Eur Respir J 1998; 12: 685 692 DOI: 1.1183/931936.98.123685 Printed in UK - ll rights reserved Copyright ERS Journls Ltd 1998 Europen Respirtory Journl ISSN 93-1936 Vlue of et-to-et lood pressure chnges, detected y pulse trnsit time, in the mngement of the ostructive sleep pnoe/ hypopnoe syndrome D.J. Pitson, J.R. Strdling Vlue of et-to-et lood pressure chnges, detected y pulse trnsit time, in the mngement of the ostructive sleep pnoe/hypopnoe syndrome. D.J. Pitson, J.R. Strdling. ERS Journls Ltd 1998. ABSTRACT: Two importnt spects of respirtory sleep study re mesure of inspirtory effort nd n estimte of the numer of rousls. These cn e derived from n indirect estimte of et-to-et lood pressure (BP), pulse trnsit time (PTT). This study investigted the reproduciility of inspirtory BP flls (reflecting inspirtory effort), nd BP rousls derived from PTT, nd the contriution they could mke to the mngement of the ostructive sleep pnoe/hypopnoe syndrome (OSAHS). Overnight PTT ws recorded t home in 4 ptients eing investigted for OSAHS, nd second PTT recording ws mde in the sleep lortory with full polysomnogrphy. Ptients were divided into three groups ccording to the severity of their sleep disorder, nd third PTT recording ws mde t home in 13 ptients susequently estlished on nsl continuous positive irwy pressure (CPAP). The reproduciility etween the home nd lortory studies ws resonle (r=.87 for inspirtory BP flls, r=.81 for BP rousls). Both derivtives showed cler progression through the three ptient groups, which returned to norml on tretment. The differences etween the groups were significnt (p<.1 for inspirtory BP flls, p=.14 for BP rousls). Receiver opertor chrcteristic curves, used to compre polysomnogrphy vriles nd PTT vriles, confirmed tht the PTT vriles were s good s pnoe-hypopnoe index (AHI), >4% rteril oxygen sturtion dip rte nd electroencephlogrphy micro-rousls t dividing ptients into two groups, either requiring nsl CPAP or not requiring CPAP. Pulse trnsit time cn provide noninvsive estimte of inspirtory effort nd mesure of rousls tht together document disese severity nd response to tretment nd my e useful in mnging ostructive sleep pnoe/hypopnoe syndrome. Eur Respir J 1998; 12: 685 692. Osler Chest Unit, The Churchill, Oxford Rdcliffe Hospitl, Oxford, UK. Correspondence: D.J. Pitson Osler Chest Unit The Churchill Oxford Rdcliffe Hospitl Oxford OX3 7LJ UK Fx: 44 1865 225221 Keywords: Ostructive sleep pnoe pulse trnsit time Received: Decemer 2 1997 Accepted fter revision April 27 1998 This work ws supported y grnt from the British Lung Foundtion. Dytime hypersomnolence is one of the clssicl fetures of the ostructive sleep pnoe/hypopnoe syndrome (OSAHS) [1] nd the symptom for which tretment is usully prescried [2]. Dytime sleepiness results from the mny trnsient rousls tht frgment sleep in OSAHS nd tht re elieved to e response to the increses in inspirtory effort eing mde in n ttempt to mintin dequte ventiltion through nrrowed or collpsed upper irwy [3, 4]. A respirtory sleep study to investigte OSAHS should therefore include either mesure of ventiltory effort or upper irwy nrrowing, nd mesure of rousls s mrker of the resulting sleep frgmenttion [5]. Becuse OSAHS is reltively common disorder ffecting 1 5% of the dult mle popultion [6], nd the finncil resources for helthcre re limited, such respirtory sleep studies should lso e cost-effective nd, preferly, domiciliry [7]. The rousl process tht occurs in response to lerting stimuli egins t the rin stem level with reflex increse in sympthetic ctivity cusing chnges in numer of mesurle vriles for exmple, lood pressure, hert rte nd skin vsoconstriction [8]. For minor lerting stimuli, this my e the only level of response lthough, for more significnt stimuli, the rousl process will rdite upwrds to involve the cortex cusing chnges in the electroencephlogrm (EEG) nd, eventully, full wkening from sleep. There is some dete s to wht level of ctivtion of the nervous system constitutes n "rousl", ut for the purposes of this pper, the term rousl is tken s generl term to reflect ctivtion of the nervous system t ny level in response to n lerting stimulus. Autonomic responses to lerting stimuli re termed "utonomic rousls" nd, more specificlly, the lood pressure (BP) rises mesured in this study re termed "BP rousls". Arousl responses involving corticl ctivtion re termed "corticl rousls", nd the rief EEG chnges mesured in this study re termed "EEG micro-rousls". The exct reltionship etween the recurrent rousls nd the consequent dytime sleepiness hs lso not yet een estlished, ut there is evidence to suggest tht rousls tht re detectle y utonomic chnges, ut where there is no EEG
