72 year old male with gastric outlet obstruction

On a Saturday call I was called in to take care of a 72 year old gentleman who had undergone a six hour paraesophageal hernia repair and extensive lysis of adhesions 10 days prior.  He was now suffering from abdominal pain with radiological studies that showed extensive colon dilation with stool and lack of movement from the stomach to the duodenum leading to a diagnosis of gastric outlet obstruction.  The patient had an NG tube in place which was on suction.  The patient was alert and orientated able to answer questions with no signs of obtundation or lethargy.

 Lab values were significant for the following:

• Sodium level of 150 mmol/L
• Potassium level of 3.2 mmol/L
• Chloride of 116 mmol/L
• BUN of 26.
• Magnesium level of 1.7 mmol/L
• Calcium (total) 7.9 mg/dL
• Phosphate 1.8 mg/dL
• Creatinine 0.8 mg/dL (down from 2.8  10 days earlier after AKI)

Post op labs

• Sodium down to 140 mmol/L
• Potassium level increased to 4 mmol/L
• Chloride slightly decreased to 114 mmol/L
• BUN  of 25
• creatinine 1.2 mg/dL
• Magnesium of 1.6 mmol/l
• Calcium 7.4 mg/dL
• Phosphate 4.0 mg/dL

Gastric outlet obstruction is typically considered a medical emergency not a surgical emergency as serious electrolyte abnormalities are typical along with significant alkalosis if not treated.  The standard textbook presentation of gastric outlet obstruction (GOO), is a dehydrated patient with a metabolic alkalosis, hypochloremia, hypokalemia and hyponatremia.  Initially, the kidneys will dump bicarbonate into the urine to avoid serious alkalosis and retain chloride ions. However, this process is overwhelmed as dehydration ensues.  At this point, the kidneys agressively attempt to retain fluid by holding onto sodium at the expense of secreting potassium and hydrogen ions.  This leads to a paradoxical acidic urine and further exacerbates the alkalemia and hypokalemia.  Alkalosis leads to a reduction in circulating calcium levels. This patient, despite a diagnosis of GOO, did not have the typical biochemical abnormalities associated with this syndrome.

  Firstly the patient had significant hypernatremia (Na greater than 144 mmol/L).  Clinically, the sensation of intense thirst that protects against severe hypernatremia in health may be absent or reduced in patients with altered mental status or with hypothalamic lesions affecting their sense of thirst and in infants and elderly people. Non-specific symptoms such as anorexia, muscle weakness, restlessness, nausea, and vomiting tend to occur early. More serious signs follow, with altered mental status, lethargy, irritability, stupor, and coma. Patients with chronic hypernatremia (hyperosmolar states) have adapted their brains via "idiogenic osmoles" to avoid neuronal shrinkage.  Idiogenic osmoles are organic molecules such as myo-inositol, taurine, glycerylphosphorylcholine, and betaine which are accumulated via extracellular fluid uptake via activation of sodium-dependent cotransporters.  Because of this adaptation to chronic hypernatremia, rapid correction of the free water deficit would result in rapid cerebral edema due to the relative hyperosmolarity inside the neuron.
In a systematic review, hypernatremia has been found to be associated with increased 30 day mortality and morbidity [1]. Hypernatremia may result in severe metabolic derangements, myocardial depression and injury, neurologic impairment, venous thromboembolism, and poor wound healing. Although there is an association, it is not proven that preoperative treatment of hypernatremia will improve outcome.  It is clear that the rate of change in sodioum levels is very important.  One study showed that slow rates of improvement (less than 0.25 mmol/L/hr) were typically inadequate with a 3-fold increased risk of 30 day mortality compared to those with a more rapid rate of correction (0.25 to 0.5 mmol/L/hr).

The most likely reason for this patient's hypernatremia was loss of hypotonic body fluids with addition of hypertonic TPN. Rapid and aggressive fluid volume replacement should have been started immediately to correct his likely volume deficit.    Also, the free water deficit needed to be replaced.  The formula to determine the free water deficit (L)=[0.6 x IBW] x [(current [Na+]/140)-1]. The deficit should be corrected in the short term by 1/2 or 50% of the deficit with the remainder corrected over the next 24 to 36 hours to minimize cerebral edema.

