BJA Advance Access originally published online on December 2, 2006
British Journal of Anaesthesia 2007 98(1):53-59; doi:10.1093/bja/ael311
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Nitrous oxide for monitoring fluid absorption in volunteers
Department of Anesthesiology, Karolinska Institute at South Hospital Stockholm, Sweden
*Corresponding author: Department of Anesthesiology, Karolinska Institute, South Hospital, S-118 83 Stockholm, Sweden. E-mail: r.hahn{at}telia.com
Accepted for publication August 14, 2006.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Background. We studied whether nitrous oxide (N2O) added to a fluid allows the infused volume to be quantified by measuring N2O in the expired air during normal breathing. If so, N2O might serve as a tracer of fluid absorption during endoscopic surgery.
Methods. Twelve male volunteers received continuous and intermittent i.v. infusions (5–45 min) of fluid containing 40 ml litre–1 of N2O. Breath N2O and CO2 concentrations were measured every second via a flared nasal cannula, a standard nasal cannula, or a Hudson mask.
Results. An expression for the amount of infused fluid was obtained by calculating the area under the N2O concentration–time curve for samples representative for exhalation (CO2>median) and then dividing this area by the median CO2 for the remaining samples. The N2O method then estimated fluid volumes of between 50 and 1400 ml within a 95% prediction interval of ±200 ml. There were differences of up to 14% in results between the airway devices tested, but the volunteers preferred the flared nasal cannula. N2O showed a distinctly higher 3 min variability during intermittent infusion, which could indicate whether fluid absorption is directly intravascular or extravascular. No adverse effects were seen.
Conclusions. N2O method does not require forced end-expiratory breath sampling but still predicts an administered fluid volume with high precision. N2O variability can probably be used to distinguish immediately between intravascular and perivesical fluid absorption during surgery.
Keywords: complications, hyponatraemia; ethanol; fluids, irrigating; measurement techniques, endscopy
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Endoscopic procedures of the genitourinary tract are usually performed using an irrigating fluid containing glycine, mannitol or sorbitol to constantly rinse blood and pieces of cut tissue out of the operating field. This fluid is occasionally absorbed in volumes of 3–4 litres that give rise to overt and potentially life-threatening symptoms summarized as the ‘transurethral resection (TUR) syndrome’ which has been best described after transurethral resection of the prostate (TURP)1 2 and transcervical resection of the endometrium (TCRE).3 However, TUR syndrome is most often incomplete and develops in response to absorption of between 1 and 2 litres of irrigant, which occurs in between 1% and 10% of the TURPs performed.4 5 A recent review is available.6
There is no generally adopted method for immediate detection of fluid absorption. Measuring fluid output and input or weighing the patient need to be performed meticulously to be reasonably accurate.7 8 Another approach is to add ethanol to the irrigating solution and to use a commercial exhaled ethanol analysed to indicate fluid absorption.9 10
The purpose of this study was to evaluate, in volunteers, the principles of a novel on-line method for monitoring fluid absorption based on adding a trace amount of nitrous oxide (N2O) to the irrigating fluid. This gas is well known, is chemically inert, has low toxicity and can be measured even at very low concentrations. The amount of infused fluid was to be quantified by measuring N2O concentration in the expired air. We hypothesized that accurate results can be obtained without active exhalation and that the rapid elimination of N2O gives a distinctly different temporal profile of data during continuous and intermittent infusions, which can help to distinguish between direct i.v. and extravascular absorption (extravasation). The amount of N2O required is so small that adverse effects are quite unlikely.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Twelve healthy male volunteers, aged 19–52 (median, 26) yr and with a body weight of 65–88 kg (median, 76), gave their oral informed consent to participate in the study, which had been approved by the Ethics Committee of Human Research in Southern Stockholm.
All volunteers completed four experiments aimed at evaluating different aspects of monitoring the administration of fluid by means of expired-breath tests using N2O as a tracer. They were (i) comparing sampling sites, (ii) continuous infusion, and (iii) intermittent infusion using a low N2O concentration (40 ml litre–1) and (iv) a high N2O concentration (100 ml litre–1).
