Diabetic Ketoacidosis: Still Serious After All These Years

Emerg Med 39(11):10, 2007

By Joel Kravitz, MD, FACEP, FRCPSC

More than 80 years after the discovery of insulin, diabetic ketoacidosis remains one of the greatest threats to diabetic patients. The author explores the management controversies and treatment recommendations.

Diabetic ketoacidosis (DKA) is one of the most serious acute complications of diabetes mellitus. Until the discovery of insulin by Banting and Best in 1922, the mortality associated with this condition was nearly 100%. Mortality improved markedly thereafter, however, dropping to 29% within 10 years. The current level of mortality is approximately 5%.

Despite our better understanding of the disease and aggressive patient education, DKA remains a concern, appearing in 5% to 10% of all discharge summaries of diabetic patients. It’s also costly: based on an average of 100,000 admissions per year, DKA alone is responsible for over $1 billion in hospital costs. Studies have estimated the combined direct and indirect costs of medical care for diabetes and its complications at $132 billion annually. Epidemiologic data indicate that one out of every four health care dollars spent on medical care for type 1 diabetes is spent on the treatment of DKA; this ratio rises to a staggering one out of every two dollars in patients with recurrent episodes of DKA.

The American Diabetes Association (ADA) has published position statements on the treatment of DKA and another of the most common severe complications of diabetes: the hyperglycemic hyperosmolar state (HHS). These comprehensive recommendations form the basis of most physicians’ treatment protocols for DKA and HHS. However, some variations still exist in physicians’ practices with respect to diagnosis, screening, and management. In this article, I’ll review the data on some of the current controversies in the management of the patient with DKA.

URINALYSIS TO SCREEN FOR DKA

Hyperglycemia is a common complaint in the emergency department, yet many of these patients present with neither acidosis nor ketosis. The emergency physician is often faced with the challenge of determining which hyperglycemic patients are in DKA and which are not. While it may be fairly simple to identify the most severe cases by virtue of the overall clinical picture (for example, by the presence of tachycardia and altered mental status), those patients who have stable vital signs and only mild symptoms pose a greater diagnostic difficulty.

Many clinicians use urine dip ketone analysis as a screening tool to identify patients with DKA. Urinary ketones are, in the majority of cases, detected by the nitroprusside method. Although beta-hydroxybutyrate (BHB) is the more prevalent ketoacid in DKA, the urine dip ketone test relies on the reaction of acetoacetate (ACA) with the nitroprusside reagent.

As acidemia develops in DKA, acetyl-CoA molecules undergo beta-oxidation, initially being converted to ACA before moving through the metabolic pathway to become BHB. With no other contravening conditions, the ratio of BHB to ACA is approximately 3:1. This ratio depends at least in part on the ratio of nicotinamide adenine dinucleotide (NADH) to NAD+ (the redox state). As this ratio rises (favoring NADH), BHB production is also favored, NAD+ being a cofactor in the conversion of acetyl-CoA to ACA. For example, in alcoholic ketoacidosis, the redox state favors NADH, so the ratio of BHB:ACA is elevated to approximately 7:1, which explains why alcoholic patients can present with ketoacidosis in the absence of urinary ketones.

It has been proposed that the urine dip ketone test could yield a negative result in the presence of DKA. Many now believe this to be a more theoretical concern, as the level of ketosis is often underestimated. A supervening lactic acidosis may also produce a false negative for ketones on the urinalysis, because lactic acid drives the chemical reaction to the right by the law of mass action, thus favoring BHB formation. Despite the hypothetical inaccuracy, current evidence seems to suggest that it may be feasible to exclude DKA using urine ketone testing. For example, according to a retrospective study of 146 patients presenting with hyperglycemia to an urban emergency department, urinary ketones showed a sensitivity of 97% for the presence of DKA.

The largest study to address this question looked at almost 700 patients with diabetes who presented to the emergency department with hyperglycemia. The researchers compared the sensitivity, specificity, and predictive value of the urine dip ketone test with the anion gap (when greater than 16) and serum bicarbonate tests in terms of their abilities to detect either ketones, acidosis, or both. They found that the sensitivity of the urine dip ketone test, at 99%, outperformed both the anion gap (92%) and serum bicarbonate (84%). The study also revealed that the urine dip ketone test was more efficient at identifying patients with ketosis in the absence of acidosis, suggesting that the test may be able to pick up diabetic patients earlier on in the spectrum of the disease (that is, before acidemia develops).

This study did not exclude the hyperosmolar nonketotic state, seen in states of severe hyperglycemia. The authors also say that using urinalysis to screen hyperglycemic patients would have resulted in cost savings for almost two-thirds of the patients in the study if no further testing had been initiated.