Therefore, in my patient, his estimated free water deficit at the start of the case was (in liters): 0.6 x 70kg= 42 x (150/140)-1 = 2.99 L.   During the case in order to avoid truly catastrophic cerebral edema while under GETA, I planned a very mild correction and estimated that slowly infusing 1/2 NS would help provide a small amout of free water while volume ressuscitation for the case with some 5% albumin (250 mL) and LR (3.5 L).  LR contains 130 mEq/L of sodium, 4 mEq/L of potassium, 2.7 mEq/L of Calcium, 109 mEq/L of Chloride and finally lactate 28 mEq/L.

The Adrogue Formula will allow you to determine how much the serum sodium level will change based on the fluid you infuse:  Change in serum Na+= (infused Na+ - serum Na+)/(TBW +1).  For my patient using 1/2 NS and plugging in the data, the expected change in serum sodium after the six hour case was (75 - 150)/(0.6 x 70kg)+1 = 1.74 mmol.  If my target is to reduce the serum sodium by about 0.3 mmol/hr, after 6 hours (approximate length of the case), I would want to have reduced the serum from 150 mmol/L to 148.2 mmol/L.  It turns out that I didn't use the above formulas, nor make the calculations listed above. I guestimated that giving 1 L of 1/2 NS would be safe and help reduce to some degree the amount of hypernatremia.  It turns out that the 1 L of 1/2 NS I infused over the six hour procedure was predicted to reduce the serum sodium by 1.74 mmol/L and my goal based on the above equations was to decrease it by 1.8 mmol after six hours.
In retrospect, I do not believe the goal should be focused at all on reducing serum sodium when taking a patient with chronic hypernatremia who is not suffering any symptoms, into major abdominal surgery.  This was likely an error of judgement on my part.

In fact, the patient also had LR infusing, and this was contributing to the reduction in Na+ as well.  I infused 3.5 L of LR and the estimated reduction in serum sodium from this therapy would be (130-150)/42+1 =  0.46 mmol/L x 3.5 L = 1.62 mmol total. This shows that I actually overshot my goal by using LR aggressively during the case. Therefore, when facing a patient with likely chronic hypernatremia undergoing major abdominal surgery, the best therapy is to use LR as needed.

My patient also suffered from hypomagnesemia. Magnesium (Mg2+) plays a major role in a number of pathways in the human body including serving as a co factor for a number of processes (i.e. protein synthesis, neuromuscular function etc.), is an endogenous regulator of several electrolytes, in particular it is a  calcium channel antagonist. In addition, Mg2+ is important in maintaining Calcium levels because low levels of Mg2+ result in end organ resistance to parathyroid hormone and in parathyroid secretion.  Furthermore, Mg2+ also modulates sodium and potassium currents thus affecting membrane potentials.  In the CNS, Mg2+ exerts depressant effects by antagonizing the NMDA receptor and also inhibiting catecholamine release.

Magnesium levels are primarily affected in the small bowel via absorption (passive and active transport) and re absorption in the thick ascending limb of the loop of henle.
Hypomagnesemia results mainly from inadequate dietary intake and/or GI and renal losses.  Hypomagnesemia is often associated with diarrhea, vomiting, use of loop and thiazide diuretics, ACE inhibitors, cisplatin, aminoglycosides,or other nephrotoxic drugs, and several endocrine like parathyroid diseases, hyperaldosteronism, and chronic alcoholism.  In addition, during major abdominal surgery, intraoperative crystalloid infusion may lead to a decrease in the passive magnesium transport, thus leading to a decrease in plasma magnesium concentrations.  The incidence of hypomagnesemia is as high as 65% for patients in the ICU, where hypoalbuminemia, TPN, and the use of magnesium-wasting medications are commonly present.  Patients with head injuries also seem to be at high risk for hypomagnesemia secondary to polyuria.  Hypokalemia and hypocalcemia are also frequently associated with hypomagnesemia.
There has been an effort to associate hypomagnesemia with increased mortality. Several studies have been able to link the two, however, careful analysis and high quality studies have suggested that hypomagnesemia may be an epiphenomenon of critically ill patients who suffer increased mortality.  Therefore, treating low magnesium levels is not likely to reduce mortality.  However, replenishing magnesium levels in surgical patients can have other benefits.  For example, reduction in cardiac dysrhythmias, primary atrial tachyarrhythmias, reduced anesthetic need (MgSO4 as an adjunct anesthetic/analgesic) inhibition of platelet dependent thrombosis, attenuate adverse cardiovascular effects during laryngoscopy, and intubation.  The treatment of hypomagnesemia takes several days of IV therapy because nearly all magnesium is intracellular.  In fact, in critically ill patients, 4 to 6 grams of IV Magnesium per day will likely be required to maintain a serum magnesium level of near 2 Gm/dL. In addition, IV magnesium can take up to 48 hours to redistribute into the cells, causing serum levels after IV administration to be falsely elevated.

