The impact of ventilation rate on end-tidal carbon dioxide level during manual cardiopulmonary resuscitation

The impac o en ila ion a e on end- idal ca bon dioxide le el du ing
manual ca diopulmona y esusci a ion
So ´ıa Ruiz de Gaunaa⇤, Jose Julio Gu i´e eza,JesusRuiz
a, Mikel Le u iondoa, Izaskun Azca a ea,
Digna Ma ´ıa Gonz´alez-O e oa,b, Ca los Co cue ac, James Knox Russelld, Mohamud Ramzan
Dayad
A ilia ion and add esses:
aUni e si y o he Basque Coun y, UPV/EHU. Bilbao, Bizkaia, Spain
bBexen Ca dio, E mua, Bizkaia, Spain
cEme gen ziak-Osakide za, Basque Coun y Heal h Sys em, Basque Coun y, Spain
dO egon Heal h & Science Uni e si y (OHSU). Po land, OR, USA
Co esponding au ho :
⇤
So ´ıa Ruiz de Gauna
email: so ia. [email protected]
Wo d coun s:
Abs ac : 237 wo ds
Manusc ip : 2859 wo ds
This is he accep ed manusc ip o he a icle ha appea ed in inal o m in Resusci a ion 156 : 215-222 (2020), which has been
published in inal o m a h ps://doi.o g/10.1016/j. esusci a ion.2020.06.007. © 2020 Else ie unde CC BY-NC-ND license (h p://
c ea i ecommons.o g/licenses/by-nc-nd/4.0/)
Abs ac
Aim: Ven ila ion a e is a con ounding ac o o in e p e a ion o end- idal ca bon dioxide
(ETCO2) du ing ca diopulmona y esusci a ion (CPR). The aim o ou s udy was o o model he
e↵ec o en ila ion a e on ETCO2du ing manual CPR in adul ou -o -hospi al ca diac a es
(OHCA).
Me hods: We conduc ed a e ospec i e analysis o OHCA moni o -de ib illa o iles wi h
concu en capnog am, comp ession dep h, ans ho acic impedance and ECG. We anno a ed
pai s o capnog am segmen s p esen ing di↵e ences in a e age en ila ion a e and a e age ETCO2
alue bu wi h o he in luencing ac o s (e.g. comp ession a e and dep h) p esen ing simila
alues wi hin he pai . ETCO2 a ia ion as a unc ion o en ila ion a e was adjus ed h ough
cu e i ing using non-linea leas squa es as a measu e o goodness o i .
Resul s: A o al o 141 pai s o segmen s om 102 pa ien s we e anno a ed. Each pai p o ided
a single da a poin o cu e i ing. The bes goodness o i yielded a coe icien o de e mina ion
R2o 0.93. Ou model desc ibed ha ETCO2decays exponen ially wi h inc easing en ila ion
a e. The model showed no di↵e ences a ibu able o he ai way ype (endo acheal ube o
sup aglo ic King-LT-D).
Conclusion: Capnog am in e p e a ion du ing CPR is challenging since many ac o s in luence
ETCO2. Fo adequa e in e p e a ion, we need o know he e↵ec o each ac o on ETCO2.Ou
model allows quan i ying he e↵ec o en ila ion a e on ETCO2 a ia ion. Ou indings could
con ibu e o be e in e p e a ion o ETCO2du ing CPR.
Key wo ds: Ca diopulmona y esusci a ion (CPR); Wa e o m capnog aphy; End- idal ca bon
dioxide (ETCO2); Ven ila ion; Ven ila ion a e; Ou -o -hospi al ca diac a es (OHCA);
Ad anced li e suppo (ALS).