ARTERIAL VERSUS VENOUS SAMPLING

Most reviews and expert opinions recommend immediately measuring arterial blood gas (ABG) as partof the normal laboratory evaluation of the DKA patient. However, the procedure can be quite painful for the patient, leading some researchers to suggest alternatives for collecting information on acid-base status.

In the mid-1980s, a small study compared the pH values of arterial and capillary blood in patients with DKA. While the two samples differed, the correlation coefficient was 98% for the pH range of 6.77 to 7.29. The difference between the two samples ranged from –0.09 to +0.02. The main disadvantage of this study was its size.

More recently, researchers assessed the effect of ABG analysis on emergency physicians’ decision-making in treating patients with DKA, as well as the correlation of venous and arterial pH values in these patients. Treating physicians were blinded to both arterial and venous pH values prior to initiating treatment for DKA, but they were prompted to change treatment in only 2.5% of the 200 patients evaluated. What’s more, the changes made were mainly in the route of administration of insulin; bicarbonate therapy was not initiated in any of the cases.

This study also found a correlation coefficient of 0.95 between arterial and venous pH samples. Subsequent bias value analysis revealed that a venous pH value would typically be less than an arterial value by an average of 0.015—an amount that would likely not alter treatment decisions.

These findings corroborate other work suggesting that venous pH can be used effectively in the evaluation and management of DKA. Recently, more review articles have suggested that venous pH can be used as an alternative to ABG analysis. Most experts still agree, however, that if information on oxygenation is specifically required, the ABG is the test of choice. Moreover, venous blood gases restrict the clinician’s ability to identify a mixed acid-base derangement.

The table below summarizes the key diagnostic criteria in the evaluation of DKA.

INSULIN THERAPY: TO BOLUS OR NOT TO BOLUS

Insulin has always been a mainstay of DKA treatment. It was not until the late 1970s that prospective studies challenged the myth that high doses of insulin were necessary to achieve effective glycemic control. Now, the pendulum has swung in the opposite direction, with the emphasis on fluid and electrolyte replacement in conjunction with low-dose insulin regimens. The theory is that fluid hydration and electrolyte repletion help dilute the effects of counter-regulatory hormones, thus rendering tissues more responsive to lower doses of insulin.

The ADA recommends treating adult DKA patients with a bolus dose of regular insulin (0.15 U/kg), followed by a continuous regular insulin infusion (0.1 U/kg/h). The ADA position statement also recommends adjusting the insulin infusion until a slow, steady glucose decline of 50 to 75 mg/dl/h is achieved. It seems counterintuitive to recommend the administration of a bolus dose of insulin when a slow reduction in glucose levels is desired. Many emergency physicians, in fact, do not administer bolus doses of regular insulin in treating DKA, believing that a smoother correction is achieved with fluid resuscitation and a continuous insulin infusion.

Some of the early studies that changed the treatment paradigm to low-dose insulin protocols did indicate that ketones were cleared more quickly when a bolus dose of insulin was given. However, others showed that the time to revert to metabolic “normality” was the same regardless of whether or not an insulin bolus was administered.

A small pediatric series revealed no difference in the rate of glucose lowering between bolus and nonbolus study groups during the first hour of therapy. The time needed to reach a serum glucose level of 250 mg/dl was similar in both groups, as was the required duration of the insulin infusion.

At my institution, ongoing studies are reexamining the utility and efficacy of bolus-dose insulin in the treatment of DKA in adults.

ALTERNATE ROUTES OF INSULIN THERAPY

Conventional treatment protocols advocate the use of a continuous infusion of regular insulin to treat the DKA patient. Occasionally, however, the emergency physician is confronted with a patient in a milder state of DKA but is forced to transfer the patient to the ICU because of hospital policy on the use of insulin infusions.

Recent studies have shown that milder forms of DKA can be successfully treated with subcutaneous injections of insulin aspart or lispro hourly (or every two hours). One study, for instance, demonstrated that hourly injections of subcutaneous regular insulin resolved ketoacidosis and controlled hyperglycemia just as quickly as an intravenous insulin infusion in patients with mild DKA. These studies, however, had small samples and excluded patients with any severe concomitant medical illness. Only patients with mild DKA were included; patients with hypotension, renal failure, cardiac ischemia, and other complications were not involved. Until larger studies are performed, it is prudent to restrict the use of subcutaneous insulin protocols to the milder DKA patients since frequent electrolyte and pH monitoring can outstrip the nursing resources of a general medical unit.

