Inclusion of peak expiratory flow for selection of the ‘ best ’ forced vital capacity manoeuvre

OB JEC TIVE: To as sess four dif fer ent cri te ria for se lect ing the ‘best’ forced vi tal ca pac ity (FVC) ma noeu vre to be used for clini cal di ag nos tic pur poses. DE SIGN: Cri te rion stan dard. SET TING: Gen eral popu la tion sur vey per formed in 198082. PA TIENTS: One thou sand, two hun dred and eightythree sub jects (age range eight to 64 years) were first strati fied into five mu tu ally ex clu sive groups ac cord ing to the fol low ing cri te ria: si mul ta ne ous larg est FVC, forced ex pi ra tory vol ume in 1 s (FEV1) and peak ex pi ra tory flow (PEF) (group 1; n=481); iso lated larg est FVC (group 2; n=223); iso lated larg est FEV1 (group 3; n=144); iso lated larg est PEF (group 4; n=299); and over lap ping cri te ria (group 5; n=136). IN TER VEN TION: Sub jects per formed spi ro me try fol low ing Ameri can Tho racic So ci ety (ATS) pro to col and filled out a stan dard ized res pi ra tory ques tion naire. MAIN OUT COME MEAS URES: Spi ro grams were ana lyzed by ex am in ing the fre quency of spi ro me try ab nor mali ties with re gard to the pres ence of res pi ra tory symp toms, first within mu tu ally ex clu sive groups of sub jects and then within the whole sam ple. The hy pothe sis of the role of PEF in ‘best test’ se lec tion was for mu lated af ter data col lec tion. MAIN RE SULTS: When the iso lated larg est PEF cri te rion was used, the fol low ing data were ob tained: the high est preva lence of spi ro met ric ab nor mali ties for each FVC pa rame ter in each mu tu ally ex clu sive group; the high est pre dic tive value for mean and in stan ta ne ous ex pi ra tory flows in sepa rat ing symp to matic from as ymp to matic sub jects; and fi nally, us ing the whole sam ple, higher lev els of sen si tiv ity and simi lar speci fic ity to other cri te ria for all test pa rame ters (all over 90%, ex cept for PEF). CON CLU SIONS: While main tain ing the cur rent ATS cri te ria of ac cept abil ity and vari abil ity for FVC tri als, it is pro posed that the curve that bet ter re flects maxi mal ex pi ra tory ef fort, ie, that with the larg est PEF, be re corded and ana lyzed for spi ro met ric vari ables.

D e spite gen eral agree ment on the Ameri can Tho racic So - ci ety (ATS) defi ni tion of forced vi tal ca pac ity (FVC) as "the maxi mal vol ume of air ex haled with maxi mally forced ef fort from a po si tion of maxi mal in spi ra tion" (1-3), the standardi za tion of the ma noeu vre has proven to be dif fi cult.A number of is sues re lated to the per form ance and analy sis of the test have been dis cussed in nu mer ous pa pers, deal ing with de ter mi na tion of the start ing point (back ex trapo la tion) (1,4) and of the end-point (2,5) of the spi ro gram; cri te ria for ac cept abil ity and re pro duci bil ity (1)(2)(3)6,7); se lec tion of the best spi ro met ric val ues (8); and ef fects of tho racic gas compres sion (9,10).
One of the re main ing con tro ver sial prob lems re lates to the se lec tion of the 'best' curve, ie, the flow-volume curve from which mean and in stan ta ne ous ex pi ra tory flows are ob tained (2,4).The In ter moun tain Tho racic So ci ety (11) rec ommended that all the pa rame ters, in clud ing FVC and forced ex pi ra tory vol ume in 1 s (FEV 1 ), should be ob tained from the curve with the larg est sum of FVC plus FEV 1 .In the guidelines pub lished by the ATS in 1979 (1) and, in re vised form, in 1987 (2) and in 1995 (3), it was rec om mended that the larg est FVC and the larg est FEV 1 be re corded, even when ob tained from dif fer ent ma noeu vres, and that the ma noeu vre with the larg est sum of FVC plus FEV 1 (ie, the ATS 'best test' curve) be used to cal cu late the mean and in stan ta ne ous forced ex pi ra tory flows.Re cently the Euro pean Res pi ra tory So ci ety pub lished a docu ment on the stan dardi za tion of lung func tion meas ure ments, which con tains the pro posal that two al ter na tive meth ods be used to ob tain flow-volume in dexes: "the first one (en ve lope method) en tails su per im pos ing the curves from to tal lung ca pac ity (TLC) to form a com pos ite maxi mal curve; the larg est FVC curve is used to de line ate the high est in stan ta ne ous flows at speci fied lung vol umes.The sec ond method is to take the high est in stan ta ne ous flow from three tech ni cally sat is fac tory FVC ma noeu vres; the FVC from the cho sen flow-volume curves should not dif fer from the larg est FVC by more than 5%" (12).Thus, none of the cur rent cri te ria for the se lec tion of the flow-volume curve from which the FVC pa rame ters are cal cu lated rec om mends that peak ex pi ra tory flow (PEF) be taken into ac count.