1. In oduc ion1
High quali y ca diopulmona y esusci a ion (CPR) imp o es ou comes o ca diac a es 2
ic ims.1–4Ideally, CPR quali y should be guided based on he pa ien ’s esponse using3
a po en ially non-in asi e haemodynamic indica o .5,6Cu en ad anced li e suppo (ALS)4
esusci a ion guidelines emphasize he use o wa e o m capnog aphy du ing CPR.7,8Wa e o m5
capnog aphy enables moni o ing o end- idal ca bon dioxide (ETCO2), he pa ial p essu e o 6
ca bon dioxide a he end o expi a ion. ETCO2is an indi ec measu e o bo h ca diac ou pu 7
and pulmona y blood low.9–11 The e olu ion o ETCO2du ing he cou se o esusci a ion has8
been ound o be aluable in moni o ing he quali y o ches comp essions,12,13 allowing o 9
ea ly ecogni ion o e u n o spon aneous ci cula ion (ROSC),14–16 and as a p edic o o pa ien 10
ou come.17–20
11
Du ing CPR he alue o ETCO2depends on he blood low gene a ed by ches comp essions, on12
he olume and a e o en ila ion and on he pa ien ’s issues me abolic ac i i y.21,22 Imp o ing13
he quali y o ches comp essions inc eases he amoun o pulmona y blood low and he e o e14
ETCO2le el, p o ided issue me abolism is ac i e. Howe e , i en ila ion olume o en ila ion15
a e inc ease du ing he cou se o esusci a ion he ETCO2le el dec eases p o ided he blood low16
emains cons an .23,24
17
The e a e ew s udies aimed a quan i ying he ela ionship be ween CPR quali y componen s18
and ETCO2du ing esusci a ion. Recen obse a ional s udies wi h human da a used mul i a ia e19
analysis o quan i y ETCO2 a ia ion in ela ion o a ia ions in ches comp ession dep h and20
a e, and en ila ion a e.25,26 An animal s udy demons a ed ha changes in en ila ion a e21
signi ican ly in luence ETCO2le el.24 The aim o ou s udy was o iden i y a quan i a i e22
ela ionship be ween ETCO2and en ila ion a e h ough e ospec i e analysis o adul 23
ou -o -hospi al ca diac a es (OHCA) episodes.24
2. Ma e ials and me hods25
2.1. Da a collec ion26
Da a we e ex ac ed om a la ge da abase o adul ou -o -hospi al ca diac a es (OHCA)27
episodes collec ed om 2006 h ough 2017 by Tuala in Valley Fi e & Rescue (TVF&R), an ALS i s 28
esponse eme gency medical se ices (EMS) agency se ing ele en inco po a ed ci ies in O egon,29
USA. The da abase is pa o he Resusci a ion Ou comes Conso ium (ROC) Epidemiological30
Ca diac A es Regis y collec ed by he Po land Regional Clinical Cen e . The da a collec ion31
was app o ed by he O egon Heal h & Science Uni e si y (OHSU) Ins i u ional Re iew Boa d32
(IRB00001736). No pa ien p i a e in o ma ion was included in he da abase.33
Episodes we e eco ded using Hea s a MRx moni o -de ib illa o s (Philips, USA), equipped34
wi h capnog aphy moni o ing using sides eam echnology (Mic os eamTM, O idion Sys ems35
L d., Is ael) and CPR quali y moni o ing (Q-CPRTM). TVF&R ield p o ide s used ei he an36
endo acheal ube (ETT) o a sup aglo ic King LT-DTM de ice, o secu e he ai way. Con inuous37
ches comp essions and en ila ions we e p o ided manually. No en ila ion olume da a was38
colec ed. Fo his s udy, we only included eco dings wi h concu en capnog am, comp ession39
dep h signal, elec oca diog am (ECG) and ans ho acic impedance (TTI) signals. We did no 40
conside a es ae iology in selec ing cases.41
2.2. Selec ion o pai ed segmen s42
The hypo hesis o he s udy was ha i is possible o assess he ETCO2 a ia ion wi h43
espec o en ila ion a e om analysis o pai s o capnog am segmen s ha di↵e only in he44