The patient was also suffering from hypophosphatemia.  Hypophosphatemia is a very common disorder in critically ill patients occurring in up to 100% of patients in some studies.  Particularly relevant to this patient, TPN can result in sudden and profound hypophosphatemia.  This usually is due to providing TPN to patients who are malnourished, have initial low phosphate levels and then suddenly receive glucose via TPN which supports a sudden increase in the formation of ATP.  With TPN initiation, a sudden acute rise in glucose transport across cell membranes occurs accompanied by oxidative phosphorylation resulting in a large demand on intracellular phosphate for support of ATP production.  This results in "refeeding syndrome" where rapid changes in fluids and electrolytes occurs.  Commonly, this changes result in hypophosphatemia, hypomagnesemia and hypokalemia.  Low phosphate levels can have wide ranging clinical effects including myocardial dysfunction, diaphragmatic weakness, seizures, coma, rhabdomyolysis, and red blood cell dysfunction from tissue hypoxia (by decreased RBC 2,3-DPG).  Furthermore, there is some evidence that hypophasphatemia is associated with longer ICU and hospital stays and increased risk of arrythmias. A recent retrospective analysis found that hypophosphatemia was associated with increased 28 day mortality in the ICU [2].  The authors of this study concluded based on their analysis that independent of illness severity, hypophosphatemia still resulted in increased mortality. This patient had significant hypophosphatemia, and I elected to replace phosphate during the case.  It is important to consider the calcium levels of patients who receive phosphate repletion because the calcium level will likely decrease as phosphate is replaced.
My patient went to the ICU intubated post op.  No problems were evident related to his severe electrolyte abnormalities.  He required prolonged post operative care in the ICU with mechanical ventilation. However, after about two weeks he was extubated and made progress.

Memorizing the dosing of all electrolyte replacements is tedious and unnecessary. I have included a simple calculator that can be stored on one's home screen of their smart phone for easy use to determine the appropriate dose to give a patient assuming IV administration.

Lyte Calculator

1.  Leung AA, McAlister FA, et al. Amer J Medicine. V. 126(10): 2013. p 877-885
2. Wang L, Xiao C, Chen L, Zhang X, Kou Q. BMC Anesthesiology. 86;2019.

preoperative hypokalemia, replace and proceed or cancel?

The other morning, while I was in the middle of my first of three scheduled cases I received a call from a nurse to report that the K+ level was 2.8 mEq/L on my next scheduled patient.   The patient was a 76 year old female who had a history of Afib, CHF, CAD. Previously, an AICD was placed for unclear reasons.  The patient stated that she believed the reason for the AICD was due to her poor cardiac function.  She reported no history of previous ventricular fibrillation/tachycardia.  She was on diuretics, with oral potassium supplements.

Basic Science review: Potassium homeostasis
Total body stores of K+ are approximately 3000 to 4000 meq with 98% of this located inside the cells.   This concentration differential is maintained by the Na+-K+ ATPase pump, located in the cell membrane which pumps out 3 Na+ ions for every 2 K+ions pumped in.  K+ plays an important role in cell metabolism, and therefore, a myriad of cellular functions deteriorate with an imbalance of K+ ion concentration. For example, significant hypokalemia can result in polyuria due to a reduced sensitivity to ADH.  Also critical is the intracellular to extracellular concentration gradient which largely determines resting membrane potential.

Em=-61 log r[K+]cell+0.01 [Na+]cell / r[K+]ecf + 0.01[Na+]ecf

Alterations in relative concentrations of K+ inside versus outside cells can significantly alter the resting membrane potential leading to cardiac arrhythmia as well as other skeletal muscular symptoms.

