The use of laparoscopy in general surgery using CO2 to create and maintain a pneumoperitoneum has become widely accepted [1,2]. Consequently, not only young, fit patients, but also older patients, are exposed to the effects of intraperitoneal CO2 for longer periods of time [3,4]. Hypercarbia and acidosis, ventilatory, i.e., decreased pulmonary compliance and vital capacity, and cardiovascular effects, i.e., decreased cardiac index and increased systemic vascular resistance, may be seen during this exposure [4-8]. In older patients with cardiopulmonary disease, it has been suggested that careful intraoperative arterial blood gas monitoring may be required .
In previous studies [3-5,7], intermittent arterial blood samples were used to obtain values for pH and PaCO2. Therefore, this method may fail to detect sudden changes in PaCO2 and pH, and merely provides isolated "snapshots" of the course of these values during the procedure. With the introduction of a new intraarterial blood gas monitor, blood gas trends can be observed accurately and other disadvantages of blood gas analyzers, such as a time delay and increased risk of bleeding and infection, can be overcome [9,10].
Totally extraperitoneal laparoscopic inguinal hernia repair is a promising technique that is increasingly performed [11,12]. During the performance of this technique the preperitoneal space is insufflated with CO2 (pneumopreperitoneum). The influence of a pneumopreperitoneum on lung compression and CO2 absorption may differ from that of a pneumoperitoneum.
This study was undertaken to compare the effect of CO2 pneumopreperitoneum with CO2 pneumoperitoneum on the development of hypercarbia and acidosis. Additionally, we evaluated continuous intraarterial blood gas monitoring in this setting.
The study was approved by the institutional ethics committee of the University Hospital Utrecht. Written, informed consent was obtained from all patients participating in this study, which was conducted in a prospective, nonrandomized fashion. Patients who were operated for inguinal hernia repair using a totally extraperitoneal laparoscopic approach involving only pneumopreperitoneum were compared with patients who underwent laparoscopic cholecystectomy involving pneumoperitoneum. Both groups met the following inclusion criteria: 1) patients older than 20 yr; 2) suitable for general anesthesia; 3) possibility of a laparoscopic approach, i.e., no history of extensive abdominal surgery, radiotherapy, or serious abdominal inflammation nor advanced pregnancy.
Patients were excluded if serious atherosclerotic peripheral arterial disease or cardiovascular disease was present . Additional exclusion criteria were: 1) intermittent claudication; 2) impossible radial artery cannulation; 3) coagulopathy; 4) use of vasoconstrictor drugs .
Patients who were selected for laparoscopic inguinal hernia repair were part of an ongoing prospective randomized study to compare totally extraperitoneal hernia repair with conventional inguinal hernia repair. These patients met the following additional inclusion criteria: 1) primary or first recurrent unilateral inguinal hernia; 2) reducible inguinal hernia.
Preoperatively, forced expiratory volume at 1 s and forced vital capacity, were measured in all patients using a Vitalograph alpha (Lameris, Buckingham, UK).
Patients were given similar general anesthesia using the following anesthetics: propofol (2 mg/kg for induction and 2 mg centered dot kg-1 centered dot h-1 for maintenance) or midazolam (0.25 mg/kg for induction and 0.2 mg centered dot kg-1 centered dot h-1 for maintenance) and sufentanil (0.5 micro gram/kg for induction and 0.5 micro gram centered dot kg-1 centered dot h-1 for maintenance) for analgesia, and vecuronium (0.1 mg/kg for induction and 0.06 mg centered dot kg-1 centered dot h-1 for maintenance) or atracrurium (0.3 mg/kg for induction and 0.3 mg centered dot kg-1 centered dot h-1 for maintenance) for muscle relaxation. Patients were ventilated with a gaseous mixture of O2 (40%) and air.
After tracheal intubation, the lungs were mechanically ventilated using a Drager 656 Narkosespiromat (Drager, Lubeck, Germany). Patients were ventilated to obtain a steady state, defined by a stable end-tidal carbon-dioxide partial pressure (PETCO2) between 27 and 31.5 mm Hg during at least 5 min. PETCO2 was measured continuously by a side-stream infrared analysis with the sensor placed between the endotracheal tube and the breathing circuit (Datex Capnomac Ultima; Instrumentation Corp., Helsinki, Finland). Minute ventilation was recorded using a flow transducer on the expiratory side of the circuit. Peak airway pressure was recorded by an aneroid pressure gauge in the ventilator. Ventilation frequency, tidal volume, FIO2, and minimum airway pressure were recorded on digital hard copy every 5 min, starting directly after induction.