Procedure
Volunteers had a light breakfast at home and were not allowed to eat or drink during the experiments, before which they rested supine for 20 min to reach a haemodynamic steady state. A cannula was placed into the antecubital vein of the left arm for fluid infusion to which N2O for medical use (AGA Linde Healthcare AB, Sundbyberg, Sweden) had been dissolved under sterile conditions shortly (at the most 30 min) before each experiment. The N2O was taken from a gas tube and withdrawn via a 3-way infusion set into a 250 ml graded glass syringe, which was sterilized daily. The valve on the infusion set was then switched, and the withdrawn gas injected manually through a bacterial filter and a sterile needle into the irrigating fluid via the injection port.
The following four series of experiments were performed on separate occasions at least 2 days apart except for the evaluation of sampling sites, which was performed shortly after Experiment 3. All randomization was made by sealed envelopes.
Comparing sampling sites (84 infusions)
The volunteers received seven 5 min continuous i.v. infusions of lactated Ringer’s solution containing 40 ml litre–1 of N2O. Infusions were administered at a constant rate of 1 ml kg–1 over 5 min with the aid of infusion pumps (Flo-Guard 6201, Baxter Healthcare Ltd, Compton, UK).
During spontaneous ventilation, the N2O concentration was measured, in random order, from a flared nasal cannula, a standard nasal cannula and a Hudson mask. These experiments were repeated while oxygen therapy was administered through a separate flared nasal cannula at a constant rate of 2 litres of O2 min–1. In addition, specific recordings were made every 30 s at the end of forced exhalations.
The N2O was added to Ringer's since hyponatraemia and breath ethanol levels were not assessed.
Continuous infusion
Forty millilitres of N2O were added to each litre of a standard electrolyte-free irrigating fluid containing mannitol 3% and ethanol 1%, which has an osmolality of 400 mOsm kg–1 and pH 6.0 (Baxter Medical AB, Kista, Sweden). Infusion was given at a constant rate of 15 ml kg–1 over 20 min with the aid of infusion pumps to each volunteer. The total infused volume was 1146 (110) ml [mean (SD)].
Intermittent, low N2O
The 12 volunteers were randomized to receive one of three intermittent patterns of infusion of the standard irrigating fluid which contained 40 ml litre–1 of N2O. The infusion patterns were selected to represent low, intermediate and high-rate direct intravascular absorption. The amounts of fluid administered were 7.5, 12 and 15 ml kg–1.
Intermittent, high N2O
The procedure for the low N2O concentration experiment was repeated with the exception that the irrigating fluid contained 100 ml litre–1 of N2O. To still provide the same amount of gas, the infused fluid volume was adjusted to be 40% of the ‘low N2O’ experiment.
Measurements continued until the N2O concentration had returned to baseline, which occurred about 10 min after the administration of N2O had ended. However, in Experiment 2 the volunteers were followed for 100 min to allow a follow-up of the fluid balance to be made.
Rationale for the experiments
The continuous infusions were intended to simulate extravasation. Intraperitoneal infusions of N2O-containing irrigating fluid in pigs show that a continuous pattern of breath N2O develops,11 similar to that when ethanol is used as a tracer.12
The intermittent infusions mimicked absorption of irrigating fluid by the direct i.v. route during intermittent filling of the bladder, as such absorption can only occur in bursts whenever the pressure exceeds the prostatic venous pressure of about 1.5 kPa.13 The three intermittent infusion patterns used are based on actual absorption cases from our clinic in which the pressure in the bladder and fluid uptake had been measured simultaneously.14 The same data have been used previously for simulation purposes.15
A maximum amount of 600 ml of N2O can be dissolved in 1 litre of water at room temperature.16 Mixing is facilitated by letting the gas diffuse through the fluid. A bubble is created in the upper part of the fluid bag, which consists mainly of displaced N2, with which irrigating fluid is always saturated. The present concentration of only 40 ml litre–1 was chosen because concentrations exceeding 100 ml litre–1 might increase a theoretical risk of explosion during electric cutting. N2O mixes with gases formed during electric cutting, such as propane, which can gather in air pockets within the area of surgery.17 18
Measurements
During all experiments, routine monitoring comprised pulse oximetry and non-invasive blood pressure. The N2O and CO2 concentrations in the exhaled breath were measured every second by an N2O monitor (NGA 2000 MLT Analyzer, Rosemount Analytical, Hasselroth, Germany), which consists of a temperature-controlled 30 ml sampling unit with pump valves. Gas concentrations are measured by an absorbance technique using infrared signals which do not interfere with each other. For N2O, the range was 5–2000 ppm. The pump was set to a suction rate of 300 ml of air per minute and sampling was performed via a flared nasal cannula (Hudson Respiratory Care Inc., Temecula, CA, USA) except in Experiment 1, in which different sampling devices were tested. Data were stored directly on a laptop computer until N2O levels had returned to baseline.