686 D.J. PITSON, J.R. STRADLING chnge (i.e. no corticl rousl), my result in dytime sleepiness [9, 1]. Thus, mesuring utonomic rousls my e s importnt s mesuring corticl rousls. Systolic lood pressure (SBP) flls during inspirtion (pulsus prdoxus), nd the size of this inspirtory fll is proportionl to the degree of inspirtory effort [11]. There re lso rises in SBP with ech rousl from sleep tht occur even in response to extremely minor lerting stimuli tht produce no discernile chnge in the EEG [12]. DAVIES et l. [13] recorded et-to-et BP using the Finpres device nd developed computer lgorithms to identify inspirtory SBP flls nd rousl-relted rises in BP from the ll night dt. These lgorithms were susequently used to identify ptients with sleep relted rething disorders. The Finpres technology used y DAVIES et l. [13] hs n inflting cuff fitting round the finger. This is uncomfortle if worn for long periods nd the control unit is lrge; therefore it is less thn idel for routine overnight monitoring, nd is unsuitle for domiciliry sleep studies. Becuse of these limittions, we hve een investigting n indirect estimte of et-to-et BP, pulse trnsit time (PTT) [14], which cn e recorded nd stored y portle device. PTT is the trnsmission time of the rteril pulse pressure wve from the ortic vlve to the periphery nd is usully recorded s the time etween the R-wve on the electrocrdiogrm (ECG) nd the susequent rrivl of the corresponding pulse shock wve t the finger. This is typiclly round 25 ms nd is thus much fster thn the time tken for the lood ejected y the left ventricle to ctully rech the periphery (usully out 25 s). The principl determinnt of the speed with which the shock wve trvels is the degree of stiffness/tension in the r-teril wlls tht in turn is determined mostly y the BP. Thus, s BP increses, there is n increse in the stiffness/tension of the rteril wlls nd the pulse wve trvels fster (PTT decreses). Conversely, s BP flls, there is less stiffness/ tension in the rteril wlls, nd the pulse wve trvels more slowly (PTT increses). We hve shown tht the inspirtory BP flls cn e detected y increses in PTT [15, 16] nd tht the size of these increses re relted to the degree of inspirtory effort [17]. In ddition, the BP rise with rousl cn e detected y decrese in PTT [18], nd the numer of BP rises per hour correlte well with other indices of sleep frgmenttion [19]. We hve dpted the lgorithms of DAVIES et l. [13] to detect inspirtory BP flls (estimte of inspirtory effort) nd rousl-relted rises in BP (estimte of sleep frgmenttion) from n ll-night PTT recording. The im of this study ws to provide some evidence regrding the reproduciility of these two derivtives of the PTT recording nd to exmine the possile contriution tht they could mke in mnging the OSAHS. Mterils nd methods Study sujects Forty ptients who were referred to the Oxford Sleep Unit for investigtion of possile ostructive sleep pnoe were recruited to the study. Thirty-four were mle, nd six were femle. Their men ge ws 48 yrs (rnge 23 66 yrs), nd their men ody mss index ws 31 (rnge 24 5). The study ws pproved y the Centrl Oxford Reserch Ethics Committee (project numer 2792). Methods Pulse trnsit time. The PTT for ech hert et ws clculted, recorded nd stored y n RM1 recorder (Prmetric Recorders, London, UK). Three thorcic electrodes were used to smple the ECG, nd the rrivl of the pulse shock wve t the finger ws detected photometriclly y n oximeter finger proe. PTT is mesured in rel time s the intervl from the QRS complex to threshold on the rising edge of the finger pulse. The threshold is clculted s 25% of the height of the pulse during time window of 28 ms fter the QRS. The QRS is detected y smpling the ECG t 5 Hz nd the pulse is lso smpled t 5 Hz, giving n ccurcy of 2 ms. A new vlue for PTT is only ville with ech hert et, nd therefore, to ensure tht none is missed, PTT vlues re smpled t 5 Hz nd these re the vlues stored in the RM1. Polysomnogrphy. One chnnel of EEG (either C3/A2 or C4/A1), seprte electro-oculogrms (EOGs) for ech eye, nd sumentl electromyogrms (EMG) were recorded. Oronsl irflow ws mesured using thermistor; ricge nd dominl movements were mesured y inductnce plethysmogrphy from elts round the chest nd domen. These signls were recorded using n MPA-2 recorder (Oxford Medicl, Oxford, UK). Arteril oxygen sturtion (S,O 2 ) nd pulse rte were mesured y n Ohmed 37 pulse oximeter (Ohmed, Boulder, CA, USA) nd were recorded long with ody movements nd snoring y the Visi-L sleep monitoring system (Stowood Scientific Instruments, Oxford, UK). Clinicl dignosis. Sujects were clssified into three groups ccording to their clinicl dignosis. Routine mngement of