After tracheal intubation, the radial artery was cannulated with a 20-gauge intraarterial cannula and the patients were connected to a continuous intraarterial blood gas system (Paratrend 7, Biomedical Sensors; Pfizer, Highwycombe, United Kingdom) for registration of PaO2, PaCO2, and pH. The intraarterial blood gas sensor was calibrated and handled strictly according to the guidelines of the manufacturer. These values were monitored continuously during the operation period and on the recovery ward and were recorded every 5 min. Furthermore, the maximum value of PaCO2 and the minimum value of pH were identified from all values seen from the continuous blood gas monitor during the operative procedure.
Two arterial blood samples were taken from each patient to analyze PaO (2), PaCO2, and pH using an ABL520 blood gas analyzer (Radiometer, Copenhagen, Danmark), and to validate the measurements obtained by the Paratrend 7. Blood samples were taken during the steady state of ventilation before induction of the pneumo(pre)peritoneum (baseline) and 15 min after induction of a pneumo(pre)peritoneum. If there was any reason to doubt the accuracy of the blood gas values obtained by the sensor, an additional blood sample was taken. This was done when blood pressure, electrocardiography, or capnography suggested the possibility of excessive hypercarbia.
Blood pressure was measured by the same intraarterial catheter as used for the continuous intraarterial blood gas sensor. If a depressed pressure curve was seen, blood pressure was measured noninvasively with a blood pressure manometer. Heart rate was measured by electrocardiography. Values were recorded every 5 min on digital hard copy.
Ventilation settings were not adjusted unless a pH of 7.20 or lower was observed by continuous intraarterial blood gas monitoring, at which time ventilator settings were adjusted by altering ventilatory frequency and tidal volume. If cardiac arrhythmias or hemodynamic instability were seen, these corrections were made sooner.
Laparoscopic cholecystectomy and totally extraperitoneal laparoscopic hernia repair were performed in standardized fashions as have been described previously [11,12,14]. Both pneumoperitoneum and pneumopreperitoneum were induced with CO (2) allowing a maximum intra(pre)peritoneal pressure of 15 mm Hg.
As the main outcome measure, we used PaCO2. A SD of 2.5 mm Hg was estimated from the range of PaCO2 (normal range, 34.9-45.0 mm Hg). With a sample size of 15 patients in each group, a power of 80% and an alpha of 0.05, a difference of 2.55 mm Hg between the two groups could be detected.
Baseline and other differences between groups were tested with unpaired Student's t-test where appropriate.
The method described by Bland and Altman was used previously to assess agreement between blood gas analyzer and intraarterial blood gas sensor [9,15] and was also used because, in fact, the true value remains unknown as both methods have measurement errors. This method plots the difference between two methods against their mean, and uses the mean difference (bias) and the SD of the difference (precision) to describe agreement. To assess agreement between both methods in their capacity to detect changes in time within one patient, we performed a similar analysis by calculating the change after 15 min of insufflation (15-min value minus baseline value) for each method within each patient. Accordingly, for pH and PaCO2 we plotted the differences of the change between the methods against the mean changes. Similarly, bias and precision were calculated. Additionally, the 95% confidence intervals of mean, lower, and upper limit of agreement (mean - 2 SD and mean + 2 SD, respectively) were calculated.
The main analyses compared the pneumopreperitoneum with the pneumoperitoneum group to evaluate time trends (every 5 min) in PaCO2 and pH using either the multivariate or the split-plot approach analysis of variance for repeated measures. As we were interested in changes, we subtracted baseline values from all following measurements. Additionally, heart rate, systolic and diastolic arterial blood pressure, FIO2, PaO2, tidal and minute volume, peak expiratory pressure and positive end-expiratory pressure, pulmonary function, age, weight, and length were used as covariates in the corresponding analysis of covariance.
