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

The impact of ventilation rate on end-tidal carbon dioxide level during manual cardiopulmonary resuscitation Sof´ıa Ruiz de Gaunaa⇤, Jose Julio Guti´erreza,JesusRuiz a, Mikel Leturiondoa, Izaskun Azcaratea, Digna Mar´ıa Gonz´alez-Oteroa,b, Carlos Corcuerac, James Knox Russelld, Mohamud Ramzan Dayad Affiliation and addresses: aUniversity of the Basque Country, UPV/EHU. Bilbao, Bizkaia, Spain bBexen Cardio, Ermua, Bizkaia, Spain cEmergentziak-Osakidetza, Basque Country Health System, Basque Country, Spain dOregon Health & Science University (OHSU). Portland, OR, USA Corresponding author: ⇤ Sof´ıa Ruiz de Gauna email: [email protected] Word counts: Abstract: 237 words Manuscript: 2859 words This is the accepted manuscript of the article that appeared in final form in Resuscitation 156 : 215-222 (2020), which has been published in final form at https://doi.org/10.1016/j.resuscitation.2020.06.007. © 2020 Elsevier under CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/) Abstract Aim: Ventilation rate is a confounding factor for interpretation of end-tidal carbon dioxide (ETCO2) during cardiopulmonary resuscitation (CPR). The aim of our study was to to model the e↵ect of ventilation rate on ETCO2during manual CPR in adult out-of-hospital cardiac arrest (OHCA). Methods: We conducted a retrospective analysis of OHCA monitor-defibrillator files with concurrent capnogram, compression depth, transthoracic impedance and ECG. We annotated pairs of capnogram segments presenting di↵erences in average ventilation rate and average ETCO2 value but with other influencing factors (e.g. compression rate and depth) presenting similar values within the pair. ETCO2variation as a function of ventilation rate was adjusted through curve fitting using non-linear least squares as a measure of goodness of fit. Results: A total of 141 pairs of segments from 102 patients were annotated. Each pair provided a single data point for curve fitting. The best goodness of fit yielded a coefficient of determination R2of 0.93. Our model described that ETCO2decays exponentially with increasing ventilation rate. The model showed no di↵erences attributable to the airway type (endotracheal tube or supraglottic King-LT-D). Conclusion: Capnogram interpretation during CPR is challenging since many factors influence ETCO2. For adequate interpretation, we need to know the e↵ect of each factor on ETCO2.Our model allows quantifying the e↵ect of ventilation rate on ETCO2variation. Our findings could contribute to better interpretation of ETCO2during CPR. Key words: Cardiopulmonary resuscitation (CPR); Waveform capnography; End-tidal carbon dioxide (ETCO2); Ventilation; Ventilation rate; Out-of-hospital cardiac arrest (OHCA); Advanced life support (ALS). 1. Introduction1 High quality cardiopulmonary resuscitation (CPR) improves outcomes of cardiac arrest2 victims.1–4Ideally, CPR quality should be guided based on the patient’s response using3 a potentially non-invasive haemodynamic indicator.5,6Current advanced life support (ALS)4 resuscitation guidelines emphasize the use of waveform capnography during CPR.7,8Waveform5 capnography enables monitoring of end-tidal carbon dioxide (ETCO2), the partial pressure of6 carbon dioxide at the end of expiration. ETCO2is an indirect measure of both cardiac output7 and pulmonary blood flow.9–11 The evolution of ETCO2during the course of resuscitation has8 been found to be valuable in monitoring the quality of chest compressions,12,13 allowing for9 early recognition of return of spontaneous circulation (ROSC),14–16 and as a predictor of patient10 outcome.17–20 11 During CPR the value of ETCO2depends on the blood flow generated by chest compressions, on12 the volume and rate of ventilation and on the patient’s tissues metabolic activity.21,22 Improving13 the quality of chest compressions increases the amount of pulmonary blood flow and therefore14 ETCO2level, provided tissue metabolism is active. However, if ventilation volume or ventilation15 rate increase during the course of resuscitation the ETCO2level decreases provided the blood flow16 remains constant.23,24 17 There are few studies aimed at quantifying