Bongard Frederic , Darryl Sue. Diagnosis and Treatment Critical Care

Подождите немного. Документ загружается.



DIABETES MELLITUS, HYPERGLYCEMIA, & THE CRITICALLY ILL PATIENT

587

Acetoacetate crossreacts in some automated methods for

measuring creatinine and thus may artificially elevate creati-

nine concentrations; if owing to this mechanism, elevated

serum creatinine concentrations diminish rapidly as acetoac-

etate levels are cleared with therapy.

Hypertriglyceridemia occurs quite frequently in diabetic

ketoacidosis. This is of importance not only because of the

clinical manifestations of hypertriglyceridemia (eg, acute

pancreatitis) but also because high triglyceride levels inter-

fere with accurate measurement of sodium, chloride, and

bicarbonate. When hypertriglyceridemia is present, pseudo-

hyponatremia may be seen (serum is composed of a sodium-

containing water phase and a large lipid-triglyceride phase

that contributes non-sodium-containing volume). What is

less well recognized is that hypertriglyceridemia can interfere

with the colorimetric measurement of chloride and bicar-

bonate, although there is no interference with more com-

monly used ion-specific electrode assays. The effect is to

report falsely high serum chloride and bicarbonate concen-

trations that may mask the increase in anion gap expected in

ketoacidosis and lead to incorrect evaluation of the patient.

This possibility should be considered whenever the anion

gap is not elevated in an otherwise typical clinical situation

suggestive of diabetic ketoacidosis.

Differential Diagnosis

Clinical manifestations of diabetic ketoacidosis, such as dys-

pnea, nausea, vomiting, or abdominal pain, may mimic non-

diabetic acute disease. In known diabetic patients with coma

or altered mental status, ketoacidosis must be distinguished

from hypoglycemia. This is usually not difficult clinically

because the circumstances and clinical findings are usually

quite different. The patient with diabetic ketoacidosis is

volume-depleted and may be acidotic, whereas the hypo-

glycemic patient is usually cold and clammy. In either case,

bedside diagnosis of these conditions by direct measurement

of blood glucose can be made quickly, so clinical differentiation

is less relevant. In comatose patients with diabetic ketoacidosis

in whom serum osmolality is less than 340 mOsm/Kg, another

cause of coma should be considered.

A common difficulty arises when differentiating alcoholic

ketosis from diabetic ketoacidosis in a patient with diabetes

who has ingested alcohol. In such situations, whatever the

cause of ketosis, the management is that of diabetic ketoacido-

sis. Severe abdominal pain from diabetic ketoacidosis, often

accompanied by vomiting, may mimic an acute abdomen.

Acute pancreatitis is often in the differential diagnosis, and its

diagnosis is complicated by the fact that serum amylase levels

are often elevated in diabetic ketoacidosis as a result of an

increase in serum amylase from the salivary isoenzymes.

Treatment

The principles of management of diabetic ketoacidosis are to

replace losses of water and electrolytes, give sufficient insulin

(ie, stop ketogenesis, lipolysis, and gluconeogenesis), correct

blood pH, closely monitor the patient through all stages of

management, and eliminate or treat precipitating causes.

Rapid initial evaluation takes place in the emergency room in

most patients with diabetic ketoacidosis, although at times

diabetic ketoacidosis does develop in the hospital. Table 26–7

sets forth the initial steps in management when diabetic

ketoacidosis is suspected. Figure 26-3 provides an overview

of management.

A. Monitoring—Close monitoring is the key to successful

management of diabetic ketoacidosis. Most management

decisions are fairly straightforward if the data are available

in timely fashion. Errors in management occur most often

when there is a lapse in monitoring so that a “catch-up” sit-

uation develops or the effects of overtreatment need to be

corrected.

Blood glucose should be monitored hourly at the bedside.