The flow-volume curve in clud ing the larg est PEF should meet the ATS defi ni tion of FVC, since an FVC ma noeu vre can not be con sid ered really 'forced' if the maxi mal ex pi ratory ef fort (ie, with a sat is fac tory start of the spi ro gram and high PEF) is not achieved.
Sub maxi mal ef forts (in di cated by a slow start to the spi rogram and low PEF) can re sult in ei ther higher or lower FEV 1 val ues than maxi mal ef forts (3,9,13) due to less dy namic com pres sion of air ways or fail ure to reach a maxi mal TLC, re spec tively.Based on the cur rent ATS cri te ria, computercontrolled rou tines should se lect mean and in stan ta ne ous expi ra tory flows from the 'best test', whether per formed with maxi mal or with sub maxi mal ef fort; the lat ter may re sult in an in cor rect clini cal in ter pre ta tion.In fact, ab nor mali ties of mean and in stan ta ne ous ex pi ra tory flows are not usu ally taken into ac count when FEV 1 and FEV 1 /vi tal ca pac ity percent age (FEV 1 /VC%) are within the ex pected range (14).Nev er the less, in the pres ence of a bor der line value for FEV 1 /VC%, they may help to con firm the pres ence of air way ob struc tion, even tak ing into ac count the wide intraindividual vari abil ity pres ent in nor mal sub jects (14).
The aim of our study was to as sess whether a dif fer ent crite rion for the se lec tion of the 'best test' curve, based on the FVC ma noeu vre per formed with the maxi mal PEF, may further en hance the di ag nos tic power of this test.This method was com pared with ATS cri te ria by ex am in ing the fre quency of ab nor mali ties in FVC pa rame ters in re la tion to the presence of res pi ra tory symp toms.

PATIENTS AND METHODS
Data set: FVC trac ings were ac quired dur ing the first crosssectional epi de mi ol ogi cal sur vey con ducted in 1980-82 on a gen eral popu la tion sam ple (n=3285, 47.9% males, age range eight to 64 years) liv ing in the Po River Delta area (20 km south of Ven ice, It aly).The study pro to col in cluded ac qui sition of the fol low ing res pi ra tory data: FVC; slow vi tal ca pacity; sin gle breath dif fus ing ca pac ity; sin gle breath ni tro gen pres ence of res pi ra tory symp toms (cough, phlegm, ef fort dysp nea, at tacks of short ness of breath with wheeze, wheezing dur ing and apart from com mon colds); pres ence of cardiac or pul mo nary dis ease (chronic bron chi tis, em phy sema, asthma, heart dis ease, etc); pres ence of rhini tis or al ler gic dis or ders; pres ence of child hood res pi ra tory dis ease; presence of fam ily his to ries of asthma, chronic bron chi tis, emphy sema, tu ber cu lo sis, lung can cer, atopy or al ler gies; oc cu pa tional ex po sure and smok ing his tory; and so cio economic in for ma tion.Trained nurses ad min is tered the question naire us ing stan dard pro to col.Sub jects were con sid ered to be symp to matic if they answered af firma tively to any ques tion re fer ring to res pi ra tory symp toms or dis eases.Sub jects who an swered af firma tively to stan dard ques tions re gard ing a di ag no sis of chronic bronchi tis and/or pul mo nary em phy sema were de fined as suf fering from chronic ob struc tive pul mo nary dis ease (COPD).Spi ro me try: A com put er ized pneu mo ta cho graph (Pul monary Sys tem 47804S, Hewlett-Packard, Mas sa chu setts) was used for the ac qui si tion and on-line analy sis of lung func tion data dur ing the field sur vey.The sys tem con sisted of a Fleisch pneu mo ta cho graph number 3 (Hewlett-Packard) for flow meas ure ments, linked via an ana logue to digi tal (A/D) Con verter (47310A-HP) to a 9825 A Hewlett-Packard cal cula