the relationship between CPR quality components18 and ETCO2during resuscitation. Recent observational studies with human data used multivariate19 analysis to quantify ETCO2variation in relation to variations in chest compression depth and20 rate, and ventilation rate.25,26 An animal study demonstrated that changes in ventilation rate21 significantly influence ETCO2level.24 The aim of our study was to identify a quantitative22 relationship between ETCO2and ventilation rate through retrospective analysis of adult23 out-of-hospital cardiac arrest (OHCA) episodes.24 2. Materials and methods25 2.1. Data collection26 Data were extracted from a large database of adult out-of-hospital cardiac arrest (OHCA)27 episodes collected from 2006 through 2017 by Tualatin Valley Fire & Rescue (TVF&R), an ALS first28 response emergency medical services (EMS) agency serving eleven incorporated cities in Oregon,29 USA. The database is part of the Resuscitation Outcomes Consortium (ROC) Epidemiological30 Cardiac Arrest Registry collected by the Portland Regional Clinical Center. The data collection31 was approved by the Oregon Health & Science University (OHSU) Institutional Review Board32 (IRB00001736). No patient private information was included in the database.33 Episodes were recorded using Heartstart MRx monitor-defibrillators (Philips, USA), equipped34 with capnography monitoring using sidestream technology (MicrostreamTM, Oridion Systems35 Ltd., Israel) and CPR quality monitoring (Q-CPRTM). TVF&R field providers used either an36 endotracheal tube (ETT) or a supraglottic King LT-DTM device, to secure the airway. Continuous37 chest compressions and ventilations were provided manually. No ventilation volume data was38 colected. For this study, we only included recordings with concurrent capnogram, compression39 depth signal, electrocardiogram (ECG) and transthoracic impedance (TTI) signals. We did not40 consider arrest aetiology in selecting cases.41 2.2. Selection of paired segments42 The hypothesis of the study was that it is possible to assess the ETCO2variation with43 respect to ventilation rate from analysis of pairs of capnogram segments that di↵er only in the44 applied ventilation rate. We hypothesised that this variation in ventilation rate is responsible45 for the di↵erences in ETCO2level between segments, when other influencing factors (metabolism,46 compression depth and compression rate) are controlled through a careful definition of the inclusion47 criteria.48 Using a custom-made Matlab program (Mathworks, USA), two biomedical experts (JJG and49 JMR) visually inspected the four signals extracted from the OHCA defibrillator recordings. They50 identified pairs of segments of capnography signal di↵ering in ventilation rate and ETCO2values51 but with the other confounding factors presenting similar behaviour in both segments. A reliable52 capnogram was required to identify individual ventilations (with the support of the TTI signal)27 53 and to annotate a single representative ventilation rate and ETCO2value for each segment in the54 pair. To provide that the observed ETCO2variation was caused only by the change in ventilation55 rate, the following inclusion criteria were established for each identified pair:56 •Absence of a pulse-generating rhythm confirmed by inspection of the ECG and review of57 ROSC annotations recorded by the ALS providers in case notes. In cases where there was58 doubt about the presence of a perfusing rhythm and pulseless electrical activity, we evaluated59 the TTI signal for evidence of a perfusing rhythm.28 60 •Similar metabolic activity so that a maximum of 2-min separation between segments was61 allowed.62 •Similar chest compression depth and rate between the two segments so that a similar amount63 of pulmonary blood flow could be assumed. Only segments with di↵erences of less than 4 mm64 in mean compression depth and 8 compressions per minute (cpm) in mean compression rate65 were included.66 Figure 1 shows two examples of segment selection (panels A and B). Capnogram and67 compression depth signals are depicted for two of the pairs included in the study. Annotated68 values for each segment were: the start and end of each segment, in