In Experiments 2 and 3, the end-expiratory ethanol concentration was measured every 5 min. The volunteer was asked to take a deep breath and perform a single end-expiratory breath alcohol test using a Lion Alcolmeter S-D2 (Lion laboratories, Wales, UK) at 0, 5, 10, 20, 30, 40, 50, 60, 80 and 100 min. Blood samples (9 ml) were drawn on the same occasions from a separate i.v. cannula, placed in the antecubital vein of the right arm, for measurement of the blood Hb and serum sodium concentrations. Hb was analysed on a Technicon H2 (Bayer, Tarrytown, NY, USA) and electrolytes on a Hitachi Modular (Hitachi Hi-Tech Corp., Tokyo, Japan).
Calculations
Nitrous oxide
Adjustments of N2O output for CO2 level were made to account for different breathing patterns and admixture with ambient air. In the series of 12 continuous infusions, several ways of minimizing the variability in N2O were tested and compared, and we arrived at the following approach. For the end of each 5 min interval of these 12 experiments, the individual area under the concentration curve (AUC) for N2O was calculated based on the samples in which the CO2 was above the median value. The reason for discarding all values associated with a low CO2 concentration was to ensure that only the most representative data for exhalation were included. This AUC for N2O was then divided by the median CO2 for the remaining sampling points. This algorithm was then applied in all 120 experiments, and its ability to predict the infused fluid volume evaluated by residual plots for each series.
To make the low- and high-concentration experiments comparable, fluid volumes used in the latter were too low and therefore multiplied by 2.5 to compensate for the difference in N2O content between infusions.
Ethanol
For the ethanol-monitoring method, the absorbed volume over 10 min was calculated using the following previously published equation, derived during TURP:19
 |
Ethanol is the change in concentration during the same 10 min interval.
Statistics
Results were expressed as the mean (SD). Correlations were evaluated by linear regression analysis and the differences between sampling sites by repeated-measures ANOVA. Changes from baseline were studied by repeated-measures ANOVA followed by Dunnett’s test. AUC was calculated by the linear trapezoid method. The coefficient of variation (CV) for breath N2O was obtained as SD divided by the mean over time. CV was recalculated every 30 s, and the Wilcoxon test was used to compare results between continuous and intermittent infusions. P<0.05 was statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
No adverse events occurred. There were fairly small but still statistically significant differences in the CO2-adjusted AUC for the N2O concentration depending on whether sampling was done from a flared nasal cannula (reference), a standard nasal cannula (+14%) or a Hudson mask (+8%). Adding O2 via a flared nasal catheter marginally changed the AUC obtained via any of these airway devices (Fig. 1). In contrast, end-expiratory sampling yielded a 25% lower value (P<0.001). All volunteers agreed that the flared nasal cannula was the most comfortable sampling method.
|
The individual N2O time–concentration profiles had a different appearance depending on the mode of infusion (Fig. 2A). Quite variable patterns were used for the three intermittent infusions (Fig. 2B). The CV for the N2O concentration over 3 min intervals was much higher for the intermittent than the continuous infusions (mean 0.70 vs 0.26, P<0.001; Fig. 3). The 3 min interval was the shortest that still offered an excellent distinction between the two types of infusions.