This provides information about whether the insulin dose is

adequate to cause a fall in blood glucose at the expected rate

of about 100 mg/dL per hour. Later, glucose measurement

prevents overshooting the target serum glucose level of

250–300 mg/dL, at which time infusion of dextrose should

be started. Serum electrolytes and ketones should be moni-

tored every 2 hours, and this should include measurement of

serum phosphorus as well. Arterial blood gases should be

repeated as necessary if progress in the clearing of acidosis is

slow or if there are associated pulmonary problems. If phosphate

therapy is used, serum calcium should be measured at least

once (after an initial measurement) during the first 12 hours

to detect any large decrease in serum calcium. All the data

obtained, including fluid balance measurements, should be

maintained on a flowchart for easy review and evaluation.

1. Correction of hyperglycemia—Correction of hyper-

glycemia occurs by four different mechanisms (Table 26–8).

First, the concentration of the extracellular glucose is diluted

by fluid replacement, expanding the extracellular space. The

second is by continuing—and usually increased—urinary

Rapid clinical examination.

Bedside blood glucose and urine ketone determinations.

Obtain blood for laboratory determination of glucose, ketones, electrolytes,

urea nitrogen, creatinine, phosphorus, magnesium.

Obtain arterial blood for blood gases.

Chest x-ray.

ECG (also to monitor K

levels).

Search for precipitating factor.

Monitoring: Hourly bedside glucose and two-hourly measurement of

anion gap, serum electrolytes, serum ketones, serum phosphorus.

Obtain arterial blood gases as needed.

Table 26–7. Initial diagnosis and management of diabetic

ketoacidosis.



CHAPTER 26

588

loss of glucose that occurs with improved renal perfusion

following expansion of the intravascular space. Third is the

effect of insulin to diminish the hepatic glucose production

rate. Considering that this is one of the important pathophys-

iologic mechanisms producing hyperglycemia in diabetic

ketoacidosis, inhibition of this process must be one of the

management goals. Indeed, it has been demonstrated that

insulin infused at the routine doses is very effective in reduc-

ing glucose production by the liver during recovery from dia-

betic ketoacidosis. Increased glucose utilization is the fourth

mechanism and is also an insulin-dependent process.

2. Correction of ketosis—Ketosis is corrected more

slowly than hyperglycemia, so while serum glucose levels

usually will reach 250–300 mg/dL within an average of 6–8

hours, it can take up to 12 hours or more for ketosis to clear.

It is particularly important to continue insulin infusion dur-

ing this period despite the fact that one major goal of ther-

apy (reduction of hyperglycemia) has been attained.

Intravenous insulin therefore should be stopped only when

ketosis has cleared.

It is not always easy to decide when a patient is no longer

in a state of ketoacidosis. Serum ketone measurements do

not provide the entire answer because they can remain posi-

tive for many hours after the acidosis is resolved, probably

because acetone is cleared much more slowly than the two

keto acids. Acetone, however, is also measured by the nitro-

prusside reaction and can give rise to persistently positive

serum ketone measurements even though it does not alter

acid-base balance. For this reason, monitoring the anion gap

is a better way of determining when ketosis has cleared.

Serum bicarbonate concentrations are useful for monitoring

progress of therapy but less useful when used alone to determine



Figure 26–3. Management of adult patients with diabetic ketoacidosis. [Na]

= corrected serum sodium concentra-

tion (for each 100 mg/dL glucose >100 mg/dL, add 1.6 meq/L to measured serum sodium.) (Modified from American

Diabetes Association. Hyperglycemic crises in diabetes. Diabetes Care 2004;27:S94–102.)

Dilution by expansion of extracellular volume

Urinary losses

Diminished glucose production rates

Increased glucose utilization

Table 26–8. Correction of hyperglycemia: Mechanisms.



DIABETES MELLITUS, HYPERGLYCEMIA, & THE CRITICALLY ILL PATIENT

589

that ketoacidosis has cleared because bicarbonate is slow to

normalize completely. This is so because hyperchloremic aci-

dosis accompanies saline therapy in almost all patients and

because hyperventilation persists after the patient is no

longer acidotic. Arterial blood pH measurement can be use-

ful when it is difficult to decide whether ketosis has cleared—

particularly if the anion gap has normalized and

hyperglycemia is resolved, yet the patient still appears to be

ill. The overall clinical picture is useful—in a patient whose

appetite has returned and is able to eat and retain a meal, this

generally signals resolution of ketoacidosis.