seconds (ts1, te1; ts2, te2)69 which are depicted in the figure with vertical red lines; the number of ventilations within each70 segment (nv1, nv2); the mean ETCO2value per segment (ET1, ET2); the mean compression71 depth and mean compression rate per segment (dc1, rc1; dc2, rc2), and the time interval between72 segments (tbs). The elapsed time between segments was used to check if the separation between73 segments was below 2 min, and the mean compression depth and mean compression rate of each74 segment of the pair (dc1,rc1; dc2, rc2) were used to check if the criteria of similar compression75 depth and rate were met. Once a pair met all the inclusion criteria it was characterised by three76 features:77 •Ventilation rate for segment 1: vr1 = 60 ·nv1/(te1ts1) vpm.78 •Ventilation rate for segment 2: vr2 = 60 ·nv2/(te2ts2) vpm.79 •ETCO2ratio between segments: ET2/ET1.80 2.3. Model fitting81 The experimental values were adjusted through the analytic expression:82 ET2 ET1=1Kvr1 1Kvr2,(1) where the coefficient Kis the only parameter determinant of the adjustment. This83 mathematical model was proposed in a previous work as an hypothesis for explaining the influence84 of ventilation rate in ETCO2during CPR.29 85 To facilitate physical interpretation of the results, we normalised Equation 1 with respect to86 the reference ventilation rate of vr1 = 10 vpm (the rate recommended by current resuscitation87 guidelines for ALS7,8). Thus, the normalised model used for the data adjustment was:88 ET2 ET1    10 =ET2 ET1·1K10 1Kvr1=1K10 1Kvr2(2) Equation 2 provides a family of decaying curves as ventilation rate vr2 increases (a single curve89 for each value of K). Figure 2 depicts the family of curves resulting from Equation 2 for vr290 varying from 3 to 35 vpm and Kfrom 0.75 to 0.99 in 0.03 increments. The adjustment of K91 allows modelling the decay rate. The model responds to the expected behaviour of ETCO2when92 ventilation rate increases.93 The experimental data were normalised accordingly: for each pair of segments we obtained one94 data point for the adjustment characterised by a value for vr2 and a normalised value for the ratio95 ET2/ET1.96 2.4. Statistical analysis97 Values were reported as median (interquartile range, IQR). The model was fitted using the98 weighted bi-squared method for robustness. This method minimised a weighted sum of the squared99 residuals, finding a curve that fits the bulk of the data using the least-squares approach and,100 simultaneously, minimising the e↵ect of the outliers.30 Goodness of fit of the model was evaluated101 using the coefficient of determination R2, which provides a measure of the ET2/ET 1 variation102 that is explained by the model.103 We also analysed di↵erences in the model fit with respect to the airway management technique,104 ETT or supraglottic King LT-D.105 3. Results106 Concurrent signals of interest (capnogram, compression depth, ECG and TTI signals) were107 available in 928 adult OHCA episodes (see Figure 3). Episodes with capnograms substantially108 distorted by chest compression artefact27,31 or poor signal quality were discarded (n= 88) because109 in those conditions ventilation rate and ETCO2values could not be accurately measured. Episodes110 with less than 250 s of concurrent signals were excluded (n= 127) since they corresponded to very111 short resuscitative attempts or late advanced airway placement. We also excluded episodes where112 ETCO2values were lower than 10 mmHg most of the time (n= 71) because small inaccuracies in113 the manual measurement of ETCO2 could cause significant percent errors distorting the input data114 for our model. Finally, episodes where the capnogram duration before any ROSC annotation was115 less than 150 s were also excluded (n= 134). Thus, we avoided selecting segment pairs near ROSC116 where the presence of pulse could a↵ect the ETCO2 level. A total of 508 adult OHCA episodes117 were included in the study with a total capnogram duration of 12,819 min.118 A total of 141 pairs of segments from 102 episodes met all inclusion criteria and were annotated.119 The number of ventilations per segment was 4 (3-5). Patients’ median age was 69 years (59-78)120 and 36 of the