tor (HP 9825, Cali for nia).In this sys tem the pres sure change, in duced by the res pi ra tory flow pass ing through the pneu mo ta cho graph, was trans lated into mil li volts.The A/D con verter dig it ized the meas ured mil li volts and the com puter in te grated the vol ume sig nal from the flow sig nal.Pneu mo tacho graph re sponse was lin ear (±3% be tween 1 and 13 L/s).The vol ume of the pneu mo ta cho graph was cali brated daily with a 3.0 L stan dard sy ringe.Be cause these stud ies were car ried out be fore in ter na tional stan dardi za tion of flow meas ure ments, the sys tem was re cently tested dur ing a range of flow rates, be tween 1.61 and 7.56 L/s, gen er ated by a standard de com pres sive pump and four re sis tances.The per centage varia tion be tween re peated read ings among re sis tors was be tween 0.70% and 1.83% (0.03 to 0.08 L/s), well within the 5% or 0.15 L/s rec om mended by Crapo et al (3).Ac cu racy of read ings com pared with meas ure ments made with a watersealed spi ro me ter var ied be tween 0.3% and 3.5%, with maximum dif fer ences of 0.04 and 0.22 L/s.
Each sub ject per formed at least three ac cept able and repro duci ble FVC ma noeu vres, as speci fied by ATS pro to col; the end-point of the FVC ma noeu vre was de ter mined us ing feed back re quir ing con secu tive sam ples to de ter mine a flow of less than 15mL/s, and no time limi ta tion was im posed by the al go rithm (5).There was no limit to the number of ac quirable FVC ma noeu vres.For each FVC ma noeu vre, the fol lowing pa rame ters were ob tained: FVC; FEV 1 ; forced ex pi ra tory flow be tween 25% and 75% of FVC (FEF 25-75% ); forced expi ra tory flow be tween 75% and 85% of FVC (FEF 75-85% ); maxi mal ex pi ra tory flow at 50% of FVC (MEF50%); maximal ex pi ra tory flow at 75% of FVC (MEF75%); and PEF.Group analy ses: Spi ro met ric trac ings of 1283 sub jects were re viewed and grouped first ac cord ing to se lec tion cri te ria and then to the to tal group.Analy ses ap plied to mu tu ally ex clu sive groups: Fol low ing a hi er ar chic pri or ity or der, each sub ject was as signed to one of five groups, ac cord ing to the fol low ing cri te ria for 'best' curve se lec tion: larg est FVC, FEV 1 and PEF in the same curve, thus in clud ing ATS cri te ria for se lec tion of the curve to be used for di ag nos tic pur poses (group 1); larg est FVC in a dif fer ent curve from the one with the larg est sum of FVC and FEV 1 (group 2); larg est FEV 1 in a dif fer ent curve from the one with the larg est FVC and from the one with the larg est sum of FVC and FEV 1 (group 3); iso lated larg est PEF in a dif fer ent curve from those char ac ter iz ing the pre vi ous three  .Analy sis of vari ance was used to com pare anthro pomet ric pa rame ters and the χ 2 test to com pare smok ing cate go ries.Paired t test was used to com pare in tra group mean val ues of FVC in dexes, and a t test for in de pend ent vari ables was used to com pare sen si tiv ity val ues among groups and cate go ries.Each pa rame ter of each group was clas si fied as nor mal or ab nor mal if it was higher or lower, respec tively, than the nor mal 95th per cen tile limit from pre diction equa tions de rived within the same gen eral popu la tion sam ple (16).Sen si tiv ity, speci fic ity and pre dic tive value were used to test the abil ity of the flow-volume curve parame ters of dif fer ent groups to dis crimi nate be tween symp tomatic and as ymp to matic sub jects (19).

RESULTS
No dif fer ence was pres ent in sex, an thro pomet ric pa rameters and smok ing hab its among the five mu tu ally ex clu sive groups (Ta ble 2).