Values are expressed as mean +/- SD. A P value of <0.05 was considered statistically significant. All reported P values are two-tailed.
Fourteen consecutive patients undergoing elective totally extraperitoneal laparoscopic hernia repair and 13 consecutive patients undergoing elective laparoscopic cholecystectomy were included from September 1994 to January 1995 Table 1. During this period, three patients undergoing laparoscopic cholecystectomy refused to participate. After informed consent no data could be collected from two patients undergoing cholecystectomy and one patient undergoing hernia repair due to failure of the intraarterial blood gas sensor.
Baseline values of PaCO2 (37.5 +/- 2.6 mm Hg vs 37.7 +/- 3.3 mm Hg, n = 27, P = 0.81) and pH (7.43 +/- 0.04 vs 7.44 +/- 0.03, n = 27, P = 0.50) for pneumopreperitoneum and pneumoperitoneum groups measured by the sensor did not differ between the two groups.
The agreement between measurements made by the intraarterial sensor and blood gas analyzer are shown in Figure 1 A-D. There was a consistent bias in PaCO (2) values Figure 1B. Bias of pH values 15 min after insufflation was 0.000 and the precision was 0.024 Figure 1A. Bias of PaCO2 values at 15 min after insufflation was -2.32 mm Hg and precision of PaCO2 was 2.07 mm Hg Figure 1B.
Bias and precision of changes in pH were -0.011 and 0.023 Figure 1C. Bias and precision of changes in PaCO2 were 0.29 mm Hg and 2.35 mm Hg Figure 1D.
(Table 2) displays hemodynamic variables during both procedures. Blood pressure increased slightly in both groups, but increased more during pneumoperitoneum than during pneumopreperitoneum.
Mean maximum values of PaCO2 for both procedures did not differ statistically significantly (56.1 +/- 9.2 mm Hg for pneumopreperitoneum, n = 14, vs 51.8 +/- 9.2 mm Hg for pneumoperitoneum, n = 13; P = 0.24), nor did mean minimum values of pH (7.31 +/- 0.7, n = 14, vs 7.32 +/- 0.7, n = 13; P = 0.92). If we omitted the results of three patients in whom the true maximum value of PaCO2 and true minimum value of pH were not reached because we intervened according to our protocol (i.e., a pH of 7.2) the results did not reach significance (54.6 +/- 9.1 mm Hg for pneumopreperitoneum, n = 12, vs 49.9 mm Hg +/- 6.2 for pneumoperitoneum, n = 12; P = 0.15), nor did mean minimum values of pH (7.33 +/- 0.05 for pneumopreperitoneum vs 7.33 +/- 0.06 for pneumoperitoneum; P = 0.74). The maximum observed value of PaCO (2) during pneumopreperitoneum occurred sooner than the maximum value for pneumoperitoneum (41 +/- 10 vs 65 +/- 29 min after induction of pneumo(pre)peritoneum). During pneumopreperitoneum no clear plateau could be observed.
The insufflation period for all patients during herniorrhaphy was much shorter than during cholecystectomy (42 +/- 12 min, range 21-62 min, vs 84 +/- 34 min, range 45-151 min; P = 0.001).
With analysis of variance for repeated measures there was an expected general increase in PaCO2 (F = 19.8; P < 0.0005) and decrease in pH (F = 17.1; P < 0.0005) averaged over both groups. Furthermore, we found that patients in the pneumopreperitoneum group developed hypercarbia and acidosis more rapidly. This result was statistically significant. There were differences for linear time trends between the two groups: a larger increase, i.e., a steeper slope, in PaCO2 (F = 3.37; P = 0.023; first order polynomial, F = 15.69; P = 0.0007) and a larger decrease in pH (F = 3.23; P = 0.027; first order polynomial, F = 8.55; P = 0.008) for the pneumopreperitoneum group. These results were unaltered when corrected for with one of the covariates and thus none of the covariates could explain the results (analysis of covariance). Four patients undergoing hernia repair were excluded from this analysis, since the insufflation period of their operation lasted less than 40 min.
Mean values of PaCO2 at 5-min intervals are shown in Figure 2 for all patients.