The maintenance of whole body potassium  stores is handled by the kidney, in particular by the principal cells of the cortical collecting tubule.  Here aldosterone (secreted in response to a minimal rise in serum potassium) stimulates increased activity of the Na+-K+ ATPase pump located in the basolateral membrane of the principal cell.   This pump pulls K+ from the extracellular space and pumps it into the principal cell while exchanging this for Na+ (3 Na+ : 2 K+ ratio).  As the intracellular K+ ion concentration increases inside the principal cells,  K+ passively leaves the principal cell into the collecting tubule lumen as it follows its concentration gradient. Also in the collecting tubules are located the intercalated cells.  These cells work to reabsorb K+ (exactly opposite the Principal cells).  These cells are particularly important in hypokalemic patients where an H+-K+ ion pump located in the luminal membrane actively exchanges intracellular H+ ions for K+ ions, allowing K+ to be reabsorbed back into the plasma.  As mentioned above, ADH, plays a role in K+ ion secretion by increasing the number of luminal K+ ion channels in the collecting tubules. The K+ ion itself can mimic all of the changes in the kidney that aldosterone initiates.  In other words, elevated K+ levels produce an aldosterone like effect where K+ excretion and Na+ reabsorption are enhanced in the principal cells.  This is due to increased luminal membrane permeability to Na+ and K+ (by increasing the number of open channels for passive diffusion) and increased activity of the Na+-K+ ATPase pump located on the basolateral membrane.

Also critical to K+ homeostasis is the urine flow rate in the cortical collecting tubule.  There are two mechanisms that allow a greater urinary flow rate in the cortical collecting tubule to increase net K+ secretion leading to hypokalemia.  1) increased flow lowers the intraluminal (inside the collecting tubule) K+ concentration, favoring the passive movement of K+ out of the principal down its concentration gradient via the K+ channels in the luminal membrane of the principal cells. 2) increased flow through the kidney brings more Na+ to the collecting tubule leading to increased Na+ reabsorption in the collecting tubule causing the lumen to become more electronegative favoring passive K+ diffusion to maintain electroneutrality.  Furthermore, as more Na+ is reabsorbed into the principal cell, the Na+-K+ ATPase pump removes the Na+ from the cell back into the body, leading to increased K+ entry into the principal cell which increases the intracellular K+ concentration of the principal cells of the collecting tubule. Increased distal flow of urine in the collecting tubule seems to be the mechanism by which the loop and thiazide diuretics induce hypokalemia. These agents increase distal flow by diminishing Na+ and water reabsorption in the loop of Henle and distal tubule respectively.

Clinically, chronic K+ alterations from so called normal levels are not as likely to cause outward clinical symptoms.  This is because, the intracellular to extracellular gradient is the more important than the overall measured serum concentration.   Nevertheless, hypokalemia does result in symptoms even if chronic in nature.  When considering the hypokalemic patient in the immediate preoperative period, it is important to consider three main things:  1)  degree of hypokalemia, 2) Invasiveness of surgery and 3) patient's co morbid conditions (i.e. concomitant coronary artery disease or congestive heart failure).   Per Miller’s Anesthesia, p. 1107, “As a rule, all patients undergoing elective surgery should have normal serum potassium levels.  However, we do not recommend delaying surgery if the serum potassium level is above 2.8 mEq/L or below 5.9 mEq/L, if the cause of the potassium imbalance is known, and if the patient is in otherwise optimal condition.”

Patients without underlying cardiac disease are unlikely to suffer myocardial effects, even at levels below 3.0 mEq/L. However, those with ischaemic heart disease, heart failure or left ventricular dysfunction are at risk of arrhythmias with only mild or moderate hypokalaemia. This fact was highlighted in 1981 when Hulting followed patients admitted for treatment of an MI.  This paper showed that patients who had a baseline risk of arrhythmia of 3.5% increased to 8% if their serum K+ was less than 3.5 mEq/L. They also found that no patients suffered arrhythmias if their serum potassium was greater than 4.6 mEq/L.

In patients undergoing very quick surgeries such as cataract, egd, etc, it is routine to not check labs in the first place so that hypokalemia if it existed would be unknown to the provider.  In patients who require diuretics undergoing intermediate or high risk surgery, checking potassium levels is common and hypokalemia should prompt a decision tree based on the degree of hypokalemia, patient co morbidities, and type of surgery.