In or der to ex am ine data with out over lap ping se lec tion cri te ria, a number of analy ses were per formed on the first four groups (n=1147).First, the av er age for each of the seven pa rame ters of the flow-volume curve ob tained in each group   was com pared with the av er age ob tained by ap ply ing ATS cri te ria in that sub group (Ta ble 3).For group 1 (n=481; 42%) the pa rame ters of only one curve were used be cause these sub jects, com pris ing the larg est group, showed the si mul tane ous pres ence of the larg est FVC, FEV 1 and PEF in the same curve.For group 2 (n=223; 19%), group 3 (n=144; 13%) and group 4 (n=299; 26%), all pa rame ters dif fered signifi cantly, as as sessed by paired t test, be tween curves selected us ing the ATS-Σ (ie, the curve with the larg est sum of FVC plus FEV 1 ) and curves se lected by the al ter na tive cri terion.Thus, larger mean val ues of all pa rame ters were found in the ATS-Σ curve in group 2 (with the ob vi ous ex cep tion of FVC) and in group 4 (with the ob vi ous ex cep tion of PEF), and in the curve of the al ter na tive cri te rion for group 3 (with the ex cep tion of FVC).The high est per cent ages of ab nor mal val ues (Ta ble 4), using the nor mal 95th per cen tile as a thresh old value, were shown by group 1 for FEF 25-75% and MEF75%, and by group 4 for FVC, FEV 1 , FEF 75-85% and MEF50%.As re gards PEF, the high est fre quency of ab nor mal ity, as as sessed by the ATS-Σ curve, was shown by group 4.
Ta ble 5 shows the sen si tiv ity, speci fic ity and pre dic tive val ues of pa rame ters de rived from the flow-volume curve in de tect ing symp to matic sub jects, ie, those with at least one res pi ra tory symp tom.The high est sen si tiv ity with re gard to FEV 1 was shown by group 4, while the high est speci fic ity for the same pa rame ter was pres ent in groups 1 and 3. Fur thermore, the high est sen si tiv ity for mean and in stan ta ne ous expi ra tory flows (ex cept FEF 75-85% ) was ex hib ited by group 1, fol lowed by group 4; the lat ter showed the same sen si tiv ity value for FEF 25-75% as group 1 and the high est speci fic ity val ues for MEF50%, MEF75% and PEF.Moreo ver, the high est sen si tiv ity and speci fic ity val ues for FVC were shown by group 3, who had also the high est speci fic ity for FEF 75-85% , while the high est sen si tiv ity for this pa rame ter was dem on strated by group 2. The high est pre dic tive val ues for FVC (50%) and FEV 1 (54.2%) were reached by groups 3 and 1, re spec tively, fol lowed by group 4. The lat ter showed the high est pre dic tive val ues for all mean and in stan ta ne ous ex pi ra tory flows and PEF (48.9%).Dif fer ences in sen si tiv ity be tween groups were not sta tis ti cally sig nifi cant by t test.
Preva lence rates of res pi ra tory symp toms and dis eases are   FEF25-75% Forced ex pi ra tory flow be tween 25% and 75% of forced vi tal ca pac ity (FVC); FEF75-85% Forced ex pi ra tory flow be tween 75% and 85% of FVC; FEV1 Forced ex pi ra tory vol ume in 1 s; MEF50% Maxi mal ex pi ra tory flow at 50% of FVC; MEF75% Maxi mal ex pi ra tory flow at 75% of FVC; PEF Peak ex pi ra tory flow.The high est val ues for sen si tiv ity (Sens), speci fic ity (Spec) and pre dic tive value (PV) reached by each pa rame ter in groups 1 to 4 are un der lined.Thresh old value is the nor mal 95th per cen tile.Sen si tiv ity dif fer ences be tween groups were not sta tis ti cally sig nifi cant by t test shown in Fig ure 1.All preva lence rates were higher in group 4, ex cept for at tacks of short ness of breath with wheeze and dysp nea, which were higher in group 1.How ever, the dif ference among the rates was sig nifi cant only for the di ag no sis of COPD.
Re sults of analy ses per formed on the four cate go ries, each con sti tuted from the whole sam ple of 1147 sub jects, are shown in Ta ble 6.The high est sen si tiv ity for all FVC parame ters, ex cept MEF75% and PEF, was pres ent in cate gory 4. The sen si tiv ity dif fer ences were not sta tis ti cally sig nificant among the four cate go ries by t test.As re gards speci ficity, the high est val ues were ex hib ited by cate gory 3 for all pa rame ters, with the ex cep tion of FVC and PEF, the high est val ues for which were pres ent in cate go ries 2 and 4, re spec -tively.The high est pre dic tive val ues for FVC (32.5%) and PEF (39.0%) were reached by cate gory 4, while cate gory 3 showed the high est pre dic tive val ues for all the other pa rameters.