BICARBONATE THERAPY

Bicarbonate has been recommended to improve pH in order to reverse the acidosis-induced impairment in myocardial function, reduce the already elevated ventilatory drive, and also potentially improve tissue responsiveness to insulin administration. Recommendations from the ADA had advocated sodium bicarbonate to help reverse the metabolic acidosis of DKA, but these have changed in recent years based on available evidence.

Administration of sodium bicarbonate carries the risk of several deleterious effects, including paradoxical central nervous system (CNS) acidosis, potential hypokalemia due to overly rapid intracellular potassium flux, and a shift of the hemoglobin-oxygenation curve to the left resulting in impaired oxygen delivery. Studies have suggested that bicarbonate infusion may delay the recovery in blood ketone levels when compared with saline-infusion controls.

Two other retrospective series have argued against the routine use of bicarbonate in the treatment of DKA. Green and others conducted a review of more than 100 pediatric DKA admissions, intentionally selecting for greater severity. Patients with a pH above 7.15 or a serum glucose below 300 mg/dl were excluded. Approximately half the patients received bicarbonate. The rate of complications was no different between the two groups. While the rate of serum bicarbonate improvement was faster in the bicarbo-nate group, there was also a trend toward longer hospitalizations (a trend that became statistically significant in the matched pair analysis). The average pH in both groups ranged from 7.02 to 7.06, suggesting bicarbonate use for a pH above this level does not change outcome.

Viallon and colleagues evaluated the use of bicarbonate in two groups of severe DKA patients in 1999. The pH range for both the group that received bicarbonate and the group that did not was between 6.9 and 7.1. The researchers found no difference in the time to normalization of both pH and serum glucose between the two groups.

No controlled studies have found an advantage in administering bicarbonate to patients with DKA whose blood pH is 6.9 or higher. However, there likely exists a point on the metabolic spectrum where the hazards of severe acidemia are of greater concern than those of bicarbonate administration. End-organ responsiveness to catecholamines declines with progressive acidemia and becomes more pronounced with a pH below 7.2. With progressive acidemia, the normal tachycardia seen in these physiologically stressed states evolves into a brady- cardia, due at least in part to acid’s inhibition of acetylcholinesterase and its blockage of the effects of cathecholamines on the heart. Further, protons exert a more direct negative inotropic effect on the myocardium, inhibiting myocardial calcium uptake.

Some researchers have questioned the true severity of the potential adverse effects of bicarbonate administration. Two studies in the 1970s suggested that CNS acidification is, to some degree, expected during the reversal of DKA and that it has little influence on the patient’s ultimate outcome. The administration of insulin results in an overall net utilization of ketones, which should counterbalance any possible delayed ketone clearance caused by the bicarbonate. Others have argued that the hypokalemia caused by bicarbonate administration is easily avoided with close, vigilant observation of the patient’s electrolytes and that “overshoot alkalosis” is a rare event, even in the most aggressive management protocols.

Current clinical evidence, therefore, points away from the use of bicarbonate in the treatment of DKA for patients with a pH above 6.9. The ADA recommendations are likely set at this level because, to date, no studies have been done on patients with an acidosis below that level. Given the metabolic evidence for the adverse effects of acidemia cited above, it seems prudent to start a bicarbonate infusion in DKA patients with acidemia below a pH of 6.9 until evidence emerges to suggest otherwise.

RISK OF LIFE-THREATENING HYPOKALEMIA

While many patients with DKA may have normal or even elevated serum potassium levels, their bodies’ potassium stores have been moved extracellularly due to the acidosis and subsequently lost as a result of the osmotic diuresis, resulting in a total body depletion of potassium on the order of 4 to 5 mEq/kg. Hypokalemia is the most life-threatening electrolyte disturbance seen in the course of DKA treatment. Insulin itself plays a dual role in the development of hypokalemia in these patients; not only does the correction of hyperglycemia and acidosis play a role, but insulin itself is also capable of mediating the intracellular flux of potassium. These effects, coupled with the dilutional effects of fluid repletion and volume expansion and the potential hypokalemic effects of bicarbonate administration (when appropriate), make careful monitoring of potassium levels in the DKA patient a crucial component of therapy.

Potassium should not be added to the initial volume used for resuscitation and should not be given until urine output has been established (see table below). The largest errors in treatment occur at lower potassium levels. At a potassium level below 3.3 mEq/L, insulin therapy should be held and potassium administration begun (or increased if already started).