applied en ila ion a e. We hypo hesised ha his a ia ion in en ila ion a e is esponsible45
o he di↵e ences in ETCO2le el be ween segmen s, when o he in luencing ac o s (me abolism,46
comp ession dep h and comp ession a e) a e con olled h ough a ca e ul de ini ion o he inclusion47
c i e ia.48
Using a cus om-made Ma lab p og am (Ma hwo ks, USA), wo biomedical expe s (JJG and49
JMR) isually inspec ed he ou signals ex ac ed om he OHCA de ib illa o eco dings. They50
iden i ied pai s o segmen s o capnog aphy signal di↵e ing in en ila ion a e and ETCO2 alues51
bu wi h he o he con ounding ac o s p esen ing simila beha iou in bo h segmen s. A eliable52
capnog am was equi ed o iden i y indi idual en ila ions (wi h he suppo o he TTI signal)27
53
and o anno a e a single ep esen a i e en ila ion a e and ETCO2 alue o each segmen in he54
pai . To p o ide ha he obse ed ETCO2 a ia ion was caused only by he change in en ila ion55
a e, he ollowing inclusion c i e ia we e es ablished o each iden i ied pai :56
•Absence o a pulse-gene a ing hy hm con i med by inspec ion o he ECG and e iew o 57
ROSC anno a ions eco ded by he ALS p o ide s in case no es. In cases whe e he e was58
doub abou he p esence o a pe using hy hm and pulseless elec ical ac i i y, we e alua ed59
he TTI signal o e idence o a pe using hy hm.28
60
•Simila me abolic ac i i y so ha a maximum o 2-min sepa a ion be ween segmen s was61
allowed.62
•Simila ches comp ession dep h and a e be ween he wo segmen s so ha a simila amoun 63
o pulmona y blood low could be assumed. Only segmen s wi h di↵e ences o less han 4 mm64
in mean comp ession dep h and 8 comp essions pe minu e (cpm) in mean comp ession a e65
we e included.66
Figu e 1 shows wo examples o segmen selec ion (panels A and B). Capnog am and67
comp ession dep h signals a e depic ed o wo o he pai s included in he s udy. Anno a ed68
alues o each segmen we e: he s a and end o each segmen , in seconds ( s1, e1; s2, e2)69
which a e depic ed in he igu e wi h e ical ed lines; he numbe o en ila ions wi hin each70
segmen (n 1, n 2); he mean ETCO2 alue pe segmen (ET1, ET2); he mean comp ession71
dep h and mean comp ession a e pe segmen (dc1, c1; dc2, c2), and he ime in e al be ween72
segmen s ( bs). The elapsed ime be ween segmen s was used o check i he sepa a ion be ween73
segmen s was below 2 min, and he mean comp ession dep h and mean comp ession a e o each74
segmen o he pai (dc1, c1; dc2, c2) we e used o check i he c i e ia o simila comp ession75
dep h and a e we e me . Once a pai me all he inclusion c i e ia i was cha ac e ised by h ee76
ea u es:77
•Ven ila ion a e o segmen 1: 1 = 60 ·n 1/( e1 s1) pm.78
•Ven ila ion a e o segmen 2: 2 = 60 ·n 2/( e2 s2) pm.79
•ETCO2 a io be ween segmen s: ET2/ET1.80
2.3. Model i ing81
The expe imen al alues we e adjus ed h ough he analy ic exp ession:82
ET2
ET1=1K 1
1K 2,(1)

whe e he coe icien Kis he only pa ame e de e minan o he adjus men . This83
ma hema ical model was p oposed in a p e ious wo k as an hypo hesis o explaining he in luence84
o en ila ion a e in ETCO2du ing CPR.29
85
To acili a e physical in e p e a ion o he esul s, we no malised Equa ion 1 wi h espec o86
he e e ence en ila ion a e o 1 = 10 pm ( he a e ecommended by cu en esusci a ion87
guidelines o ALS7,8). Thus, he no malised model used o he da a adjus men was:88
ET2
ET1



10
=ET2
ET1·1K10
1K 1=1K10
1K 2(2)
Equa ion 2 p o ides a amily o decaying cu es as en ila ion a e 2 inc eases (a single cu e89