DISCUSSION
Sev eral authors (20)(21)(22) have as sessed the fea si bil ity of us ing PEF rate as an in dex of a sub ject's ef fort dur ing an FVC ma noeu vre.Krowka et al (9) dem on strated "a sig nificant posi tive cor re la tion be tween PEF rate and the mag nitude of the transpul mon ary pres sure⋅area prod uct dur ing the first sec ond of the FVC ma noeu vre."Re sults of large epi demi ol ogi cal stud ies (13) sug gest that the use of PEF re produci bil ity may fur ther en hance the tech ni ci an's abil ity to de tect poorly per formed FVC ma noeu vres.Also, the European Res pi ra tory So ci ety, in a 1993 docu ment on the standardi za tion of lung func tion (12), sug gested the use of PEF re pro duci bil ity (within 10% of the maxi mal value) for computer se lec tion of ac cept able spi ro grams.Fi nally, very recently (3) the ATS en cour aged in ves ti ga tors to meas ure time-to-PEF or rise-time of PEF when as sess ing a sub ject's cor rect per form ance of FVC ma noeu vres.
Pa rame ters ob tained from FVC ma noeu vres in which the larg est PEF oc curs should re flect per form ance as so ci ated with maxi mal ef fort.Thus, the in clu sion of this cri te rion might im prove the clini cal in ter pre ta tion of the FVC test; indeed, our data show that dif fer ent se lec tion cri te ria for the 'best' curve lead to dif fer ent val ues for sev eral of the de rived vari ables.In fact, within groups 2, 3 and 4, we ob served statis ti cally sig nifi cant dif fer ences be tween val ues de rived from the ATS-Σ curve and those from spi ro grams se lected by the al ter na tive cri te rion.The dif fer ences in the four groups were not due to bi ases re lated to sex, or an thro pomet ric or smok ing char ac ter is tics, since these vari ables were not sig nifi cantly dif fer ent.
Most sub jects were as signed to group 1 (larg est FVC, FEV 1 and PEF in the same curve), which in cluded ATS cri teria for se lec tion; thus all the cri te ria were si mul ta ne ously pres ent in the same curve, em pha siz ing the use ful ness of 306 Can Respir J Vol 3 No 5 Sep tem ber/Oc to ber 1996

Di Pede et al
Fig ure 1) Preva lence rates of res pi ra tory symp toms and dis eases in four mu tu ally ex clu sive groups.Group 1: si mul ta ne ous larg est forced vi tal ca pac ity (FVC), forced ex pi ra tory vol ume in 1 s (FEV 1 ) and peak ex pi ra tory flow (PEF); Group 2: iso lated larg est FVC; Group 3: iso lated larg est FEV 1 ; Group 4: iso lated larg est PEF.All preva lence rates were higher in group 4, ex cept for at tacks of shortness of breath with wheeze (SOBWHZ) and dysp nea, which where higher in group 1. *P<0.05 by χ 2 test.COPD Chronic ob struc tive pul mo nary dis ease ATS cri te ria in se lect ing the best per formed ma noeu vre, ie, that with maxi mal ef fort.Nev er the less, ATS rec om men dations do not in clude PEF among the se lec tion cri te ria (1-3), and an FVC ma noeu vre ac com plished with sub maxi mal effort may be se lected as 'best' curve.
The meth ods pro posed by Quan jer et al (12) do not seem to im prove the se lec tion cri te ria for the 'best' FVC.The first (en ve lope method) is based on a theo reti cal FVC curve.As dem on strated by Bouhuys and Jon son (23) and con firmed by Knud son et al (20), one would ob tain an FVC curve with higher maxi mal flow val ues than those meas ured from any sin gle FVC curve ac tu ally ex haled by the sub ject.Thus, di agnos tic in ter pre ta tion would be based on a model, rather than the ac tual, FVC ma noeu vre.On the other hand, the sec ond method pro posed by Quan jer et al (12) uses three dif fer ent curves to se lect the high est in stan ta ne ous flows.This might mean that, for ex am ple, PEF is ob tained from one flowvolume curve and MEF50% from an other.Se lect ing in stanta ne ous flows in this way would lead to higher in tra-i nd ividual vari abil ity and an un der es ti ma tion of ab nor mal ity.In fact, many pa tients with se vere air flow ob struc tion yield better re sults for FEV 1 and midex pi ra tory flow rates when they make sub maxi mal ef forts, thereby avoid ing air way col lapse.Thus, the sen si tiv ity of an FVC ma noeu vre might be decreased were this method to be used.