B. Fluid Replacement—Fluid losses in an adult with dia-

betic ketoacidosis average 5–7 L and may be as much as

10–15% of body weight. Fluid replacement should be initi-

ated immediately after diagnosis to prevent further deterio-

ration of hemodynamic status. Fluid therapy is begun with

intravenous 0.9% NaCl solution. A rough guide to volume

replacement is as follows: (1) 2 L can be given over the first 2

hours, (2) this can be followed by 2 L over the next 4 hours

using 0.9% or 0.45% NaCl solution, and (3) over the next 8

hours, another 2 L can be given using either 0.9% or 0.45%

NaCl. This schedule should be modified according to ongoing

assessment of volume replacement needs and serum sodium

levels. In hemodynamically compromised patients and the

elderly, close monitoring of volume status may be necessary,

and consideration should be given to use of a central venous

pressure or pulmonary artery catheter. If severe hypotension

is present, maintaining intravascular volume with albumin or

other plasma expanders also should be considered.

When the plasma glucose concentration reaches 250–300

mg/dL, 5% dextrose solution should be started with an

appropriate amount of NaCl according to the needs of the

patient at that time.

There is disagreement about the best approach when

serum sodium exceeds 150 meq/L at presentation. When

0.45% NaCl is given at the outset, some patients have devel-

oped hypotension during treatment. This is likely due to the

insulin effect to drive glucose intracellularly; water follows,

and the intravascular space is depleted. When hypotonic

saline is used, there is less addition of solute that will remain

in the intravascular and extracellular spaces. Therefore, a

greater proportion of the infused fluid will be ineffective in

expanding the intravascular volume.

Administering 0.9% NaCl solution provides a greater

osmotic load, and thus it is more likely to protect intravascu-

lar volume in the face of fluid shifts in diabetic ketoacidosis

patients who are strikingly volume-depleted. The concern

with infusing 0.9% NaCl (normal saline) is that hyperosmo-

lality will not be corrected. However, this approach conforms

to the physiologic principle that during states of severe vol-

ume depletion, volume maintenance takes precedence over

maintaining normal serum osmolality. Thus 0.9% NaCl is

recommended as the volume replacement fluid of first choice

even when the patient is initially hypernatremic: 0.9% NaCl

is hypo-osmolar relative to serum when serum osmolality is

greater than 308 mOsm/Kg. Once volume depletion has been

corrected, administration of 0.45% NaCl or 5% dextrose in

water to correct hypernatremia is indicated.

Fluid replacement has been given without insulin therapy

to test its effects in the absence of insulin. In a few studies

there has been considerable improvement in hyperglycemia

but with no consistent effect on serum ketones. The

improvement in serum glucose concentrations in the

absence of insulin is probably the result of increased renal

glucose losses and a dilution effect (see Table 26–8).

C. Insulin

1. Initial therapy—Insulin is started at the time of diagno-

sis at a rate of 0.1 unit/kg per hour by continuous intravenous

infusion. A bolus dose of 0.15 unit/kg may be administered

initially. Blood glucose levels should be checked at hourly

intervals using bedside (glucometer) measurements, and lab-

oratory serum glucose determinations can be obtained every

2 hours, along with serum electrolytes, to provide confirma-

tion of the glucometer measurements.