these ptients involves n outptient ssessment history, Epworth Sleepiness Score (ESS) [2], exmintion, nd Visi-L sleep study. The sleep study is reviewed to ssess ostructive respirtory prolems (hypoxic dips nd snoring) nd sleep disturnce (ody movements nd hert rte oscilltions). The clinicl dignosis is then mde, sed on the integrtion of history, sleepiness nd sleep study findings. The three clinicl groups, reflecting their susequent mngement, were: 1) norml, virtully no snoring nd no sleep disturnce due to OSAHS; 2) mild/moderte, snoring, limited evidence of sleep disruption nd/or insufficient dytime sleepiness for the clinicin nd ptient to feel tril of nsl CPAP ws justified; nd 3) severe disorder, evidence of upper irwy ostruction (snoring, hypopnoes, pnoes, hypoxic dips) nd dytime symptoms (usully sleepiness) to justify tril of nsl continuous positive irwy pressure (CPAP). This clinicl lloction into mngement groups is used in our unit ecuse of the filure of ny prticulr sleep study derivtive (e.g. pnoe-hypopnoe index (AHI)) to ccurtely identify those ptients likely to enefit from nsl CPAP [2]. Tle 1 shows the medin nd upper/lower qurtiles of four severity indictors for the three groups, oth those ville t the time of dignosis (ESS, >4% S,O 2 dip rte), nd those used only for this study (AHI, EEG micro-rousls h -1 ).
OSA AND INDIRECT BEAT-TO-BEAT BP 687 Tle 1. Clinicl chrcteristics of ptients grouped ccording to the severity of their sleep disorder Clinicl group ESS medin (25 75%) S,O 2 dips h -1 medin (25 75%) AHI medin (25 75%) EEG micro-rousls h -1 medin (25 75%) Norml (n=11) Mild/moderte (n=11) Severe disorder (n=18) 1 (7 15) 9 (5 13) 16 (11 18) 1.3 (.9 2.5) 2.5 (1. 4.5) 17.9 (4.3 46.5) 4.5 (2.3 9.5) 8.5 (3.3 1.5) 32. (22. 53.) 19. (12. 31.) 2. (14. 28.) 37.5 (24.5 61.3) ESS: Epworth Sleepiness Score; AHI: pnoe/hypopnoe index; S,O 2 : rteril oxygen sturtion; EEG: electroencephlogrm. Protocol The study ws conducted in three stges. Initilly, recording of overnight PTT ws undertken in the ptient's own home. On seprte occsion, the ptient ws dmitted to the sleep unit where second overnight PTT recording ws mde in conjunction with full reserch polysomnogrphy sleep study nd Visi-l study. At this stge, ptients were lso grouped ccording to the severity of their sleep disorder into mngement groups s ove, without reference to the polysomnogrphy results. Finlly, those ptients who were commenced on nsl CPAP were restudied t home with further PTT recording enling us to exmine the chnge in the PTT derivtives fter tretment. Domiciliry recording. The home visits were undertken y one of the uthors (D.J. Pitson) who set up the RM1 recorder using lp-top PC (Commodore 286-LT, West Chester, PA, USA). The three thorcic electrodes (Unilect Ag/AgCI, MSB, Wiltshire, UK) were positioned to ensure cler positive R wve. The finger proe ws ttched to the index finger of the right hnd. The RM1 ws progrmmed to strt recording pproximtely 1 h efore the ptient's usul ed time nd stop recording out n hour fter their usul wke time. Once dequte signls hd een verified, the PC ws disconnected from the RM1, nd the RM1 ws disconnected from the ptient, leving just the three ECG electrodes in plce. The ptient ws instructed to reconnect the ECG leds nd finger proe when they went to ed nd to tpe over the leds with sticky tpe to try nd prevent ccidentl disconnection during the study. An instruction sheet reminding the ptient of wht they needed to do ws left together with form sking them to record how well they slept on threepoint scle: sme s norml; little less thn norml; or much less thn norml. Polysomnogrphy. Ptients were sked to ttend for their hospitl sleep study t 2: h. All the recording equipment ws ttched, nd the MPA-2, RM1 nd Visi-l sleep monitoring system were time-synchronized. Ptients were llowed to remove the monitoring equipment themselves when they woke up in the morning nd completed the sme questionnire to record how well they slept, s in the home study. Repet domiciliry recording. For those ptients in the severe group who were successfully estlished on nsl CPAP, second home RM1 study ws performed t lest 3 months fter CPAP tretment hd egun. This ws identicl to the first home study. Anlysis PTT dt. The ll-night PTT recording ws viewed, nd ny rtefctul periods t the strt of the recording (presumed to e while the ptient ws ttching to the equipment) nd t the end of the recording (presumed to e fter the ptient hd woken) were excluded. The intervening period ws utomticlly processed y the RM5 softwre version 1.51 (Prmetric Recorders, London, UK) for BP rousls nd inspirtory BP flls. If ny of the equipment ecme detched during the night, the