|
|
The CO2-adjusted AUC for N2O did not differ significantly when the fluid volume in the high-dose experiment was multiplied by 2.5, i.e. the factor with which the N2O concentration differed between low- and high-dose experiments. The results are therefore presented together in Figure 4, which show that the AUC strongly correlated with the volume of infused N2O-containing irrigating fluid, regardless of whether the fluid was infused continuously or intermittently with a high or low N2O concentration. The 95% prediction interval was ±200 ml, and overall the data showed a normal distribution around the regression line (Fig. 5A).
|
|
Breath ethanol concentrations obtained during the continuous and intermittent infusions of irrigating fluid (Fig. 6B) slightly overestimated the known infused fluid volume. The 95% prediction interval was approximately 400 ml (Fig. 6B) which is in agreement with previous findings based on a larger number of data points.10
|
Blood Hb and serum sodium concentrations decreased by 6% during the continuous-infusion experiments (repeated-measures ANOVA, P<0.001), but only Hb was restored during the follow-up period.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The study evaluated whether it would be possible to monitor i.v. fluid administration by adding a tracer amount of N2O to the fluid and sampling exhaled air during normal breathing. An algorithm was developed based on 12 of the 120 experiments, and eventually validated on all of them by demonstrating limited scattering on residual plots (Fig. 5). This algorithm adjusted the N2O data with regard to the depth of breathing and also removed measurements performed during apparent inhalation. Since the AUC for a drug changes in proportion to the dose, we finally used the CO2-corrected AUC for N2O concentration as the expression for infused fluid volume. The correlation between N2O and infused volume then became linear and, when applied to all experiments, the 95% prediction interval for any new measurement was only ±200 ml for fluid volumes up to 1400 ml (Fig. 4). However, one must consider that the regression line describing the correlation between N2O and the actually infused fluid volume is valid only for the measuring chamber size of 30 ml and a suction rate of 300 ml min–1. The slope is likely to be slightly different with other settings.
The precision of this measurement was slightly influenced by whether breath sampling was performed via a flared nasal cannula, a standard nasal cannula, or a Hudson mask. However, since the volunteers preferred the flared nasal cannula, we concluded that further experiments should be performed using this airway device. Moreover, the results were affected only marginally by simultaneous administration of oxygen, which might be important in the operating theatre. End-expiratory sampling, however, yielded lower values than the three modes of continuous measurement. This might be explained by the fact that the measuring chamber for N2O is larger than that for CO2. The N2O chamber then partially equilibrates the concentration over time which yields a falsely low peak at the end of exhalation. The volunteers also had difficulty in providing a forced end-expiratory sample every 30 s, which probably made this mode of measuring less accurate.
The N2O technique could serve as a monitoring tool when administration of fluid may occur spontaneously in unknown volumes, such as fluid absorption during endoscopic surgery. A prerequisite would be that N2O does not diffuse through the mucosa of the urinary tract. Such diffusion was not found in pigs studied prior to these experiments (R. Hahn, unpublished data). The tracer also needs to be absorbed in the same proportion as the irrigating fluid. This is well studied for ethanol,10 and a completed but yet unpublished two-centre clinical trial shows that it is also the case for N2O during TURP. Immediate detection of fluid absorption on the operating table is of clinical value as it allows steps to be taken to prevent excessive absorption. Monitoring allows the earliest possible treatment to be instituted and the optimal level of postoperative care be selected. Our goal is to automate the calculations and have the results displayed continuously to the operating team.