Group 2 and group 3 cri te ria did not se lect curves that were rep re sen ta tive of the best FVC data.In group 2 (iso lated larg est FVC), FEV 1 and all the other de rived pa rame ters were lower com pared with the ATS-Σ curve, be cause the com pu ta tion points for mean and in stan ta ne ous ex pi ra tory flows were shifted to wards the right of the flow-volume curve, ie, to wards re sid ual vol ume.In these FVC tri als, the sub maxi mal ex pi ra tory ef fort, as dem on strated by a lower mean value in PEF, proba bly re duced the dy namic ex pired gas com pres sion with a con se quent de lay in pe riph eral airway clo sure (9,10).In group 3 (iso lated larg est FEV 1 ), the small est group, higher val ues for de rived flows were obtained than in the ATS-Σ curves.In these ma noeu vres, subjects ex pired from TLC, but they were not able to empty their lungs com pletely; thus, the high est value for FVC was not reached and the com pu ta tion points for mean and in stan ta neous ex pi ra tory flows were shifted to wards the left of the flow-volume curve.In ad di tion, sub maxi mal ex pi ra tory effort (in di cated by a lower mean value in PEF) was pres ent in this group.Thus, large er rors may re sult in the ex trapo lated vol ume and con se quently, higher mean val ues of FEV 1 , and mean and in stan ta ne ous ex pi ra tory flows are ob tained, if this cri te rion is ap plied.
The group 4 cri te rion, based on the larg est PEF, would meet the ATS defi ni tion of FVC (1-3), since the high est PEF rep re sents the maxi mal ef fort made to per form an FVC manoeu vre.This group con tained the low est FVC, FEV 1 and mean and in stan ta ne ous forced ex pi ra tory flows, in di cat ing the ef fects of the air ways com pres sion achieved by maxi mal ef fort and dis tin guish ing a forced from a slow vi tal ca pac ity.
Within groups 2, 3 and 4, FVC and de rived pa rame ters were sig nifi cantly dif fer ent be tween the ATS curve and the al ter na tive cri te rion.There fore, the clas si fi ca tion of sub jects as nor mal or ab nor mal may change, es pe cially if the per centage pre dicted value of the ob served pa rame ter was near the nor mal 95% per cen tile thresh old.In our study, the group 4 cri te rion showed the high est number of ab nor mali ties for FVC, FEV 1 , FEF 75-85% and MEF50%, while group 1 showed the high est number of ab nor mali ties for FEF 25-75% and MEF50%.
It is note wor thy that the high est sen si tiv ity for FEV 1 in detect ing symp to matic sub jects was pres ent in group 4 (27.9%),fol lowed by group 1 (23.0%).Fur ther, group 4 showed the high est speci fic ity for MEF50%, MEF75% and PEF.In deed, the high est pre dic tive value for de rived flows, in sepa rat ing symp to matic from as ymp to matic sub jects, was al ways shown by group 4, while the high est pre dic tive value for FVC and FEV 1 was ex hib ited by groups 3 and 1, re spectively, fol lowed by group 4.
The study de sign based on the four mu tu ally ex clu sive groups was adopted in or der to have data with out over lapping se lec tion cri te ria and, pos si bly, to group sub jects with simi lar res pi ra tory pa tho physio logi cal con di tions.Nev er theless, when analy ses were per formed on sub jects grouped into the four cate go ries con sti tuted by the whole sam ple of 1147 sub jects, analo gous re sults were ob tained.
Dif fer ences in sen si tiv ity among groups and cate go ries were not sta tis ti cally sig nifi cant by t test.How ever, in this gen eral popu la tion sam ple liv ing in a ru ral un pol luted area with a low preva lence of res pi ra tory symp toms (17), a clinically use ful trend to wards an in crease in sen si tiv ity in de tecting symp to matic sub jects was ob tained with the use of the larg est PEF cri te rion.In deed, poor sen si tiv ity and high specific ity are two rec og nized fea tures of the FVC test.For ex ample, Sten ton et al (24), dur ing a pro gram of asthma sur veillance in the work place, found a sen si tiv ity of 21% and a speci fic ity of 92% for any ab nor mal ven ti la tory func tion test.