patients were female (35.3%). Sixty patients were intubated with ETT (58.8%) and121 39 with supraglottic King LT-D (38.2%). Intubation type was not available for the remaining 3122 episodes (2.9%). Eighty pairs of segments were extracted from ETT episodes, 56 from King LT-D123 episodes and 5 from the episodes with unknown airway type.124 The best model fit for the 141 data points was obtained for K=0.91 with R2=0.93. For125 ETT segments, best fit was obtained for K=0.91 with R2=0.93, and for King LT-D segments126 for K=0.92 with R2=0.93. Figure 4 shows the data points and the curve fitting for the global127 analysis and for the two airway types.128 The physical interpretation of the model becomes more intuitive from inspection of Figure 4129 (top panel). ETCO2alters with ventilation rate according to the exponential function driven by130 the coefficient K. As highlighted in the figure, any ETCO2value measured in an inasuterval where131 ventilations are applied with a mean ventilation rate of 5 vpm could be normalised to a reference132 ventilation rate of 10 vpm (the guidelines target) by dividing the measured value by 1.62. The133 normalised ETCO2value is then lower than the measured value. Similarly, any ETCO2value134 obtained in an interval with a mean ventilation rate of 15 vpm could be normalised to the reference135 of 10 vpm by dividing the measured value by 0.81, yielding a normalised ETCO2value higher than136 the measured value.137 4. Discussion138 Capnogram interpretation during CPR is challenging since many factors influence ETCO2 139 variation: ventilation rate and volume, chest compression depth and rate, patient metabolism, or140 drug administration. For adequate interpretation, we need to know the e↵ect of each factor on141 ETCO2variation. In this study, we propose a model for quantifying the e↵ect of ventilation rate142 on ETCO2variation.143 After extensive analysis of all signals contained in 508 OHCA episodes we selected 141 segment144 pairs from 102 patients. Figure 4 (top panel) shows the best fitting curve for all data points with145 K=0.91. The coefficient of determination R2=0.93 demonstrated the goodness of fit. According146 to our model, ETCO2decreases exponentially with increasing ventilation rate. Moreover, ETCO2 147 variation is more abrupt for rates lower than 10 vpm.148 We also analysed the influence of the advanced airway type in our results. For that aim, we149 adjusted separately the data points corresponding to episodes in which airway type was ETT and150 King LT-D (Figure 4 middle and bottom panels, respectively). Resemblance between both models151 (K=0.91 for ETT and K=0.92 for King LT-D) led us to conclude that using ETT or King LT-D152 did not influence the ETCO2variation with ventilation rate.153 Studies addressing this topic in the literature are scarce. Sheak et al. conducted a multicenter154 cohort study of 583 in-hospital and out-of-hospital cardiac arrests.25 They used a multiple linear155 regression model to predict ETCO2variation as a function of compression depth, compression156 rate and ventilation rate. They reported that for every 10 mm increase in depth, ETCO2 157 increased 1.4 mmHg, and for every 10 vpm increase in ventilation rate, ETCO2dropped 3.0 mmHg.158 Compression rate was not a predictor of ETCO2. Murphy et al. conducted an observational159 prospective study with similar objectives which included 230 patients.26 The association between160 log-transformed ETCO2and CPR variables was assessed through linear mixed e↵ect models. The161 authors concluded that a 10 mm increase in compression depth was associated with a 4.0% increase162 in ETCO2; a 10 vpm increase in ventilation rate with a 17.4% decrease in ETCO2; and a 10 cpm163 increase in compression rate with a 1.7% increase in ETCO2. Both studies assumed that the nature164 of the dependency of ETCO2with compression variables and ventilation rate is the same, linear165 or logarithmic. That assumption might compromise the accurate interpretation of the results.166 By these models, an increment of compression depth from 30 to 50 mm (from suboptimal to the167 minimum recommended depth) would only raise ETCO2by 2.8 mmHg25 or 8%26.168 The results of our