 0.0152A 0.0244A - 0.0271A - 0.0380A -0.152 A -0.145 A -4.650 -1.914 -2.190 -3.512 +2.133 + 110.49 + 112.09 0.54 0.644 1.115 0.53 0.414 0.708 0.64 0.486 0.842 0.62 0.587 1.017 0.42 0.962 1.866 0.26 4.78 8.28 0.52 2.68 4.64 60 WOMEN -FifESEWT STUDY - MORRIS --- KNUDSON -SCHOENSERG 5.0 --CHERNIACK 2 m p 4.0 Y k 3.0 -MEN , Delinitionsolabbreeviefions: FVC = forcedvital capacity; H = height. in cm; A = age in yr; R' = coefficientof determlnation; SEE = standard error of the estimate; FEV = forcedexpiratoryvolumein the number of seconds lndlcatedby subscript; FEF, = forcedexpiratory flow duringthe middlehalf of the forced vital capacity. * The 95 YO confidenceintervalis calculatedfromaonetailed t test. It is thesinglevaluerecommendedfor all heights and ages in this study. When subtractedfrom the predictedvalue, it yields the lower limit of normal. Tablesof normal Ilo mZD ,1 I& , 110 I 180 m I$ MEIGMT knl values for men and women are available from the authorson request. Fig. 1. Comparisonsof forced vital capacity (FVC) and forced expiratory volume in one second(FEV,) TABLE 5 PREDICTION EQUATIONS FOR SPIROMETRIC PARAMETERS IN WOMEN In five reference studies of lifetime nonsmokers: Morris and associates (l), Knudson and colleagues (E), Schoenberg and csworkers (7), Cherniack and ? Test, Units FVC, L BTPS FEV., L BTPS FEV,, L BTPS FEV,, L mps ,FE,F, L m p s l s FEVJFVC, % FEVJFVC, % O.0491H 0.0238H 0.0342H 0.0442H - 0.0154H 0.2020H - 0.0937H Equations - 0.0216A -- 0.0185A 0.0256A -- 0.0257A 0.WA -0.252 A -0.163 A -3.590 -0.809 - 1.578 -2.745 +2.663 + 126.58 + 118.16 95 Yo Raber 0,and the present study. Calculations were Confi- made for a person 40 yr of age. Weights for the R' SEE dence Interval. data of Schoenberg and associates are average weights from actuarlal tables for each height (21). The data of Morris and ceworkers were adjusted 0.74 0.393 0.676 to a back extrapolation technique (20). 0.72 0.294 0.506 0.80 0.80 0.326 0.360 0.561 0.620 ated using the best test from three ac- 0.60 0.792 1.363 ceptable tracings. Nathan and col- 0.43 5.26 9.06 leagues (22) have shown tE%t~Jittleis 0.48 3.11 5.36 gained by doing more than three tests. For definitionsof abbrevlations, see Table 4. In our normal subjects there was no significant difference between selecting TABLE 6 COMPARISON OF LINEAR AND NONLINEAR EQUATIONS FOR PREDICTING FVC AND FEV, the largest FVC or FEVI result or using the best test result. We therefore recommend using the best test method of I Reference Variable Formula Format SEE R' selection because of its simplicity. The Schoenberg, et el. (7) Morris, et el. (1) Schmidt, et el. (10) Present study Resent study Schoenberg, et el. (7) Morris, et el. (1) Schmidt, et el. (10) Present study Present study Schoenberg, et al. (7) Morris, et el. (1) Schmidt, etel. (10) Present study Present study Schoenberg, et 6 4 . 0 Morris, et el. (1) Schmidt, et a/. (10) Present study Present study FVC F FVCF FVC F FVC F FVCF FVC M FVCM FVC M FVC M FVCM FN, F FEV, F FEV, F FEV, F FEV, F FEV, M FEV, M FEV, M FEV, M FEV, M aHW + blnA + cAH + dW + e W + fAW + k 0.433 0.56 equations in tables 4 and 5 would, how- aH + bA + k 0.520 0.50 ever, apply to either method of test se- +aH + bA + k aHW + blnA + cAH dW + e W + fAW + k aH+bA+k aHW + blnA + CAW + d W + k aH+bA+k aH + bA + k aHW + blnA + CAW + dW2 + k aH + bA + k 0.360 0.398 0.391 0.597 0.740 0.460 0.624 0.645 0.58 0.74 0.74 0.56 0.42 0.53 0.57 0.53 lection. Comparison of equations us- ing height and age with the nonlinear equations proposed by Schoenberg and co-workers (7) (table 6) shows that the coefficient of determination and the SEE of the simpler, linear equations are comparable to, if not better than, aH aH ++ bHW bA + + k cinA + dAH + eW2 + k aH + bA + k aH + bHW + clnA + dAH + eW' + k 0.323 0.68 those of