Fluid absorption is rarely monitored in the operating theatre because of the poor precision or invasiveness of the available methods. In the volunteers, Hb and serum sodium decreased only by 6% in response to 1.1 litre of fluid. In the operating theatre these blood parameters are affected by factors other than fluid absorption. The most viable approaches are ethanol monitoring and continuous weighing of the patient on a bed-scale.9 Both are reasonably accurate, but the latter method can only be applied when the bladder is empty. As for Hb and sodium, the result is confounded by blood loss and i.v. infusion. These drawbacks are not shared by ethanol monitoring, which is in daily use in some European countries. The method is simple and safe but the patient must currently make an active exhalation to provide a breath sample, except in the 10–15% of patients in whom the endoscopic surgery is performed under general anaesthesia.20 Rapid fluid absorption raises blood alcohol sufficient to make the patient legally unfit to drive a car. Elimination from the blood is relatively slow.10
In contrast, N2O can apparently be monitored continuously without active involvement of the anaesthetist, the surgeon or the patient. Measuring N2O during general anaesthesia is a well-known technique and, in the awake patient, this study suggests that N2O could provide a more precise estimate of fluid absorption than ethanol. The wider prediction interval of ethanol is based on fewer observations but is in agreement with the 90% prediction interval of ±300 ml obtained for 160 TURPs.10
Another benefit is that N2O may distinguish between direct intravascular absorption and extravasation (Figs 2 and 3). The former type is most common during TURP4 5 and TCRE21 while the latter dominates during renal stone surgery.22 During TURP, intravascular absorption occurs only during the periods of time when the fluid pressure exceeds the periprostatic pressure,13 which happens during the latter part of each intermittent filling of the bladder.14 The fact that no uptake of fluid can occur during the first part of each filling15 explains why the concentration–time curve of N2O in the breath becomes intermittent. In contrast, extravasation implies that fluid is continuously being absorbed from a pool deposited in the periprostatic, retroperitoneal or intraperitoneal space. Here, any elevation of the N2O concentration is more stable.
Measurements were performed every second, as bursts of fluid might otherwise be missed because of the high clearance of N2O. Thanks to frequent sampling, the three different patterns of intermittent infusion provided data on N2O that correlated as well as the continuous infusion with the amount of infused fluid (Fig. 4). However, calculating the CV over 3 min yielded a clear difference between the two types of fluid administration (Fig. 3). While not tested in this study, extravasation results in N2O, ethanol and serum sodium responses that are only one-third of those found after intravascular absorption.10 11 Dangerous symptoms may develop before treatment is started if extravasation is mistaken for intravascular absorption.23
The concentrations of N2O reached in the present study are within the environmental pollution range (0–500 ppm). Dissociative mental effects of N2O occurs at a concentration >25%, which is 1000 times higher than the maximum concentrations measured in the present experiments (Fig. 2). However, the patient’s brain is exposed to even lower concentrations since mass balance calculations based on i.v. infusions of N2O in anaesthetized pigs indicate that nearly all gas is eliminated during the first passage through the lungs.11 This massive first-pass elimination probably makes accumulation of gas in the intestines unimportant. Exhaled N2O is quickly diluted in the ambient air, and any elevated concentration soon becomes impossible to detect. Needless to say, implementation of the method described in the clinic precludes the use of N2O as an adjunct to general anaesthesia.
After conclusion of the present experiments, a laboratory study revealed that N2O escapes from plastic fluid bags at a rate of 5% per hour (unpublished data). For clinical use, we believe that the N2O should be added to the fluid immediately before use via a small gas bomb attached to a port on the irrigating system. This arrangement would also prevent the need for more types of factory-made irrigating fluids.
In conclusion, the results from an i.v. infusion study in volunteers suggest that N2O can serve as a tracer for irrigating fluid absorption in patients undergoing endoscopic surgery. Potential benefits include practical aspects of monitoring, good precision and low toxicity. Further studies during actual surgical interventions are needed to evaluate whether these findings are also valid in clinical practice.
|
Acknowledgments |
|---|
The study was sponsored by AGA Linde Healthcare AB, Lidingö, Sweden.