In most patients, the recommended starting dose is suffi-

cient to treat diabetic ketoacidosis because the serum-free

insulin levels achieved with this infusion rate are 7 to 10

times normal basal concentrations. However, some patients

do require even faster infusion rates. Therefore, if serum glu-

cose concentrations have not begun to fall after the first hour,

the rate of insulin infusion should be doubled—and doubled

once again if the same should happen during continuing ther-

apy. The insulin infusion should cause blood glucose to fall at

a rate of about 75 mg/dL per hour. The rate of insulin infusion

should be continued—if glucose is falling appropriately—

until serum glucose has reached 230–300 mg/dL. At that

time, a 5% dextrose infusion should be started, but the

insulin infusion must continue until ketosis clears. Because

considerable insulin resistance remains even though some of

the factors described in Table 26–1 have normalized, insulin

infusion should be maintained at the same rate and 10%

dextrose infused, if necessary, to maintain serum glucose lev-

els while awaiting clearance of ketones. Decreasing insulin

infusion rates at this time may result in slower resolution of

ketosis. Clearing of ketosis should be monitored using a

combination of the anion gap, bicarbonate, and serum

ketone measurements and, if necessary, arterial blood gas pH

measurement.

2. After resolution of ketosis—Once ketosis has

cleared—and if the patient is eating—a subcutaneous insulin

regimen can be started. The patient should be given a subcu-

taneous injection of regular insulin about 30 minutes before

stopping the insulin infusion. This allows some of the subcu-

taneously injected insulin to be absorbed before the insulin

infusion is halted. Because intravenous insulin has a half-life

of only 5–6 minutes, stopping the intravenous insulin prior

to giving subcutaneous insulin can result in temporarily

inadequate serum insulin concentrations. Low serum insulin

levels, in turn, can cause a rebound into diabetic ketoacidosis,



CHAPTER 26

590

particularly if there is significant delay in delivering the sub-

cutaneous regular insulin injection. Subcutaneous rapid-

acting (regular or analogue) insulin is also best injected

before a meal, and the insulin infusion can be halted once the

meal begins. The bolus of regular insulin thus will be deliv-

ered at a physiologically appropriate time, and the meal will

help to compensate for any error of excess in estimating the

regular insulin dosage at that time.

Split-dose insulin therapy with a combination of regular

and NPH insulin before breakfast and dinner can be started

as soon as the diabetic ketoacidosis has resolved and the

patient is able to eat. The total daily dose can be similar to pre-

hospital doses in patients with known diabetes; in newly diag-

nosed patients, regular insulin can be given before meals and

at 12 midnight using a sliding scale for 24 hours. The total

dose required then can be used to calculate split-dose therapy,

giving two-thirds of the total dose in the morning and one-

third in the evening. NPH insulin usually constitutes two-

thirds of the morning dose and three-fourths of the evening

dose, with the remainder in each case being regular insulin.

D. Potassium Replacement—Usual potassium deficits in

diabetic ketoacidosis have been estimated to be approxi-

mately 300 meq, but the deficit may be as much as 1000 meq.

This total body potassium deficit is the result of combined

renal losses owing to osmotic diuresis and potassium shifts

from intracellular to extracellular fluid that lead to further

loss through the kidneys.

If the patient is normokalemic or hypokalemic at presen-

tation, potassium replacement should be started when

insulin therapy is initiated. If the serum potassium level is

high, potassium therapy should be started only after insulin

therapy is begun and with the second liter of fluid replace-

ment. If treatment is necessary before the serum potassium

level is reported, epidemiologic data provide a basis for

determining the frequency of serum potassium abnormali-

ties (see Tables 26–5 and 26–6). Most patients initially have

either normal or high serum potassium levels; however, up to

20% of patients will have hypokalemia at the outset. In this

last group of patients, insulin infusion alone (without potas-

sium replacement) will exacerbate the hypokalemia, with

potentially severe consequences. In contrast, the fear of giv-

ing potassium to normokalemic or hyperkalemic patients is

mitigated by the effects of the concomitant insulin infusion,

intracellular rehydration, and rise in pH. Thus, if the serum

potassium concentration is unknown, the ECG does not

demonstrate hyperkalemic changes, and the patient is pro-

ducing adequate amounts of urine, potassium infusion

should be started at the same time insulin therapy is begun.

Potassium replacement is started by the addition of

potassium chloride 20–40 meq/L to the fluid replacement

solution. Potassium phosphate also may be given (alternating

with potassium chloride) if phosphate repletion is necessary.