the more complicated, nonlin- 0.470 0.53 ear equations. When we added multiple 0.290 0.324 0.71 0.81 transformationsof the independent vari- aH + bA + k 0.328 0.80 able in both linear and nonlinear re- aHW + blnA + CW + dWz + eAHW + k aH + bA + k 0.440 0.550 0.64 0.53 gressions (including the regressions proposed by Schoenberg and associ- aH + bA + k 0.390 0.64 ates), we could not demonstrate signifi- aHW + binA + CW + d W + eAHW + k aH + bA + k 0.483 0.66 cant improvement over the linear re0.482 0.65 gressions that used only height and Delinitions01 abbreviations: SEE = standarderror of the estimate, in liters BTPS;R' = coefficientot determlnation; N C = forcedvital capacity; F = female; H i:height; in = natural logarithm;A = age; W = weight; k = constant;M = male; FEV, = forced expiratory volume in one second. age. We therefore recommend that the simpler linear regressions using height and age be used. values reported by Morris and coworkers to approximate those that would have been obtained by the back extrapolation method. These values were used in a graphic comparison of the five studies of lifetime nonsmokers (figure 1). Discussion The equations in this study were gener- The 95 070 confidence intervals for all predicted spirometric values are relatively constant as height and age vary from the mean height and age of the sample population. With almost constant confidence intervals over a wide 662 CRAPO. MORRIS. AND QARDNER range of predicted values, it is not rea- sonable to consider the predicted value -f 20 olo as normal. For example, the predicted FEVl for men varies from 5.48 to 2.11 L BTPS. The recommended confidence interval of 0.84 L (table 4) would, therefore, constitute from 15 to 40 070 of the predicted value, making it untenable to use f 20 070 to define the limits of normal. We therefore recommend that 95 olo confidence intervals (single value as in tables 4 and 5 , or cal- culated as in the Appendix) be used to define the lower limit of normal and that the practice of subtracting 20 olo be abandoned. TABLE A INVERTED VARIANCE-COVARIANCE MATRIX VALUES FOR THE CALCULATION OF SV Female spirometry Male spirometry 2.2456 1W 2.1460 lcr' 3.6042 10-' 2.8452 IC' 2.5956 le' 2.4318 It' Oetinit~onsof abbrwiatlons: See text of the Appendix. These values are the same for forced vital capacity (FVC). timed forced expiratoty volumes (FEV.,, FEV,. FEVJ, forced expiratoryflow during the middle half of the FVC (FEF,,,), FEV,iFVC. and FEVJFVC. TABLE B PREDICTED SPIROMETRIC VALUES FOR ADULT WOMEN 150 155 '% 160 165 170 175 180 20 30 40 50 60 70 20 30 40 50 80 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 3.34 3.13 2.91 2.69 2.48 2.26 3.59 3.37 3.18 2.94 2.72 2.51 3.83 3.62 3.40 3.19 2.97 2.75 4.08 3.86 3.85 3.43 3.22 3.00 4.32 4.11 3.89 3.68 3.46 3.24 4.57 4.35 4.14 3.92 3.71 3.49 4.82 4.60 4.38 4.17 3.95 3.74 FEV, -(L) 3.04 2.79 2.53 2.28 2.02 1 .?7 3.21 2.88 2.70 2.45 2.19 1.94 3.38 3.13 2.87 2.62 2.36 2.11 3.55 3.30 3.05 2.79 2.53 2.28 3.73 3.47 3.22 2.88 2.71 2.45 3.90 3.64 3.39 3.13 2.88 2.82 4.07 3.81 3.56 3.30 3.05 2.79 91.3 88.8 86.2 83.7 81.2 78.7 90.3 87.0 85.2 82.7 80.2 77.7 89.3 88.7 84.2 81.7 79.2 78.7 88.3 85.7 83.2 80.7 78.2 75.7 87.2 84.7 82.2 79.7 77.2 74.7 86.2 83.7 81.2 78.7 78.2 73.6 85.2 82.7 80.2 77.7 75.1 72.6 Definitions of abbreviations: L = liters BTPS; Us = liters s r w s . 