Se lec tion meth ods for val ues of FEV 1 and FVC, based on the ma noeu vre with the larg est FVC, the larg est FEV 1 or the larg est PEF, were ana lyzed by Wise et al (25).These authors found that none of the as sessed se lec tion meth ods was substan tially su pe rior to any of the oth ers with re spect to short term re pro duci bil ity of FEV 1 and FVC (co ef fi cient of variation rang ing from 4.1% to 4.9% and from 3.5% to 5.7%, respec tively).These authors stated that the simi lar ity of the re sults at tested to the over all re pro duci bil ity of spi ro met ric meas ure ments when per formed with good tech nique.However, it should be noted that Wise et al (25) sug gested cau tion in ex trapo lat ing from this highly se lected group to a clini cal or gen eral popu la tion sam ple; also, they did not evalu ate the re la tion ship of spi ro met ric in dexes to the pres ence of res pi ratory symp toms.Con versely, we have shown in a gen eral popu la tion sam ple that the se lec tion of the spi ro gram to be used for in ter pre ta tive pur poses should be based also on its abil ity to dis tin guish sub jects with symp toms or dis ease from those with out.As dem on strated in Ta ble 5, the cri te rion of the larg est PEF in creases by about 5% the sen si tiv ity of FEV 1 in se lect ing sub jects with the pres ence of at least one res pi ratory symp tom com pared with the group 1 cri te rion.This im -plies that, as re gards the FEV 1 se lected by ap ply ing the group 4 cri te rion, a phy si cian might con sider start ing or modi fy ing medi cal treat ment in about 5% more sub jects with symptoms.
In a pre vi ous pa per, we re ported that if a com bi na tion of lung func tion in dexes were con sid ered, such as the forced spi ro grams, sin gle breath ni tro gen test and dif fus ing ca pacity, the per cent age of symp to matic male sub jects with any lung func tion ab nor mal ity in creased to 60% to 65% (18).In rou tine clini cal use, the FVC test is of ten ap plied to evalu ate lung func tion of sub jects with res pi ra tory symp toms.It is pos si ble that, if PEF were in cluded in the se lec tion cri te ria for the best FVC curve, the dis crimi na tive power of spi rograms would be fur ther en hanced.In deed, sub jects in the group se lected by the larg est PEF cri te rion tended to have the high est preva lence rates of res pi ra tory symp toms and disease.In con clu sion, our re sults show that, al though the cri te ria rec om mended in ATS state ments for FVC ac qui si tion have good screen ing power, this power might be im proved by inclu sion of PEF in the se lec tion cri te ria for mak ing the choice of which flow-volume curve to use in an in di vid ual.Thus, while main tain ing cur rent ATS cri te ria of ac cept abil ity and re pro duci bil ity of the FVC ma noeu vre, we pro pose also that the curve that best re flects maxi mal ex pi ra tory ef fort, ie, that with the larg est PEF, be re corded and ana lyzed, thereby improv ing early de tec tion of COPD, a con di tion whose mor bidity and mor tal ity are still in creas ing (26).

TABLE 1 Sample spirometric data for one individual from each of five mutually exclusive study groups
The cri te ria used to se lect groups were mu tu ally ex clu sive; to il lus trate this pro cess, in di vid ual data from three spi ro grams, ob tained in each of five sub jects as signed to the five dif fer ent groups, are shown in Ta ble 1.