Subsequent potassium replacement depends on the results of

serum potassium determinations, which should be done at

2-hour intervals throughout the period of management of

diabetic ketoacidosis. In addition, electrocardiographic

follow-up with single-lead measurement is a useful way of

monitoring serum potassium for gross hyperkalemia or

hypokalemia so that emergency treatment can be initiated if

necessary. It should be kept in mind that even while replace-

ment of potassium is taking place, continuing potassium

losses occur in the kidney throughout the period of manage-

ment of diabetic ketoacidosis. For this reason, serum potas-

sium measurements also should be obtained daily once the

patient is no longer ketotic. A total body deficit of potassium

may persist despite initial correction of serum potassium,

and oral potassium replacement therapy should be given to

patients who continue to be hypokalemic.

E. Bicarbonate—Acidosis generally resolves with insulin ther-

apy and metabolism of keto acids. In most cases of diabetic

ketoacidosis, it is not necessary to treat acidemia with bicar-

bonate. Recent studies have demonstrated that bicarbonate

therapy does not alter the eventual outcome of diabetic

ketoacidosis, nor does it increase the rate at which pH is cor-

rected. In fact, some have found a counterproductive effect of

sodium bicarbonate in the treatment of diabetic ketoacidosis.

Clinical and animal studies have shown that bicarbonate

administration actually may increase ketone production.

Recent data suggest also that in children with diabetic ketoaci-

dosis, the mean duration of hospitalization for those receiving

bicarbonate was 23% longer than that of children who did not

receive bicarbonate. Lastly, a multicenter retrospective study of

the risk factors associated with cerebral edema in children with

diabetic ketoacidosis found that the relative risk of cerebral

edema was 4.2 in children treated with bicarbonate compared

with a matched control group that did not receive bicarbonate.

Bicarbonate therapy also has been associated with hypocal-

cemic tetany, decreased tissue oxygen delivery, a paradoxical

fall in pH of the cerebrospinal fluid, rebound alkalosis, greater

potassium needs, and sodium overload.

However, when pH values are extremely low, concern

about the hazards of acidemia begin to outweigh the con-

cerns with alkali therapy. Therefore, at a pH of less than 7.0,

particularly in a patient who is very ill, many clinicians will

administer bicarbonate. The effects of low pH are a negative

inotropic effect on the heart, vasodilatation, cerebral depres-

sion, insulin resistance, and depression of enzyme activity. If

treatment with bicarbonate is begun, the American Diabetes

Association (ADA) recommends that 100 mmol of sodium

bicarbonate should be added to 400 mL of sterile water and

given at a rate of 200 mL/h in severe acidosis (pH <6.9). In

patients with a pH of 6.9–7.0, 50 mmol of sodium bicarbon-

ate should be diluted in 200 mL of sterile water and infused

at a rate of 200 mL/h.

In general, taking into account the hazards of treatment,

the hazards of acidemia, and the probable benefits of bicar-

bonate administration, it is not necessary to treat routinely

with bicarbonate.

F. Phosphate Replacement—The routine administration of

phosphate in the management of diabetic ketoacidosis remains



DIABETES MELLITUS, HYPERGLYCEMIA, & THE CRITICALLY ILL PATIENT

591

controversial. Although there are theoretical reasons for replac-

ing phosphate during treatment, most studies have not demon-

strated any beneficial effect. Despite depletion of erythrocyte

2,3-diphosphoglycerate and the concern about decreased oxy-

gen delivery to the tissues resulting from a leftward shift in the

oxyhemoglobin dissociation curve, there are no deleterious

clinical consequences of a mildly low serum phosphorus con-

centration. For most patients with diabetic ketoacidosis, routine

administration of phosphate is unnecessary.