|
Footnotes |
|---|
Declaration of interest. Following completion of the study, the sponsors (AGA-AB) gave the patent of the N2O method to the corresponding author. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
1 Osborn DE, Rao PN, Greene MJ, Barnard RJ. Fluid absorption during transurethral resection. Br Med J 1980; 281:1549–50
2 Rothenberg DM, Berns AS, Ivankovich AD. Isotonic hyponatremia following transurethral prostate resection. J Clin Anesth 1990; 2:48–53[CrossRef][Medline]
3 Baggish MS, Brill AIO, Rosenweig B, Barbot JE, Indman PD. Fatal acute glycine and sorbitol toxicity during operative hysteroscopy. J Gynecol Surg 1993; 9:137–43
4 Olsson J, Nilsson A, Hahn RG. Symptoms of the transurethral resection syndrome using glycine as the irrigant. J Urol 1995; 154:123–8[CrossRef][Web of Science][Medline]
5 Hahn RG, Sandfeldt L, Nyman CR. Double-blind randomized study of symptoms associated with absorption of glycine 1.5% or mannitol 3% during transurethral resection of the prostate. J Urol 1998; 160:397–401[CrossRef][Web of Science][Medline]
6 Hahn RG. Fluid absorption in endoscopic surgery (review). Br J Anaesth 2006; 96:8–20
7 Boyd HR and Stanely C. Sources of error when tracking irrigation fluids during hysteroscopic procedures. J Am Assoc Gynecol Laparosc 2000; 7:472–6[CrossRef][Web of Science][Medline]
8 Hahn RG. The volumetric fluid balance as a measure of fluid absorption during transurethral resection of the prostate. Eur J Anaesth 2000; 17:559–65[CrossRef][Web of Science][Medline]
9 Shipstone DP, Inman RD, Beacock CJM, Coppinger SWV. Validation of the ethanol breath test and on-table weighing to measure irrigating fluid absorption during transurethral prostatectomy. BJU Int 2002; 90:872–5[CrossRef][Web of Science][Medline]
10 Hahn RG. Ethanol monitoring of irrigating fluid absorption (review). Eur J Anaesth 1996; 13:102–15[CrossRef][Web of Science][Medline]
11 Hahn RG. Nitrous oxide as a marker for irrigating fluid absorption. An experimental study in the pig. Scand J Urol Nephrol 2003; 37:281–5[CrossRef][Web of Science][Medline]
12 Olsson J and Hahn RG. Simulated intraperitoneal absorption of irrigating fluid. Acta Obstet Gynecol Scand 1995; 74:707–13[Web of Science][Medline]
13 Hultén J, Bengtsson M, Engberg A, Hjertberg H, Svedberg J. The pressure in the prostatic fossa and fluid absorption. Scand J Urol Nephrol 1984; 82:Suppl., 33–43
14 Hahn RG, Berlin T, Johansson H, Lewenhaupt A. Changes in intravesical pressure during irrigating fluid absorption in transurethral prostatic surgery. Urol Res 1988; 16:281–5[CrossRef][Web of Science][Medline]
15 Hahn RG, Larsson H, Ribbe T. Continuous monitoring of irrigating fluid absorption during transurethral surgery. Anaesthesia 1995; 50:327–31[Web of Science][Medline]
16 Gas Encyclopaedia. Elsevier Scientific Publishing Company, 1976; 1056
17 Ning TC, Atkins DM, Murphy RC. Bladder explosion during transurethral surgery. J. Urol 1975; 114:536[Web of Science][Medline]
18 Iversen Hansen R and Iversen P. Bladder explosion during uninterrupted transurethral resection of the prostate. Scand J Urol Nephrol 1979; 13:211–2[Web of Science][Medline]
19 Hahn RG. Calculation of irrigant absorption by measurement of breath alcohol level during transurethral resection of the prostate. Br J Urol 1991; 68:390–3[Web of Science][Medline]
20 Stalberg HP, Hahn RG, Jones AW. Ethanol monitoring of transurethral prostatic resection during inhalation anesthesia. Anesth Analg 1992; 75:983–8
21 Olsson J and Hahn RG. Ethanol monitoring of irrigating fluid absorption in transcervical resection of the endometrium. Acta Anaesthesiol Scand 1995; 39:252–8[Web of Science][Medline]
22 Gehring H, Nahm W, Zimmermann K, et al. Irrigating fluid absorption during percutaneous nephrolithotripsy. Acta Anaesthesiol Scand 1999; 43:316–21[CrossRef][Web of Science][Medline]
23 Hahn RG. Life-threatening transurethral resection syndrome despite monitoring of fluid absorption with ethanol. Eur J Anaesth 1995; 12:431–3[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Piros, T. Fagerstrom, J. W. Collins, and R. G. Hahn Glucose as a Marker of Fluid Absorption in Bipolar Transurethral Surgery Anesth. Analg., December 1, 2009; 109(6): 1850 - 1855. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||