o each alue o K). Figu e 2 depic s he amily o cu es esul ing om Equa ion 2 o 290
a ying om 3 o 35 pm and K om 0.75 o 0.99 in 0.03 inc emen s. The adjus men o K91
allows modelling he decay a e. The model esponds o he expec ed beha iou o ETCO2when92
en ila ion a e inc eases.93
The expe imen al da a we e no malised acco dingly: o each pai o segmen s we ob ained one94
da a poin o he adjus men cha ac e ised by a alue o 2 and a no malised alue o he a io95
ET2/ET1.96
2.4. S a is ical analysis97
Values we e epo ed as median (in e qua ile ange, IQR). The model was i ed using he98
weigh ed bi-squa ed me hod o obus ness. This me hod minimised a weigh ed sum o he squa ed99
esiduals, inding a cu e ha i s he bulk o he da a using he leas -squa es app oach and,100
simul aneously, minimising he e↵ec o he ou lie s.30 Goodness o i o he model was e alua ed101
using he coe icien o de e mina ion R2, which p o ides a measu e o he ET2/ET 1 a ia ion102
ha is explained by he model.103
We also analysed di↵e ences in he model i wi h espec o he ai way managemen echnique,104
ETT o sup aglo ic King LT-D.105
3. Resul s106
Concu en signals o in e es (capnog am, comp ession dep h, ECG and TTI signals) we e107
a ailable in 928 adul OHCA episodes (see Figu e 3). Episodes wi h capnog ams subs an ially108
dis o ed by ches comp ession a e ac 27,31 o poo signal quali y we e disca ded (n= 88) because109
in hose condi ions en ila ion a e and ETCO2 alues could no be accu a ely measu ed. Episodes110
wi h less han 250 s o concu en signals we e excluded (n= 127) since hey co esponded o e y111
sho esusci a i e a emp s o la e ad anced ai way placemen . We also excluded episodes whe e112
ETCO2 alues we e lowe han 10 mmHg mos o he ime (n= 71) because small inaccu acies in113
he manual measu emen o ETCO2 could cause signi ican pe cen e o s dis o ing he inpu da a114
o ou model. Finally, episodes whe e he capnog am du a ion be o e any ROSC anno a ion was115
less han 150 s we e also excluded (n= 134). Thus, we a oided selec ing segmen pai s nea ROSC116
whe e he p esence o pulse could a↵ec he ETCO2 le el. A o al o 508 adul OHCA episodes117
we e included in he s udy wi h a o al capnog am du a ion o 12,819 min.118
A o al o 141 pai s o segmen s om 102 episodes me all inclusion c i e ia and we e anno a ed.119
The numbe o en ila ions pe segmen was 4 (3-5). Pa ien s’ median age was 69 yea s (59-78)120
and 36 o he pa ien s we e emale (35.3%). Six y pa ien s we e in uba ed wi h ETT (58.8%) and121
39 wi h sup aglo ic King LT-D (38.2%). In uba ion ype was no a ailable o he emaining 3122
episodes (2.9%). Eigh y pai s o segmen s we e ex ac ed om ETT episodes, 56 om King LT-D123
episodes and 5 om he episodes wi h unknown ai way ype.124
The bes model i o he 141 da a poin s was ob ained o K=0.91 wi h R2=0.93. Fo 125
ETT segmen s, bes i was ob ained o K=0.91 wi h R2=0.93, and o King LT-D segmen s126
o K=0.92 wi h R2=0.93. Figu e 4 shows he da a poin s and he cu e i ing o he global127
analysis and o he wo ai way ypes.128
The physical in e p e a ion o he model becomes mo e in ui i e om inspec ion o Figu e 4129
( op panel). ETCO2al e s wi h en ila ion a e acco ding o he exponen ial unc ion d i en by130
he coe icien K. As highligh ed in he igu e, any ETCO2 alue measu ed in an inasu e al whe e131
en ila ions a e applied wi h a mean en ila ion a e o 5 pm could be no malised o a e e ence132
en ila ion a e o 10 pm ( he guidelines a ge ) by di iding he measu ed alue by 1.62. The133