ville time period ws nlysed. These sme time periods were used for nlysing the polysomnogrphy dt. Blood pressure rousls. The rw PTT trcing ws utomticlly processed y the RM5 softwre to remove rtefct. The differences etween djcent PTT vlues were computed, nd if they exceeded 5 ms (physiologiclly very unlikely), then the strt of n rtefct ws mrked. The softwre continued to record n rtefct until three consecutive differences were <5 ms. Missing dt were filled in y liner interpoltion for mximum of 1 smples (2 s of dt), fter which gp ws left in the record. Dt were next pssed through 17-smple (3.4 s) moving window verge to remove most of the reth-yreth oscilltions. Finlly, the dt were processed for BP rousls (fll in PTT of >15 ms lsting for >5 s nd <45 s) [19], nd the numer of rousls per hour of the PTT recording ws clculted. Inspirtory BP flls (inspirtory effort). The sme rtefct removl process ws pplied to the rw PTT dt except tht only five smples (1 s) of dt seprted y rtefct were interpolted. Dt were then pssed through three smple (.6 s) moving window verge. Finlly, the trcing ws processed for inspirtory BP flls tht were defined s n inspirtory chnge (rise in PTT) lsting for >.7 s nd <4.5 s. This definition ws dpted from the work of DAVIES et l. [13]. The progrm clculted the numer of reths detected, nd the men nd SD of the inspirtory PTT chnge (men BP fll) for the whole PTT record. Polysomnogrphy dt. Apnoe/hypopnoe index. Initil processing utilized the Oxford Medicl 92 reply system, which scored n pnoe when the mplitude of reth (oronsl irflow) ws reduced y 8% or more for t lest 1 s nd hypopnoe when the reth mplitude ws reduced y 5% for 1 s or more. The recording ws lter checked mnully to exclude ny pnoes or hypopnoes tht hd een scored during wkefulness. The numer of pnoes nd hypopnoes per hour were computed for the time period defined from the PTT record. Oxygen sturtion dip rte (S,O 2 dips). Dt processing utilized the Visi-L system (Stowood Scientific Instruments, Oxford, UK). The ll-night S,O 2 dt were stored every 12 s, using the lowest S,O 2 vlue occurring in the previous 12 s period of the recording (this is identicl to the dt tht would e stored nd downloded from the
688 D.J. PITSON, J.R. STRADLING Ohmed iox in trend output mode). The 12 s dt were then used to clculte the numer of S,O 2 dips of >4% per hour of the nlysis period. EEG rousls. The Medilog Rpide softwre (Oxford Medicl) ws used to disply the EEG, EOG's nd EMG which were scored y one of the uthors (D.J. Pitson) for rousls of 3 s or longer using the scoring rules pulished y the Americn Sleep Disorders Assocition (ASDA) [21]. The numer of EEG micro-rousls per hour of the nlysis period were computed. All mnul scoring ws done lind to the other sleep study dt nd to the ptient's clinicl dignosis. Sttistics. Exmintion of the dt nd the Shpiro-Wilk W test for non-normlity suggested tht most of the vriles were unlikely to e normlly distriuted. Non-prmetric methods were therefore used to exmine the dt. Arcus Pro-Stt (I.E. Buchn, West Lncshire, UK) ws used for ll sttistics except for the receiver opertor chrcteristics (ROC) curves which were otined using True Epistt (Epistt Services, Richrdson, TX, USA). Results The RM1 ws tolerted well y the ptients. PTT dt from two lortory studies nd one home study could not e nlysed ecuse multiple ectopic ets clerly ffected the PTT recording. The men period (±SD) of nlysle PTT dt ws 6.5±1.64 h for the home studies nd 6.1± 2.22 h in the sleep lortory. Reproduciility There ws resonle greement etween the home nd lortory studies oth for BP rousls (Spermn r=.81) nd for inspirtory BP flls (r=.87). The Blnd nd Altmn plots (fig. 1) suggest tht sujects tended to hve more BP rousls during the home study, men difference (lortory study - home study) -3.9±1.9 rousls per hour, nd lrger inspirtory BP flls, men difference -1.5±2.9 ms. Clinicl dignosis According to the clinicl dignosis, 11 ptients were found to e norml, 11 hd mild/moderte sleep disorder nd 18 hd severe sleep disorder. Thirteen of the severe group were susequently estlished on nsl CPAP. Of the remining five, three ptients tried CPAP ut did not feel tht the enefits outweighed the prolems of this therpy, nd two were unwilling to try CPAP nd pursued lterntive tretments. BP rousls nd inspirtory BP flls derived from PTT Figure 2 shows the dt for BP rousls nd inspirtory BP flls during the home study in the ptients grouped ccording to their clinicl dignosis nd in the group who were restudied hving een estlished on nsl CPAP. BP rousls h -1 (medin, 25 75% rnge) were 1 (7 17), 21 (12 32), 36 (13 63), nd 8 (7 13) for norml, mild/ moderte, severe nd on CPAP