4.07 3.81 3.15 2.89 2.23 1. n 4.15 3.89 3.23 2.77 2.31 1.86 4.23 3.77 3.31 2.86 2.39 1.93 4.30 3.84 3.38 2.m 2.46 2.00 4.38 3.92 3.46 3.00 2.54 2.08 4.46 4.00 3.54 3.08 2.62 2.16 4.53 4.07 3.61 3.15 2.69 2.23 Our study population was quite similar to that reported on by Morris and associates (I), because both studies contained lifetime nonsmokers and large numbers of Mormons. The differences between our population and theirs were that our population was urban and lived at 1,400-m altitude, whereas their population was rural and lived near sea level. The study of Morns and colleagues has been criticized for using an unusual population (7). We believe that it was unusual only in that it consisted of healthy, lifetime nonsmokers. Mormons generally come from a diverse northern and middle European background. In comparison with other studies, our population was more highly screened, which might explain the higher spirometric values of the present study when compared to 3 of the 4 studies using nonsmokers (figure 1). The results reported by Morris and as- sociates, when adjusted to approximate the back extrapolation technique, were quite similar to ours. For female FVC and FEV, and male FVC,our study re- sults were almost identical to the adjusted values of Morris and colleagues (1). The correlation between the two studies was not as good for male FEV, but was still excellent until the extremes of age and height were reached. The main difference betweeen the male FEVl equations was the age coefficient, a difference that might be explained by the more even age distribution in our study. Both the present study and Morris and co-workers' study (adjusted for back extrapolation) conform to ATS technique and equipment recommendations. The high correlation between these two studies of subjects with similar ethnic backgrounds is important because they are separated by the passage of 10 years and by 1,400-m altitude, and because one concerns rural and the other, urban subjects. Either the present study or Morris and colleagues' study (6) (adjusted to approximate the back extrapolation technique) can be recommended for the prediction of ref- erence values for rural or urban whites of northern and middle European extraction from sea level to 1,400-m altitude. Appendix For these multiple linear regression equations, the standard error (Sp)of a single new predicted value ( f ) for a given height and age is calculated from the fol- REFERENCE VALUES FOR SPIROMETRY .. TABLE C PREDICTED SPIROMETRIC VALUES FOR ADULT MEN 683 Berlin, Margaret Crapo, and Ann Gennaro for their assistance in this study. Height (cm) 155 20 4.22 30 4.01 40 3.79 50 3.58 60 3.37 70 3.15 160 20 4.52 30 4.31 40 4.09 50 3.88 60 3.61 70 3.45 165 20 4.82 30 4.61 40 4.39 50 4.18 60 3.97 70 3.75 170 20 5.12 30 4.91 40 4.69 50 4.48 60 4.27 70 4.05 175 20 5.42 30 5.21 40 4.99 50 4.78 80 4.57 70 4.35 180 20 5.72 30 5.51 40 529 50 5.08 80 4.87 70 4.65 185 20 6.02 30 5.87 40 5 . s 50 5.38 Bo 5.17 70 4.95 im 20 6.32 30 6.11 40 5.89 50 5.88 Bo 5.47 70 525 FM deflnltlms of abbrsu!ations,aoe Table 8. FEV, (L) 3.74 3.49 3.25 3.01 2.76 2.52 3.95 3.70 3.46 3.21 2.97 2.73 4.15 3.91 3.66 3.42 3.18 2.93 4.36 4.12 3.87 3.63 3.36 3.14 4.57 4.32 4.08 3.84 3.59 3.35 4.77 4.53 429 4.04 3.80 3.55 4.9s 4.74 4.49 4.25 4.01 3.76 5.19 4.94 4.70 4.46 4.21 3.97 87.4 4.53 85.8 4.16 84.3 3.78 82.8 3.40 81.3 3.02 79.7 2.64 86.7 4.64 85.2 4.26 83.7 3.88 82.1 3.50 80.6 3.12 79.1 2.74 86.1 4.74 84.5 4.36 83.0 3.98 81.5 3.60 80.0 3.22 78.4 2.84 85.4 4.84 83.9 4.46 82.4 4.08 80.8 3.70 79.3 3.32 77.8 2.94 84.8 4.94 83.2 4.56 81.7 4.18 80.2 3.80 78.7 3.42 77.2 3.04 84.1 5.04 82.8 4.66 81.1 4.28 79.5 3.90 78.0 3.52 76.5 3.14 83.5 5.15 81.9 4.77 80.4 4.39 78.9 4.01 77.4 3.63 75.9 3.25 82.8 5.25 81.3 4.07 79.8 4.49 7.8.2 A..1.1. 