no malised ETCO2 alue is hen lowe han he measu ed alue. Simila ly, any ETCO2 alue134
ob ained in an in e al wi h a mean en ila ion a e o 15 pm could be no malised o he e e ence135
o 10 pm by di iding he measu ed alue by 0.81, yielding a no malised ETCO2 alue highe han136
he measu ed alue.137
4. Discussion138
Capnog am in e p e a ion du ing CPR is challenging since many ac o s in luence ETCO2
139
a ia ion: en ila ion a e and olume, ches comp ession dep h and a e, pa ien me abolism, o 140
d ug adminis a ion. Fo adequa e in e p e a ion, we need o know he e↵ec o each ac o on141
ETCO2 a ia ion. In his s udy, we p opose a model o quan i ying he e↵ec o en ila ion a e142
on ETCO2 a ia ion.143
A e ex ensi e analysis o all signals con ained in 508 OHCA episodes we selec ed 141 segmen 144
pai s om 102 pa ien s. Figu e 4 ( op panel) shows he bes i ing cu e o all da a poin s wi h145
K=0.91. The coe icien o de e mina ion R2=0.93 demons a ed he goodness o i . Acco ding146
o ou model, ETCO2dec eases exponen ially wi h inc easing en ila ion a e. Mo eo e , ETCO2
147
a ia ion is mo e ab up o a es lowe han 10 pm.148
We also analysed he in luence o he ad anced ai way ype in ou esul s. Fo ha aim, we149
adjus ed sepa a ely he da a poin s co esponding o episodes in which ai way ype was ETT and150
King LT-D (Figu e 4 middle and bo om panels, espec i ely). Resemblance be ween bo h models151
(K=0.91 o ETT and K=0.92 o King LT-D) led us o conclude ha using ETT o King LT-D152
did no in luence he ETCO2 a ia ion wi h en ila ion a e.153
S udies add essing his opic in he li e a u e a e sca ce. Sheak e al. conduc ed a mul icen e 154
coho s udy o 583 in-hospi al and ou -o -hospi al ca diac a es s.25 They used a mul iple linea 155
eg ession model o p edic ETCO2 a ia ion as a unc ion o comp ession dep h, comp ession156
a e and en ila ion a e. They epo ed ha o e e y 10 mm inc ease in dep h, ETCO2
157
inc eased 1.4 mmHg, and o e e y 10 pm inc ease in en ila ion a e, ETCO2d opped 3.0 mmHg.158
Comp ession a e was no a p edic o o ETCO2. Mu phy e al. conduc ed an obse a ional159
p ospec i e s udy wi h simila objec i es which included 230 pa ien s.26 The associa ion be ween160
log- ans o med ETCO2and CPR a iables was assessed h ough linea mixed e↵ec models. The161
au ho s concluded ha a 10 mm inc ease in comp ession dep h was associa ed wi h a 4.0% inc ease162
in ETCO2; a 10 pm inc ease in en ila ion a e wi h a 17.4% dec ease in ETCO2; and a 10 cpm163
inc ease in comp ession a e wi h a 1.7% inc ease in ETCO2. Bo h s udies assumed ha he na u e164
o he dependency o ETCO2wi h comp ession a iables and en ila ion a e is he same, linea 165
o loga i hmic. Tha assump ion migh comp omise he accu a e in e p e a ion o he esul s.166
By hese models, an inc emen o comp ession dep h om 30 o 50 mm ( om subop imal o he167
minimum ecommended dep h) would only aise ETCO2by 2.8 mmHg25 o 8%26.168
The esul s o ou s udy explain he a ia ion o ETCO2di↵e en ly, and a e much mo e in line169
wi h he conclusions o he expe imen al swine s udy by Gazmu i e al.24 In Gazmu i’s s udy, he170
use o mechanical en ila ion du ing CPR allowed en ila ion a e and olume o be con olled. An171
adjus ed cu e o he expe imen al da a allowed o he es ima ion o ETCO2le el in mmHg as a172
unc ion o he exchanged en ila ion olume in li es pe minu e (l/min), acco ding o exp ession:173
ETCO2(mmHg) = 10.3+ 61.2
minu e olume (l/min),(3)
Conside ing an a e age swine weigh o 33 kg and a cons an idal olume o 6 ml/kg, he ETCO2