groups, respectively. Inspirtory BP flls were 9 (8 12), 11 (9 15), 18 (13 21) nd 8 (6 1) for the sme groups. Both BP rousls nd inspirtory BP flls therefore show cler progression from norml to severe tht returns to the norml level on tretment. The Kruskl-Wllis test confirmed tht there were significnt differences etween the clinicl groups (p=.14 for BP rousls, p<.1 for inspirtory BP flls). The multiple comprisons procedure confirmed tht for oth BP rousls nd inspirtory BP flls, there were no significnt differences etween the group clssified s norml nd the group restudied whilst on nsl CPAP, nd etween the normls nd those with mild/moderte disorder. Significnt differences were found etween the norml nd severe group, the mild/moderte group nd the group restudied on nsl CPAP, nd the severe nd on CPAP group. Inspirtory BP flls did differentite etween the mild/ moderte group nd the severe group, ut BP rousls did not (tle 2). Comprison etween informtion derived from PTT nd polysomnogrphy dt Dt from the lortory sleep study enled us to exmine whether there were ny differences etween the ility ) Difference (l-home) ) Difference (l-home) 3 2 1-1 -2-3 -4-5 6 4 2-2 -4-6 -8-1 -12 2 4 6 8 1 12 Men BP rousl h -1 4 8 12 16 2 24 28 32 36 4 44 Men inspirtory BP fll (PTT ms) Fig. 1. Agreement etween the home nd lortory (l) studies for the two derivtives of pulse trnsit time (PTT). ) Blood pressure (BP) rousls, nd ) inspirtory BP flls.
OSA AND INDIRECT BEAT-TO-BEAT BP 689 ) 12 BP rousls h -1 ) Men inspirtory BP fll PTT ms 1 8 6 4 2 5 4 3 2 1 Norml Mild/mod Severe Norml Mild/mod Severe On CPAP On CPAP Fig. 2. Box nd whisker plots for the two pulse trnsit time (PTT) derivtives; ) lood pressure (BP) rousls, nd ) inspirtory BP flls in ptients grouped ccording to the severity of their sleep disorder nd in 13 of the severe group who were restudied whilst using nsl continuous positive irwy pressure (CPAP). The lower oundry of the ox indictes the 25th percentile, the line within the ox indictes the medin nd the upper oundry of the ox, the 75th percentile. The 1th nd 9th percentiles re indicted y the error rs elow nd ove the ox, nd vlues outside the 1th nd 9th percentile re shown s individul points. of the PTT derived vriles nd other commonly mesured sleep study vriles (AHI, EEG micro-rousls, S,O 2 dips) to differentite etween the clinicl groups. This ws initilly exmined y the Kruskl-Wllis test. For ll vriles, significnt differences were found etween the groups, S,O 2 dips p<.1, AHI p=.2, inspirtory BP flls p=.33, EEG micro-rousls p=.37, nd BP rousls p=.25. Tle 3 shows the results of the multiple comprisons procedure tht confirmed tht ll vriles could discriminte etween the norml nd severe group, nd ll except BP rousls could discriminte etween the Tle 2. Kruskl-Wllis nlysis for pulse trnsit time vriles (home studies) Clinicl group BP rousls h -1 Inspirtory BP flls On CPAP Norml Mild/moderte Severe disorder Clinicl groups with the sme letter re not significntly different. Clinicl groups with different letters re significntly different (p<.5). BP: lood pressure; CPAP: continuous positive irwy pressure. c c c Tle 3. Kruskl-Wllis nlysis for ll vriles (lortory studies) Clinicl group Norml Mild/ moderte Severe disorder Inspirtory BP flls EEG microrousls h AHI S,O 2 BP -1 dips h -1 rousls h -1 Clinicl groups with the sme letter re not significntly different. Clinicl groups with different letters re significntly different (p<.5). BP: lood pressure; EEG: electroencephlogrm; AHI: pnoe/hypopnoe index; S,O 2 : rteril oxygen sturtion. mild/moderte nd severe group. None of the vriles could discriminte etween the normls nd mild/moderte group. A further comprison etween the polysomnogrphy nd PTT vriles ws mde y plotting ROC curves for ech vrile. In order to do this, the norml nd mild/moderte groups were comined, resulting in two groups: ptients offered tril of nsl CPAP (n=18) nd ptients not offered CPAP (n=22). Anlysis of the re under ech ROC curve reveled tht the respirtory vriles (AHI nd S,O 2 dips) hd the highest res under the curve followed y inspirtory BP flls, EEG micro-rousls nd BP rousls. All vriles were, however, highly sttisticlly etter thn chnce t differentiting etween the two groups (fig. 3). Discussion The im of this study ws to exmine whether inspirtory effort nd BP rousls, mesured indirectly from PTT, could contriute to the mngement of the OSAHS. The possile dvntges of this technique re tht it includes noninvsive mesure of oth inspirtory effort nd sleep frgmenttion, using fully portle technology. In ddition, since the mesurement of PTT only requires ECG electrodes nd n oximeter finger proe, nd oth of these re lredy used in other sleep monitoring