ATS-Σ Ameri can Tho racic So ci ety (ATS) 'best test' curve (ie, the curve with the larg est re sult of forced vi tal ca pac ity [FVC] + forced expi ra tory vol ume in 1 s [FEV1]) for sub jects who had an al ter na tive crite rion for the se lec tion of the 'best' FVC ma noeu vre; PEF Peak ex pi ra tory flow.Un der lined val ues were used for the se lec tion of the group and of the ATS-Σ curve.Sub jects 1 to 4 were as signed to groups 1 to 4 ac cord ing to the hi er ar chi cally and mu tu ally ex clu sive pres ence of se lec tion cri te ria.Sub ject 5 showed the larg est val ues of FVC, FEV1 and PEF in three dif fer ent curves, and was thus in cluded in group 5 and ex cluded from the sub se quent analy ses groups (group 4); and more than one of the se lec tion cri te ria for groups 2, 3 and 4 si mul ta ne ously pres ent in the three curves (group 5).The as sign ment of sub jects to groups followed this pri or ity or der (from group 1 to group 4); thus, for ex am ple, a sub ject was in cluded in group 3 if the larg est FEV 1 and the larg est PEF were pres ent in the same curve.PEF (cate gory 4).Each cri te rion was ap plied to the whole sam ple of 1147 sub jects; thus, in this analy sis the four catego ries were not mu tu ally ex clu sive.Sta tis ti cal analy ses:Analy ses were per formed at the computer fa cili ties of the Uni ver sity of Pisa and the CNR Computer Center-CNUCE us ing the Sta tis ti cal Pack age for the So cial Sci ences (SPSS/PC + Up date, V3.0 and V3.1, SPSS Inc, Il li nois)

TABLE 2 Main characteristics of the five mutually exclusive study groups
There were no sta tis ti cally sig nifi cant dif fer ences among the mu tu ally ex clu sive groups for any an thro pomet ric mean value by ANOVA, nor for sex or smok ing habit by χ 2 test

TABLE 3 Flow-volume curve parameters in the first four study groups
ATS-Σ Ameri can Tho racic So ci ety (ATS) 'best test' curve (ie, the curve with the larg est re sult of forced vi tal ca pac ity [FVC] + forced ex pi ra tory volume in 1 s [FEV1]) for sub jects who had an al ter na tive cri te rion for the se lec tion of the 'best' FVC ma noeu vre; FEF25-75% Forced ex pi ra tory flow between 25% and 75% of FVC; FEF75-85% Forced ex pi ra tory flow be tween 75% and 85% of FVC; MEF50% Maxi mal ex pi ra tory flow at 50% of FVC; MEF75% Maxi mal ex pi ra tory flow at 75% of FVC; PEF Peak ex pi ra tory flow.>FVC, FEV1, PEF is de fined as the si mul ta ne ous larg est FVC, FEV1 and PEF cri te rion (in clud ing ATS cri te ria for se lec tion of the curve to be used for di ag nos tic pur poses); >FVC is de fined as iso lated larg est FVC crite rion; >FEV1 is de fined as iso lated larg est FEV1 cri te rion; >PEF is de fined as iso lated larg est PEF cri te rion.The high est value found for each parame ter in groups 2 to 4 is un der lined.*P<0.05;**P<0.01 by paired t test

TABLE 4 Prevalence (%) of spirometric abnormalities in the first four mutually exclusive study groups
85% Forced ex pi ra tory flow be tween 75% and 85% of FVC; MEF50% Maxi mal ex pi ra tory flow at 50% of FVC; MEF75% Maxi mal ex pi ra tory flow at 75% of FVC; PEF Peak ex pi ra tory flow.>FVC, FEV1, PEF is de fined as the si mul ta ne ous larg est FVC, FEV1 and PEF cri te rion (in clud ing ATS cri te ria for se lec tion of the curve to be used for di ag nos tic pur poses).The high est ab nor mal ity val ues shown by each pa rame ter in groups 1 to 4 are un der lined.Thresh old value is the nor mal 95th per cen tile

TABLE 6 Sensitivity (%), specificity (%) and predictive value (%) of FVC parameters for the presence of at least one respiratory symptom in the four categories, each composed of the whole sample of 1147 subjects
Cate gory 1 sig ni fies Ameri can Tho racic So ci ety cri te ria 'best test' curve; Cate gory 2 sig ni fies the curve with the larg est forced vi tal ca pac ity (FVC); Cate gory 3 sig ni fies the curve with the larg est forced ex pi ra tory vol ume in 1 s (FEV1); Cate gory 4 sig ni fies the curve with the larg est peak ex pi ratory flow (PEF).The high est val ues for sen si tiv ity (Sens), speci fic ity (Spec) and pre dic tive value (PV) reached by each pa rame ter in cate go ries 1 to 4 are un der lined.Sen si tiv ity dif fer ences were not sta tis ti cally dif fer ent by t test.FEF25-75% Forced ex pi ra tory flow be tween 25% and 75% of FVC; FEF75-85% Forced ex pi ra tory flow be tween 75% and 85% of FVC; MEF50% Maxi mal ex pi ra tory flow at 50% of FVC; MEF75% Maxi mal ex pi ra tory flow at 75% of FVC