174
le el ob ained expe imen ally by Gazmu i e al. could be exp essed as a unc ion o en ila ion175
a e as:176
ETCO2(mmHg) = 10.3+ 61.2
0.198 · en . a e ( pm) (4)
Va ying he en ila ion a e om 3 o 30 pm and no malising he esul s o a en ila ion a e177
o 10 pm, a cu e can be easily ob ained and compa ed wi h he cu e ob ained om ou da a.178
Figu e 5 depic s join ly ou model (showed in he op panel o Figu e 4) and he cu e adap ed179
om he expe imen s wi h swine by Gazmu i e al. T ends o bo h models a e simila and he180
di↵e ences in he ange o usual en ila ion a es a e small: 7.2% o 2 = 5 pm and 13.2% o 181
2 = 20 pm.182
The main clinical applica ions o ou indings a e wo old. Fi s , in s udies on he ole o ETCO2
183
as an ea ly de ec o o ROSC o as an indica o o p ognosis o CPR,14–20 en ila ion a e is an184
impo an con ounding ac o ha mus be adjus ed o . The cu e depic ed in he op panel o 185
Figu e 4 could be used as a ma hema ical ool o e e all measu ed ETCO2 alues o he same186
en ila ion a e (ETCO2no malisa ion) and hus help co ec o he e↵ec s o his con ounding187
ac o . Fo example, i a ce ain ETCO2 alue is measu ed (ET1) a a a e o 5 pm, he alue188
ha would ha e been measu ed unde simila condi ions a a a e o 10 pm would be ET1/1.62.189
Simila ly, i a alue ET1 is measu ed a 15 pm, he alue measu ed a a a e o 10 pm would be190
ET1/0.81. Second, a p io knowledge o ETCO2 a ia ion wi h en ila ion a e should acili a e191
Figu e 1

Figu e 2
Moni o -de ib illa o episodes
wi h concu en signals
(n=928) Excluded:
capnog am dis o ion
o poo signal quali y
(n=88)
Episodes wi h
good signal quali y
(n=840) Excluded:
episode du a ion <250s
(n=127)
Episodes eligible o
capnog am analysis
(n=713) Excluded:
ETCO2<10 mmHg
(n=71)
Episodes wi h
ETCO2>= 10 mmHg
(n=642) Excluded:
capnog am du a ion
be o e ROSC <150 s
(n=134)
Episodes in-
cluded in he s udy
(n=508)
Figu e 3
Figu e 4
Adjus ed
Gazmu i e al.
Figu e 5
Con lic o in e es s a emen
Au ho Digna Ma ´ıa Gonz´alez-O e o is employed by Bexen Ca dio, a Spanish medical de ice
manu ac u e .
Bexen Ca dio had no ole in s udy unding, o s udy design, da a collec ion and analysis, decision
o publish, o p epa a ion o he manusc ip .
Au ho s So ´ıa Ruiz de Gauna, Jos´e Julio Gu i´e ez, Jesus Ma ´ıa Ruiz, Mikel Le u iondo, Izaskun
Azca a e, Ca los Co cue a, James Knox Russell, and Mohamud Ramzan Daya decla e no con lic
o in e es .
*Con lic o In e es S a emen
Click he e o download Con lic o In e es S a emen : Con lic o In e es .pd

C edi Au ho S a emen
So ´ıa Ruiz de Gauna: Concep ualiza ion; Funding acquisi ion; Me hodology; P ojec
adminis a ion; Resou ces; Supe ision; Valida ion; Roles/W i ing - o iginal d a ; W i ing - e iew
& edi ing
Jos´e Julio Gu i´e ez: Concep ualizac ion; Da a cu a ion; Fo mal analysis; In es iga ion; So wa e;
Valida ion; W i ing - e iew & edi ing.
Jesus Ma ´ıa Ruiz: Concep ualiza ion; Funding acquisi ion; Me hodology; Fo mal analysis;
In es iga ion; Resou ces; So wa e; W i ing - o iginal d a .
Mikel Le u iondo: Da a cu a ion; So wa e; W i ing - e iew & edi ing
Izaskun Azca a e: Da a cu a ion; So wa e; W i ing - e iew & edi ing
Digna Ma ´ıa Gonz´alez-O e o: Da a cu a ion; So wa e; W i ing - e iew & edi ing
Ca los Co cue a: Concep ualiza ion; W i ing - e iew & edi ing
James Knox Russell: Resou ces; Da a cu a ion; W i ing - e iew & edi ing
Mohamud Ramzan Daya: Resou ces; W i ing - e iew & edi ing; Supe ision
C edi Au ho S a emen