By far the most common cause of hypokalemia in the preoperative period is diuretic use. Diuretic induced hypokalemia has been associated with an increased rate of arrhythmias [4]. Furthermore, diuretic therapy in hypertension and heart failure has been associated with an increased rate of arrhythmic death that can be prevented by a K+ sparing diuretic and therefore, may be related to K+ depletion [5,6].  In the Framingham Heart Study [10], they reported increased frequency of PVCs to be associated with hypokalemia.   They estimated that the arrhythmia increased by 27% wit each 0.5 mEq/L decrease in K+ level. Understanding diuretic physiology makes sense given its importance in both inducing and treating hypokalemia.  A brief primer on diuretics follows:

Loop Diuretics (furosemide, bumetanide, torsemide, and ethacrynic acid)

• may lead to the excretion of 20 to 25% of filtered Na+ at max doses.
• Act in thick ascending limb of loop of henle. 
• loop diuretics inhibit Na+ reabsorption by binding to Cl- site of the Na+-K+-2Cl- carrier on luminal membrane.  The carrier only works when all four sites are occupied.
• Secondarily inhibit Ca2+ reabsorption which is passive down an electronegative gradient by the absorption of Na+/K+, which leads to a caliuresis.
• Ca2+ effects make loop diuretics excellent method of treating hypercalcemia when combined with saline loading.

Thiazide type diuretics

• primarily inhibit NaCl transport in the DCT
• at max doses only inhibit up to 3 to 5% of filtered Na+ (far less potent than loop diuretics).
• Diuresis offset by increased reabsorption in the cortical collecting tubule.
• Thiazide type diuretics also compete with Cl- at the Na+-Cl- cotransporter
• In contrast to loop diuretics, thiazide type diuretics can increase the reabsorption of Ca2+ in the DCT and early collecting tubule which may be useful in the treatment of recurrent kidney stones due to hypercalciuria.

Potassium Sparing Diuretics (Amiloride, Spironolactone, Triamtere)

• Act in principal cell in the cortical collecting tubule where Na+ reabsorption occurs passively through Na+ channels (which are increased via aldosterone)
• amiloride and triamterene directly decrease the number of open Na+ channels decreasing Na+ reabsorption leading to decreased K+ (and H+) secretion.  
• Spironolactone competitively inhibits aldosterone resulting in same result.
• Weak natriuretic effect (1 to 2% of filtered Na+).
• Amiloride is very effective in the treatment of polyuria/polydipsia from lithium-induced nephrogenic diabetes insipidus where the tubular cells of the collecting ducts become insensitive to ADH from the accumulation of lithium.
• Triamterene is a potential nephrotoxin, possibly leading to crystalluria and cast formation which in severe cases has lead to renal failure particularly if given in conjunction with NSAIDs.

 In a patient with hypokalemia not taking any loop or thiazide diuretics, other causes should be considered.  The next most common cause would be increased GI losses, either from diarrhea, vomiting or NG suction.  Magnesium levels should also be checked.  The loop and thiazide diuretics also result in Mg2+ wasting, and hypomagnesemia promotes K+ wasting. The mechanism whereby Mg2+ effects K+ homeostasis is unclear.  However, replacement with MgSO4 (as is common in the perioperitve period) could be problematic.  The sulfate acts as a non resorbable anion in the collecting tubule (leading to a greater negative intraluminal charge).  The negative charge in the lumen prevents the passive diffusion of K+ out of the lumen into luminal cells of the collecting tubules leading to increased K+ losses. Repletion with magnesium chloride or magnesium lactate would avoid this problem. 

In the patient who is hypokalemic prior to anesthesia, it is important for the anesthesiologist to consider all of the ways in which acute shifts of K+ into the intracellular space may occur exacerbating the pre existing hypokalemia.  Avoidance of respiratory or metabolic alkalemia. is very important since alkalemia results in a transfer of H+ ions from the intracellular space into the plasma to counter act a rising pH.  To preserve electroneutrality, Na+ and K+ ions  enter cells.  In general the plasma concentration falls less than 0.4 meq/L for 0.1 unit increase in pH.  Therefore, a patient with a serum potassium of 3.1 meq/L who develops a respiratory alkalosis to a pH of 7.5 due to inadvertent hyperventilation, will potentially see an acute intracellular shift of K+ ions leading to a serum K+ concentration of 2.7 meq/L. 

Insulin directly stimulates the entry of K+ ions into skeletal muscle and hepatic cells via a Na+-K+ ATPase pump. Thusly, it would be important to avoid using a dextrose solution to replace potassium, as the dextrose can stimulate insulin release causing a paradoxical further decrease in the serum potassium concentration.