PHOSPHATE REPLACEMENT

Like potassium, phosphate is an intracellular ion that is shifted extracellularly and is depleted by the osmotic diuresis of the hyperglycemic state. Phosphate levels, like potassium levels, also drop in response to insulin therapy. The ADA recommendations give somewhat conflicting information with respect to phosphate administration. In the discussion of potassium replacement, they advise giving one-third of the potassium as potassium phosphate. However, in the section discussing phosphate depletion, they state that routine phosphate administration is unnecessary.

Severe hypophosphatemia can cause muscle weakness and hemolytic anemia and can theoretically shift the oxygen dissociation curve to the left by decreasing 2,3-DPG levels in red blood cells. Symptomatic hypophosphatemia (less than 1 mg/dl) in DKA is rare; the vast majority of DKA patients manifest mild to moderate hypophosphatemia at levels unlikely to cause symptoms. While few studies exist on the incidence of hypophosphatemia, the largest revealed only a 0.43% incidence of low phosphate levels in patients admitted to a general medical unit. This percentage may be slightly higher for some patient populations, including DKA patients.

Randomized studies have not shown any benefit to the routine administration of phosphate in DKA patients. Further, administration of large doses of phosphate can cause hypocalcemia-induced tetany and soft tissue calcifications. Phosphate should be replaced if the serum level is less than 1 mg/dl or if the patient shows signs of hypophosphatemia. Supplementation can be given as potassium phosphate.
The box below lists general principles to keep in mind regarding the treatment of DKA.

COMPREHENSIVE ALGORITHM

Current ADA recommendations for the treatment of DKA provide a comprehensive treatment algorithm for the severe metabolic derangements that ensue. Still, research data regarding some of the questions surrounding available diagnostic and therapeutic options continues to accumulate, requiring the emergency clinician to be aware of emerging evidence to provide optimal treatment for these patients.

back to top

Suggested Reading Basu A, et al.: Persisting mortality in diabetic ketoacidosis. Diabet Med 10(3):282, 1993. Chiasson JL, et al.: Diagnosis and treatment of diabetic ketoacidosis and the hyperglycemic hyperosmolar state. CMAJ 168(7):859, 2003. Green SM, et al.: Failure of adjunctive bicarbonate to improve outcome in severe pediatric diabetic ketoacidosis. Ann Emerg Med 31(1):41, 1998. Hale PJ and Nattrass M: A comparison of arterial and non-arterialized capillary blood gases in diabetic ketoacidosis. Diabet Med 5(1):76, 1988. Hendey GW, et al.: Urine ketone dip test as a screen for ketonemia in diabetic ketoacidosis and ketosis in the emergency department. Ann Emerg Med 29(6):735, 1997. Kelly AM: The case for venous rather than arterial blood gases in diabetic ketoacidosis. Emerg Med Australas 18(1):64, 2006. Kitabchi AE and Wall BM: Diabetic ketoacidosis. Med Clin North Am 79(1):9, 1995. Kitabchi AE, et al.: Hyperglycemic crises in diabetes. Diabetes Care 27 Suppl 1:S94, 2004. Kitabchi AE, et al.: Management of hyperglycemic crises in patients with diabetes. Diabetes Care 24(1):131, 2001. Lindsay R and Bolte RG: The use of an insulin bolus in low-dose insulin infusion for pediatric diabetic ketoacidosis. Pediatr Emerg Care 5(2):77, 1989. Ma OJ, et al.: Arterial blood gas results rarely influence emergency physician management of patients with suspected diabetic ketoacidosis. Acad Emerg Med 10(8):836, 2003. Morris LR, et al.: Bicarbonate therapy in severe diabetic ketoacidosis. Ann Intern Med 105(6):836, 1986. Okuda Y, et al.: Counterproductive effects of sodium bicarbonate in diabetic ketoacidosis. J Clin Endocrinol Metab 81(1):314, 1996. Riley LJ Jr, et al.: Alkali therapy of diabetic ketoacidosis: biochemical, physiologic, and clinical perspectives. Diabetes Metab Rev 5(8):627, 1989. Schwab TM, et al.: Screening for ketonemia in patients with diabetes. Ann Emerg Med 34(3):342, 1999. Umpierrez GE, et al.: Efficacy of subcutaneous insulin lispro versus continuous intravenous regular insulin for the treatment of patients with diabetic ketoacidosis. Am J Med 117(5):291, 2004. Umpierrez GE, et al.: Treatment of diabetic ketoacidosis with subcutaneous insulin aspart. Diabetes Care 27(8):1873, 2004. Viallon A, et al.: Does bicarbonate therapy improve the management of severe diabetic ketoacidosis? Crit Care Med 27(12):2690, 1999.

Printable version of this article

E-mail this article