We were forced to adjust ventilation settings in three patients because unacceptable hypercarbia and acidosis developed. Two of them underwent herniorrhaphy and pneumopreperitoneum. The first was an 80-yr-old male with marginal pulmonary function (forced vital capacity = 1.83, 64% of predicted value; forced expiratory volume at 1 s = 0.97, 47% of predicted value), the second, however, was a 60-yr-old male with adequate pulmonary function. Ventilation had to be adjusted after 45 and 43 min, respectively. The third patient, a healthy 40-yr-old female, underwent cholecystectomy and pneumoperitoneum. After an accidentally increased intraabdominal pressure up to 20 mm Hg, profound hypercarbia developed within 5 min and ventilation had to be adjusted 56 min after insufflation. Hypercarbia and acidosis in all three patients were corrected without complications.
During laparoscopic herniorrhaphy, in most cases, two slopes were observed during the increase in PaCO2 and the decrease in pH. During the first part of the operation (preperitoneal dissection) the slope was steeper than during the second part (positioning of the mesh). No sudden changes of PaCO2 and pH after position changes (Trendelenburg and reverse Trendelenburg position) were observed in any patient.
Postoperatively, no complications of radial artery cannulation nor the operation occurred; recovery and hospital stay were not prolonged.
We found that pneumopreperitoneum for laparoscopic herniorrhaphy resulted in a more rapid increase in PaCO2 and consequent decrease in pHa than pneumoperitoneum for laparoscopic cholecystectomy. However, the maximum observed value for PaCO2 and the minimum observed value for pHa during the two procedures were not statistically significantly different. The observed difference in increase in PaCO (2) between the two procedures can be explained by a more rapid CO2 absorption during pneumopreperitoneum. Several other differences between pneumopreperitoneum and pneumoperitoneum could possibly also account for the difference in PaCO2 and pHa.
Firstly, the lungs may be differently affected by the two procedures, resulting in different dead space or alveolar ventilation changes. A faster and higher increase in PaCO2 may be explained by increased dead space or less alveolar ventilation. However, one would expect this to be less during pneumopreperitoneum than during pneumoperitoneum. Nevertheless, we used tidal and minute volume and peak expiratory pressure as covariates to exclude a possible influence, but no significant change in outcome was observed. Thus, this explanation is not likely to account for our observations.
Secondly, there may have been differences between the two groups in metabolic carbon dioxide production. This may have been induced by different mechanical (compressive) effects on venous return and systemic vascular resistance which would alter cardiac output [5,6,16]. In both groups, PaCO2 increased. If this had been the result of changes in the metabolic rate it would have been indicated by more significant changes in oxygen consumption, heart rate, and blood pressure in the pneumopreperitoneum group. Blood pressure slightly increased in both groups, but blood pressure increased more during pneumoperitoneum Table 2. FIO2 was unaltered and PaO2 was stable throughout the procedure. Nevertheless, we used blood pressure, heart rate, FIO2, and PaO2 as covariates to exclude a possible influence, but no significant change in outcome was observed. Therefore, the observed changes cannot be accounted for by differences in metabolic carbon dioxide production.
Hence, we believe that the observed effect is entirely due to absorption of CO2, which is more extensive in the preperitoneal space than in the intraperitoneal space. This is probably due to a larger total gas exchange area as a result of the absence of a natural border. CO2 can easily diffuse into and between subcutaneous tissues or into the scrotum. In contrast, this is prevented during pneumoperitoneum as the peritoneum functions as a natural border. Furthermore, during laparoscopic herniorrhaphy before commencing insufflation, a preperitoneal dissection is performed. Lateral dissection for mesh placement during the procedure results in an increasing total gas exchange area .
This explanation can be considered in terms of the components of the formula for diffusion of gases between two compartments. Diffusion of gases is dependent on a pressure gradient between the two compartments and not dependent on the infused volume into the compartments . Therefore, it was relevant to control only for intra(pre)peritoneal pressure of CO2. In this setting, especially the larger gas exchange area but probably also a shorter anatomic distance (diffusion through the vascular wall instead of diffusion through the vascular wall and peritoneum) may result in an increased influx into the circulation. This is further supported by our observation of a biphasic PaCO2 and pH curve during herniorrhaphy. Change in the slope of the curve during increase in PaCO2 and decrease in pH occurred at the time when preperitoneal dissection was finished and thus no active increase in total gas exchange area occurred. However, the larger exchange area remained. Additionally, increasing gas exchange area and resultant CO2 uptake would explain the slight increases in PaCO2 sometimes observed when the gall bladder was dissected of its liver bed.