The Na+-K+ -ATPase pump is also stimulated by beta 2 adnergic receptors.  This is a particular concern in the preoperative period when stress related events causes a surge in catecholamines. In fact, a catecholamine surge can acutely lower the plasma K+ concentration by approximately 0.5 to 0.6 meq/L.  This large change in serum K+ ion concentration may be partially due to the effects of insulin which is secreted in response to increased B2 adrenergic activity. Prompt recognition of this may be treated by adequate opioids +/- non selective beta blockers (i.e. propranolol). Administration of beta agonists such as albuterol for a breathing treatment, may induce a 0.5 to 1 meq/L drop in serum K+ concentration.

Clinical studies looking at preoperative K+ levels and patient outcomes

One study done on patients with coronary artery disease who underwent non-cardiac surgery suggested that a pre-operative serum potassium level of less than 3.5meq/ L was independently associated with peri-operative mortality. Other studies have failed to find any increased incidence of arrhythmia in patients at high risk (major vascular or cardiac surgery with cardiac disease) who also had significantly decreased K+ levels [2].  However, this study was probably under powered.  They examined only 447 patients, and of these only 9% had significant hypokalemia (less than or equal to 3.0 mEq/L). In contrast to this, another study of 2402 patients undergoing CABG, were followed for a variety of outcomes.   In this study, a serum potassium level less than 3.5 mEq/L was a predictor of serious perioperative arrhythmia (OR 2.2), intraoperative arrhythmia (OR 2.0) and post operative atrial fibrillation/flutter (OR 1.7) [8].  In another study looking at patients undergoing non cardiac surgery were analyzed for predictors of preoperative myocardial infarction (PMI) or cardiac death. They found that  among several risk factors hypokalemia (serum level less than 3.5 mEq/L) was identified as a predictor of these outcomes [9]. Myocardial ischemia seems to be a significant risk factor  leading to arrhythmias in the setting of hypokalemia [3,7]. Therefore, it becomes particularly important to monitor potassium levels in any patient at significant risk for preoperative ischemia.

Repletion of K+ prior to surgery is fraught with problems due to the logistics of K+ administration.  KCl is painful in peripheral IVs and can cause severe phlebitis.  In my patient, she could not tolerate KCl being infused in her peripheral IV at a concentration of 20 mEq per 100 mL any faster than 50 mL per hour (or 20 mEq per 2 hour time period).  Therefore, it was decided to delay surgery until later that afternoon to allow adequate time for her to receive 40 mEq (requiring 4 hours) and to recheck her potassium level.  After 5 hours, we returned to perform her surgery.  Her repeat potassium was 3.3 mEq/L.  She underwent GETA with Sevoflurane.  Because she had a medtronic AICD, defibrillator pads were placed prior to surgery and a magnet was placed over the device to disable anti tachycardia therapy.  The anesthetic was uneventful and she was observed overnight in the hospital.



1.  Shah, K.B., Klienman, B.S., Rao, T.L., Jacobs, H.K., Mestan, K. and Schaafsma, K. (1990) Angina and other risk factors in patients with cardiac diseases undergoing non-cardiac operations. Anesth. Analg., 70, 240-247. 

2.Hirsch IA1, Tomlinson DL, Slogoff S, Keats AS.  Anesth Analg. 1988 Feb;67(2):131-6.
3.  Hulting, J. (1981) In hospital ventricular fibrillation and its relation to serum potassium. Acta Med. Scand. Suppl., 647, 109-116.
4. Kuller LH, Hülle SB, Cohen JD, Neaton J. Circulation 73:114, 1986
5. Siscovick DS, Raghunathan TE, Psaty BM, et al. N Engl J Med 330:1852, 1994
6. Pitt B, Zannad F, Remme WJ, et al. N Engl J Med 341:709, 1999
7. Nordrehaug JE, von der Lippe G. Hypokalaemia and ventricular fibrillation in acute myocardial infarction.  Br Heart J.1983;50:525-529.
8. Wahr JA, Parks R, Boisvert D et al. JAMA 1999; 28(23):2203-10.
9. Shah KB, Kleinman BS, Rao TL, Jacobs HK, Mestan K, Schaafsma K.  Anesth Analg.1990;70:240-247.
10. Tsuji H, Venditti FJ Jr, Evans JC, et al. Am J Cardiol 74:237, 1994