study explain the variation of ETCO2di↵erently, and are much more in line169 with the conclusions of the experimental swine study by Gazmuri et al.24 In Gazmuri’s study, the170 use of mechanical ventilation during CPR allowed ventilation rate and volume to be controlled. An171 adjusted curve to the experimental data allowed for the estimation of ETCO2level in mmHg as a172 function of the exchanged ventilation volume in litres per minute (l/min), according to expression:173 ETCO2(mmHg) = 10.3+ 61.2 minute volume (l/min),(3) Considering an average swine weight of 33 kg and a constant tidal volume of 6 ml/kg, the ETCO2 174 level obtained experimentally by Gazmuri et al. could be expressed as a function of ventilation175 rate as:176 ETCO2(mmHg) = 10.3+ 61.2 0.198 ·vent. rate (vpm) (4) Varying the ventilation rate from 3 to 30 vpm and normalising the results to a ventilation rate177 of 10 vpm, a curve can be easily obtained and compared with the curve obtained from our data.178 Figure 5 depicts jointly our model (showed in the top panel of Figure 4) and the curve adapted179 from the experiments with swine by Gazmuri et al. Trends of both models are similar and the180 di↵erences in the range of usual ventilation rates are small: 7.2% for vr2 = 5 vpm and 13.2% for181 vr2 = 20 vpm.182 The main clinical applications of our findings are twofold. First, in studies on the role of ETCO2 183 as an early detector of ROSC or as an indicator of prognosis of CPR,14–20 ventilation rate is an184 important confounding factor that must be adjusted for. The curve depicted in the top panel of185 Figure 4 could be used as a mathematical tool to refer all measured ETCO2values to the same186 ventilation rate (ETCO2normalisation) and thus help correct for the e↵ects of this confounding187 factor. For example, if a certain ETCO2value is measured (ET1) at a rate of 5 vpm, the value188 that would have been measured under similar conditions at a rate of 10 vpm would be ET1/1.62.189 Similarly, if a value ET1 is measured at 15 vpm, the value measured at a rate of 10 vpm would be190 ET1/0.81. Second, a prior knowledge of ETCO2variation with ventilation rate should facilitate191 Figure 1 Figure 2 Monitor-defibrillator episodes with concurrent signals (n=928) Excluded: capnogram distortion or poor signal quality (n=88) Episodes with good signal quality (n=840) Excluded: episode duration <250s (n=127) Episodes eligible for capnogram analysis (n=713) Excluded: ETCO2<10 mmHg (n=71) Episodes with ETCO2>= 10 mmHg (n=642) Excluded: capnogram duration before ROSC <150 s (n=134) Episodes included in the study (n=508) Figure 3 Figure 4 Adjusted Gazmuri et al. Figure 5 Conflict of interest statement Author Digna Mar´ıa Gonz´alez-Otero is employed by Bexen Cardio, a Spanish medical device manufacturer. Bexen Cardio had no role in study funding, or study design, data collection and analysis, decision to publish, or preparation of the manuscript. Authors Sof´ıa Ruiz de Gauna, Jos´e Julio Guti´errez, Jesus Mar´ıa Ruiz, Mikel Leturiondo, Izaskun Azcarate, Carlos Corcuera, James Knox Russell, and Mohamud Ramzan Daya declare no conflict of interest. *Conflict of Interest Statement Click here to download Conflict of Interest Statement: ConflictofInterest.pdf Credit Author Statement Sof´ıa Ruiz de Gauna: Conceptualization; Funding acquisition; Methodology; Project administration; Resources; Supervision; Validation; Roles/Writing - original draft; Writing - review & editing Jos´e Julio Guti´errez: Conceptualizaction; Data curation; Formal analysis; Investigation; Software; Validation; Writing - review & editing. Jesus Mar´ıa Ruiz: Conceptualization; Funding acquisition; Methodology; Formal analysis; Investigation; Resources; Software; Writing - original draft. Mikel Leturiondo: Data curation; Software; Writing - review & editing Izaskun Azcarate: Data curation; Software; Writing - review & editing Digna Mar´ıa Gonz´alez-Otero: Data curation; Software; Writing - review & editing Carlos Corcuera: Conceptualization; Writing - review & editing James Knox Russell: Resources; Data curation; Writing - review & editing Mohamud Ramzan Daya: Resources; Writing - review & editing; Supervision Credit Author Statement