systems, it is reltively simple to dd PTT to existing systems. The equipment ws esy to set up nd use in the home, nd ptients tolerted it well. The dt qulity ws good, with 93% of studies hving nlysle dt. Our results confirmed tht there ws resonle reproduciility of the two PTT derivtives over the two study nights, which ws similr to tht found y others [22]. When ptients were sudivided ccording to the severity of their sleep disorder, oth derivtives clerly reflected the increse in severity cross the groups nd showed return to norml in ptients who were restudied on tretment. When compred with some of the commonly mesured polysomnogrphy vriles, the PTT-derived inspirtory effort ws s good s EEG rousls, AHI nd S,O 2 dips t discriminting etween the three ptient groups, nd oth PTT vriles were comprle with the polysomnogrphy vriles when dividing the ptients into the two mngement groups (offered tril of nsl CPAP or not offered tril of CPAP). Only two ptients' PTT dt could not e nlysed. In oth cses, multiple ectopic ets, occurring t frequency of etween 7 nd 1 min -1, clerly ffected the PTT. These resulted in sudden increses in PTT tht either led to rtefct or, if they were less thn the 5 ms threshold
69 D.J. PITSON, J.R. STRADLING ) 1..8.6.4.2 AUC=.92 p<.1.2.4.6.8 1. ) 1..8.6.4.2 AUC=.91 p<.1.2.4.6.8 1. c) 1..8.6.4.2 AUC=.85 p<.1.2.4.6.8 1. d) 1..8.6.4.2 AUC=.81 p=.6.2.4.6.8 1. e) 1..8.6.4.2 AUC=.73 p=.5.2.4.6.8 1. Fig. 3. Receiver opertor chrcteristic (ROC) curves of the pulse trnsit time derivtives nd polysomnogrphy vriles in detecting ptients who were recommended tril of nsl continuous positive irwy pressure (CPAP). ) pnoe/hypopnoe index, ) rteril oxygen sturtion (dips h -1 ), c) inspirtory lood pressure (BP) flls, d) electroencephlogrm micro-rousls h -1, nd e) BP rousls h -1. : proportion of ptients recommended to try nsl CPAP correctly identified; : proportion of ptients who were not recommended to try nsl CPAP flsely identified. The re under the curve (AUC) cn vry from.5 (vrile is no etter thn chnce) to 1. (vrile is s good s clinicl dignosis). required for rtefct rejection, flsely long PTT. Becuse the ectopics were occurring within the respirtory frequency, this would hve produced high vlue for the verge inspirtory BP fll if the studies hd een nlysed. So fr s the ssessment of true reproduciility is concerned, this study is, of course, potentilly flwed y the fct tht the two recordings were undertken under different environmentl conditions. The first took plce in the ptient's home nd the second in the sleep lortory with the ddition of further monitoring equipment. Both of the PTT derivtives suggested tht the ptient exhiited more normlities on the home study s the men difference (lortory study - home study) ws negtive for oth. Agreement ws etter for inspirtory effort (men difference -1.5±2.9 ms) thn for BP rousls (men difference -3.9±1.9 rousls h -1 ). A possile explntion for the higher numer of rousls in the home study might e tht the sujects slept longer thn in the lortory with less periods of wkefulness during the night. This grees with the questionnire tht ptients completed out their sleep. Ptients were sked to rte how well they slept on ech night compred with norml. Sme s norml ws scored
OSA AND INDIRECT BEAT-TO-BEAT BP 691 s 1, little less thn norml s 2 nd much less thn norml s 3. The men scores for home nd lortory studies were 1.6±.7 nd 2.1±.7, respectively, difference tht ws significnt (p<.1). The lrger vlues for inspirtory effort would lso suggest tht sleep t home ws more norml, llowing longer periods of ostructed rething. Thus, we conclude tht ptients proly slept etter with the portle equipment t home thn they did during polysomnogrphy sleep study in the sleep lortory, nd tht the results my therefore e more representtive of their norml sleep. The numer of BP rousls per hour nd the size of the inspirtory BP flls (degree of inspirtory effort ) monitored t home clerly incresed with incresing disese severity (fig. 2). The one cler outlier in the norml group for BP rousls ws suject who ws norml from the respirtory point of view ut who hd sleep disturnce from cuse tht could not e identified. Both BP rousls nd inspirtory BP flls lso clerly documented the response to nsl CPAP tretment. The outlier who remins in the nsl CPAP group is ptient with severe ostructive sleep pnoe whose PTT derivtives suggest tht their CPAP pressure is sutherpeutic. Despite this, the ptient ws considerly improved y CPAP (ESS efore tretment 24, on tretment 4) nd ws unwilling to tolerte higher pressure s it led to no further reported improvement in well-eing. Anlysis of the lortory sleep study dt reveled tht the respirtory