76.7 3.73 75.2 3.35 lowing equationJl8): Sp = [(SEE)' (1 _+ +l/n C I I ( H - H ) 2 f C l z ( H - H ) ( A - A ) + C2I (A - A ) (H - H)+ C22 (A - -Because C12= CzIthe equation plified to: Sp = ((SEZ)'(1 +J/n ca+n be sim- CtI (H H)' + 2 Ci2 ( H - H) (A - A) + Czz (A - &2)]yz. For both equations, SEE is the standard error of the estimate; n, the popu- lation size; H, the height, in cm; A, age, in yr, at which $e new prediction is to be made; H and A, mean height and age, re- spectively, in the sample used to generate the regression equation; G I ,C12,GI.and CZ2are the values from the inverted vari- ance-covariance matrix corresponding to the appropriate combinations of height and age. The subscript 1 corresponds to height; 2, to age. Tables 2 and 3 contain the appropriate values of n and the mean heights and ages. Table A contains the appropriate C I I ,C l z , G I ,and C2*values for the prediction equa- tions in this study. The 95 070 confidence in- tervals using a one-tailed t test in samples of 125 or 126 subjects (present study) can be calculated using the formula: dicted P i 1.66 Sp. = pre- Acknowledgment The writers thank Sonnie Adams, Steven References 1. Morris JF, Koski A, Johnson LC. Spirometric standards for healthy nonsmoking adults. Am Rev Respir Dis 1971; 10357-67. 2. Permutt S, Chester E,Anderson W,Cugell D,Petty TL, Sharp JT. Office spirometryin clinical practice. Statement of the American College of Chest Physicians' Committee on clinic and office pulmonary function testing. Chest 1978; 74: 298. 3. Kanner RE, Morris AH, eds. Clinical pulmonary function testing: a manual of uniform laboratory procedures for the intermountain area. Salt Lake City: Intermountain Thoracic Society, !915. 4. Ferris BG (Principal Investigator). Epidemiology standardization project. Am Rev Respir Dis 1978; 118(Part 2):55-88. 5. Gardner RM, et al. Snowbird workshop on standardization of spirometry: a statement by the American Thoracic Society. Am Rev Respir Dis 1979; 119~831-8. 6. Gardner RM. Hankinson JL, West BJ. Evaluating commercially available spirometers. Am Rev Respir Dis 1980; 121:73-82. 7. Schoenberg JB. Beck GL, Bouhuys A. Growth and decay of pulmonary function in healthy blacks and whites. Respir Physiol 1978; 33: 367-93. 8. Knudson RJ. Slatin RC, LebowitqMD. Burrows B. The maximal expiratory flokr-volume curve. Am Rev Respir Dis 1976; 113:587600. 9. Cherniack RM, Raber MB. Normal standards for ventilatory function using an automated wedge spirometer. Am Rev Respir Dis 1972; 106. 38-44. 10. Schmidt CD, Dickman ML, Gardner RM, Brough FK. Spirometric standards for healthy elderly men and women. Am Rev Respir Dis 1973; 108~933-9. 11. Kory RC, Callahan R, Boren HG, Syner JC. The Veterans Administration-Army coop erative study of pulmonary function. Clinical spirometry in normal men. Am J Med 1961; 30. 243-58. 12. Cotes JE,Rossiter CE,Higgins ITT, Gilson JC. Average normal values for the forced expiratory volume in white Caucasian males. Br Med J 1966; 1:1016-19. 13. Fems BO, Anderson DO, Zickmantel R. Prediction values for screening tests of pulmonary function. Am Rev Respir Dis 1965; 91:25261. 14. Sobol BJ, Sobol PG. Per cent of predicted as the limit of normal in pulmonary function testing: a statistically valid approach. Thorax 1979; 34:l-3. 15. Medical Research Council's Committee on the aetiology of clinic bronchitis: Standardized questionnaireson respiratory symptoms. Br Med J 1960; 2:1665. 16. Dickman ML, Schmidt CD, Gardner RM. Marshall HW, Day WC, Warner HR. On-line computerized spirometry in 738 normal adults. Am Rev Respir Dis 1%9; 100:780-90. 17. Warner HR. Computer-assisted medical decision-making. New York: Academic Press. 1979. 18. Zar JH. Biostatistical analysis. Englewood Cliffs, NJ: Prentice-Hall, 1974. I 684 19. Afifi AA. Azen SP. Statistical analysis: a computer oriented approach. New York: Academic Press, 1973. 20. Smith AA, Gaensier EA. Timing of forced expiratory volume in one second. Am Rev Respir Dis 1975; 112:882-5. 21. Association of Life Insurance Medical Directors and the Society of Actuaries. 1979 Build CWW, MORRIS, AND Q A R ,,~ ~ \ and Blood Pressure Study. 22. Nathan SP. Lebowitz Spirometric testing: number MD. KnudPoe of tests rcqt&d 4 selection of data. Chest 1979; 76:384-8.