Our observations are also in concordance with those of several others. Mullet et al.  measured VCO2 and PETCO2 but not PaCO2 and found a continued increase in these values during extraperitoneal pelviscopy. They concluded that it takes longer during extraperitoneal absorption to reach a steady state and that there is a continued recruitment of gas exchange area during the continued dissection. Hall et al.  detected a case of hypercarbia during laparoscopic cholecystectomy with continuous capnometry due to extensive subcutaneous emphysema which increased the gas exchange area.
The maximum difference between pneumopreperitoneum and pneumoperitoneum of PaCO2 amounted to 4.7 mm Hg. This difference was not statistically significant. However, the measured difference is appreciable (49% vs 38% increase from baseline values) and might have reached significance if pneumopreperitoneum was prolonged. This might be expected because maximum PaCO2 and minimum pHa occurred at the end of the procedure and no clear plateau could be observed. At that time preperitoneal CO2 had not yet reached an equilibrium with PaCO2. In other words the insufflation period during laparoscopic herniorrhaphy could have been too short to measure significant differences.
In three cases it was necessary to adjust ventilation to correct hypercarbia and acidosis. The first patient had poor pulmonary function. Wittgen et al.  found preoperative pulmonary function tests to be predictive of whether acidosis develops during the procedure. Although we measured only vital and forced expiratory capacity, their assumption is in agreement with our observation in the 80-yr-old patient. However, it cannot explain why ventilation in the other two patients in our study had to be adjusted due to acidosis. One case could be explained by an abnormally high intraperitoneal pressure; in the other case the patient had adequate pulmonary function. The acidosis in this case might be explained by the use of sharp preperitoneal dissection during the procedure and therefore a larger wound bed than usual.
Position changes are likely to be of minimal influence on PaCO2 changes. The Trendelenburg position may compromise pulmonary function, i.e., decrease of functional residual capacity and decrease in compliance . The fact that no clinically important effects on PaCO2 were seen during insufflation directly after position changes either from Trendelenburg or reverse Trendelenburg to supine position or vice versa, suggests that different lung compression caused by position changes is of minimal influence on PaCO2.
We used continuous intraarterial blood gas monitoring to detect maximum values of PaCO2 and minimum values of pHa exactly and not to miss any trends. There was a consistent bias in absolute values of PaCO2, but there was good agreement between continuous and traditional blood gas monitoring to detect changes in PaCO2 and pHa within one patient. Therefore, this bias did not influence our main analyses, since we were interested in changes. Although not the main goal of this study, our analysis of ability to detect changes with continuous blood gas monitoring has not been performed previously, as the validation of previous studies was done during blood gas stability . During laparoscopy the expected changes after 15 min of CO2 insufflation were detected by both intermittent and continuous blood gas monitoring Figure 1 C and D.
The advantage of continuous intraarterial monitoring was seen clearly in only one case. The acute development of hypercarbia in this patient was detected by intraarterial blood gas monitoring, but would have also been detected if only capnography was used. However, PETCO2 may sometimes remain constant while PaCO (2) further increases  or may inadequately reflect changes in PaCO2[3,4]. We believe that the clinical value of continuous intraarterial blood gas monitoring must be limited during laparoscopy, but that it can be useful for further investigations.
In conclusion, it may be expected that laparoscopic herniorrhaphy will be used more often, as more surgeons are trying to master this technique. Consequently, operation time may often be prolonged, especially in difficult cases. Given this, and the development of more rapid hypercarbia and resultant acidosis, patients undergoing this procedure must be monitored adequately and intensely.
We are indebted to Yolanda van der Graaf, MD, PhD, Department of Clinical Epidemiology, for statistical advice and for carefully reviewing the manuscript. We acknowledge the helpful comments of Nigel Turner, MD. Finally, we thank Walter Hermans, Baxter BV, The Netherlands, for logistic support and Ronald van der Meer, Datex, medical electronics BV, The Netherlands, for lending us the video printer for the Datex Capnomac Ultima.