vriles AHI nd S,O 2 dips were the strongest discrimintors etween the normls nd severe group nd etween the mild/moderte nd severe group. None of the vriles could distinguish etween the norml nd mild/moderte group. The results for the PTT vriles were consistent with the home studies with oth le to differentite etween norml nd severe, ut only inspirtory BP flls showing seprtion etween mild/moderte nd severe. Once the ptient group ws divided into two, either offered tril of nsl CPAP or not offered CPAP, ll vriles were significntly etter thn chnce t detecting ptients who cliniclly were felt to require tril of nsl CPAP. Once gin, the respirtory vriles performed est, with res under the curve >.9. Inspirtory BP flls performed mrginlly etter thn EEG microrousls (re under the curve=.85 compred with.81) with BP rousls performing lest well (re under the curve =.73). The min limittion of ny study looking t new dignostic techniques in ostructive sleep pnoe is the sence of gold stndrd definitions of the disese. Originl definitions sed on pnoe, hypopnoes, hypoxic dips or micro-rousls hve not een shown to identify ccurtely those ptients with sleep frgmenttion syndrome due to upper irwy ostruction likely to respond to nsl CPAP. For exmple, GUILLEMINAULT et l. [23] identified ptients with miniml pnoes, hypopnoes, hypoxic events nd ovious rousls (upper-irwy resistnce syndrome) who clerly hd hypersomnolence, incresed upper-irwy nrrowing with inspirtory effort, nd responded symptomticlly to nsl CPAP. Thus, in our clinicl grding of ptients, there were some in the severe group who hd reltively low >4% S,O 2 dip rtes nd AHI vlues, yet were very sleepy nd responded well to nsl CPAP. Although PIETERS et l. [24] hve found wek correltion etween CPAP use nd movement rousl index, in our unit t present, lloction of ptients into the three ctegories is very much clinicl decision, sed on history, degree of hypersomnolence nd sleep study designed to identify upper irwy nrrowing tht leds to sleep frgmenttion. This study ws conceived to estlish whether mesurements of PTT-derived inspirtory effort nd sleep frgmenttion were s good s other, more conventionl, indices t differentiting those put forwrd for tril of nsl CPAP from those not. Clerly >4% S,O 2 dip rte is likely to e very predictive, since this index ws used s prt of the dignostic process, long with the other indices ville from the Visi-L sleep study (hert rte oscilltions, snoring nd ody movements), the ESS nd the history. The PTT-derived vriles were s good s the other polysomnogrphy vriles t dividing ptients into these two groups ccording to nsl CPAP requirement, lthough when ptients were divided into three groups, BP rousl numers were unle to differentite etween the mild/ moderte nd severe groups wheres ll the other vriles were. If indirect BP monitoring ws to form prt of portle sleep monitoring system, it would e unlikely tht EEG would e recorded, nd therefore no informtion out true sleep onset nd sleep end would e ville. For this reson, "sensile" strt nd stop time for the nlysis of the ll night PTT dt ws estlished y excluding ny oviously rtefctul dt t the strt nd end of the study if they existed nd nlysing the remining dt. In order to e le to compre the two PTT recordings (home nd lortory), the sme pproch ws dopted for the lortory recording, nd this time period ws pplied to the nlyses of EEG, AHI nd S,O 2 dips. In summry, recording overnight PTT in the home ws esy to set up, well tolerted y the ptients nd produced relile nd resonly reproducile results from its two derivtives, BP rousls nd inspirtory BP flls (inspirtory effort). These two derivtives reflected the increse in disese severity cross three cliniclly defined ptient groups nd the improvement on tretment nd were similr to EEG micro-rousls, AHI nd S,O 2 dips in their ility to differentite etween the ptient groups. We conclude tht the mesurement of indirect et-toet lood pressure y using pulse trnsit time my e useful ddition to portle monitoring system for the mngement of the ostructive sleep pnoe/hypopnoe syndrome. References 1. Peter JH, Koehler U, Grote L, Podszus T. Mnifesttions nd consequences of ostructive sleep pnoe. Eur Respir J 1995; 8: 1572 1583. 2. Bennett LS, Strdling JR. Who should receive tretment for sleep pnoe. Thorx 1997; 52: 13 14. 3. Gleeson K, Zwillich CW, White DP. The influence of incresing ventiltory effort on rousl from sleep. Am Rev Respir Dis 199; 142: 295 3. 4. Gleeson K, Zwillich CW. Adenosine stimultion, ventiltion, nd rousl from sleep. Am Rev Respir Dis 1992; 145: 453 457. 5. Strdling JR. Sleep studies for sleep-relted rething disorders. A consensus report. J Sleep Res 1992; 1: 223 23.
692 D.J. PITSON, J.R. STRADLING 6. Dvies RJO, Strdling JR. The epidemiology of sleep pnoe. Thorx 1996; 51: Suppl 2, S65 S7. 7. Dougls NJ. Sleep-relted rething disorders: 3 How to rech dignosis in ptients who my hve the sleep pnoe/hypopnoe syndrome. Thorx 1995; 5: 883 886. 8. Hilton SM. The defence-rousl system nd its relevnce for circultory nd respirtory control. J Exp Biol 1982; 1: 159 174. 9. Mrtin SE, Drery IJ, Dougls NJ. The effect of utonomic rousls on dytime function. [Astrct]. Am J Respir Crit Cre Med 1996; 153(2): 354. 1. Mrtin SE, Writh PK, Dery IJ, Dougls NJ. The effect of nonvisile sleep frgmenttion on dytime function. Am J Respir Crit Cre Med 1997; 155: 1596 161. 11. Le S, Ali NJ, Goldmn M, et l. Systolic lood pressure swings reflect inspirtory effort during simulted ostructive sleep pnoe. In: Horne J, ed. Sleep 9. Bochum, Pontengel Press, 199; pp. 178 181. 12. Dvies RJO, Belt PJ, Roert SJ, Ali NJ, Strdling JR. Arteril lood pressure responses to grded trnsient rousl from sleep in norml humns. J Appl Physiol 1993; 74: 1123 113. 13. Dvies RJO, Vrdi-Visy K, Clrke M, Strdling JR. Identifiction of sleep disruption nd sleep disordered rething from the systolic lood pressure profile. Thorx 1993; 48: 1242 1247. 14. Griin B, Steptoe A, Sleight P. Pulse wve velocity s mesure of lood pressure chnge. Psychophysiology 1976; 13: 86 9. 15. Pitson DJ, Chhin N, Knijn S, vn Herwden M, Strdling JR. Mechnism of pulse trnsit time lengthening during ostructed inspirtion. J Am Mon 1995; 8: 11 15. 16. Brock J, Pitson D, Strdling JR. Use of pulse trnsit time s mesure of chnges in inspirtory effort. J Am Mon 1993; 6: 295 32. 17. Pitson DJ, Sndell A, vn den Hout R, Strdling JR. Use of pulse trnsit time s mesure of inspirtory effort in ptients with ostructive sleep pnoe. Eur Respir J 1995; 8: 1669 1674. 18. Pitson D, Chhin N, Knijn S, vn Herwden M, Strdling J. Chnges in pulse trnsit time nd pulse rte s mrkers of rousl from sleep in norml sujects. Clin Sci Colch 1994; 87: 269 273. 19. Pitson DJ, Strdling JR. Autonomic mrkers of rousl during sleep in ptients undergoing investigtion for ostructive sleep pnoe, their reltionship to EEG rousls, respirtory events nd sujective sleepiness. J Sleep Res 1998; 7: 53 59. 2. Johns MW. A new method for mesuring dytime sleepiness: the Epworth sleepiness scle. Sleep 1991; 14: 54 545. 21. Bonnet M, Crley D, Crskdon M, et l. EEG rousls: scoring rules nd exmples. A preliminry report from the Sleep Disorders Atls Tsk Force of the Americn Sleep Disorders Assocition. Sleep 1992; 15(2): 173 184. 22. Flemons WW, Remmers JE. The dignosis of sleep pne: questionnires nd home studies. Sleep 1996; 19: S243 S247. 23. Guilleminult C, Stoohs R, Clerk A, Cetel M, Mistros P. A cuse of excessive dytime sleepiness. The upper irwy resistnce syndrome. Chest 1993; 14: 781 787. 24. Pieters T, Collrd P, Auert G, Dury M, Delguste P, Rodenstein DO. Acceptnce nd long-term complince with ncpap in ptients with ostructive sleep pnoe syndrome. Eur Respir J 1996; 9: 939 944.