However, it is important to separate out one group of

patients in whom phosphate therapy is essential—those with

low serum phosphorus levels at presentation, which includes

about 10% of patients (see Table 26–6). Serum phosphorus

decreases in all patients, sometimes dramatically, after insti-

tution of insulin therapy. If the initial serum phosphorus

concentration is 1.5 mg/dL, however, it may fall with treat-

ment into the range of concentrations that are associated

with the hypophosphatemia syndrome (Table 26–9). This

syndrome may include decreased myocardial contractility,

respiratory muscle weakness and respiratory failure, hemol-

ysis, and rhabdomyolysis. It is important to note that none of

the studies evaluating the efficacy of phosphate therapy in

diabetic ketoacidosis included patients with severely

depressed serum levels at presentation, so the conventional

wisdom that phosphate therapy is unnecessary should not be

extended to include the group with low serum phosphate

levels at diagnosis.

Severe hypophosphatemia may not be recognized unless

frequent serum phosphorus measurements are made. These

should be obtained every 2 hours during treatment of dia-

betic ketoacidosis in all patients whose serum phosphorus

level is less than 2.5 mg/dL at presentation. Patients with

hypophosphatemia should be given supplemental phospho-

rus in the form of potassium phosphate.

Potassium phosphate can be added to the fluid replacement

solution, alternating with potassium chloride. Hypopho-

sphatemic patients generally will require 500–1000 mg

elemental phosphorus given over 12–24 hours depending on

the severity of phosphorus depletion. This is equivalent to

about 15–30 mmol of phosphate, or a total of 5–10 mL of

potassium phosphate solution (3 mmol/mL) to be added to

the fluid replacement solutions (1–4 mL potassium phos-

phate solution per liter of replacement fluid). A concern

about phosphate administration in patients with diabetic

ketoacidosis is hypocalcemic tetany, which has been

described in some patients who receive phosphate therapy. It

should be kept in mind that calcium levels may fall during

the management of diabetic ketoacidosis. Small decreases in

serum calcium therefore may not be a contraindication to

continued phosphate repletion in patients who are

phosphorus-depleted, but they do require more frequent

monitoring of serum calcium to prevent hypocalcemia.

Patients with normal or high serum phosphorus levels

at diagnosis will have a decrease in serum phosphorus

during therapy, but only to levels that do not usually

require treatment.

Complications of Diabetic Ketoacidosis

A number of complications of diabetic ketoacidosis may

occur during treatment. Hypocalcemia, hypokalemia or

hyperkalemia, hypophosphatemia, and hypoglycemia need

to be watched for during continuous monitoring of patients

with diabetic ketoacidosis (Table 26–10). Another recognized

complication is thromboembolism that occurs as a result of

Table 26–9. Hazards of hypophosphatemia and

phosphorus repletion in diabetic ketoacidosis.

Hypophosphatemia

Decreased red cell 2,3-diphosphoglycerate with decreased

delivery

Muscle weakness

Hypophosphatemia syndrome ([P] <1.0 mg/dL) with respiratory

insufficiency, decreased myocardial contractility, rhabdomyolysis,

hemolysis

Phosphorus administration

Hypocalcemia and tetany

Complication Possible Mechanism

Hypokalemia Bicarbonate therapy, inadequate replacement,

insulin

Hyperkalemia Anuria, excessive replacement

Hypophosphatemia Insulin therapy (if [P] starts out

<1.5 mg/dL)

Hypoglycemia Inadequate glucose infusion with insulin

therapy

Other

Thromboembolism

Aspiration

ARDS

Cerebral edema

Mucormycosis

Increased platelet adhesiveness, hypervis-

cosity, poor perfusion

Gastric stasis, vomiting; lack of nasogastric

suction

Use of hypotonic replacement fluids or

excess crystalloid infusion or increased

pulmonary epithelial permeability

Excessive correction of hyperglycemia

Use of hypotonic replacement fluids(?)

Acidemia

Table 26–10. Complications of diabetic ketoacidosis.



CHAPTER 26

592

increased platelet adhesiveness and hyperviscosity.

Aspiration of gastric contents is associated with gastroparesis

and vomiting. Nasogastric suction should be initiated in all

patients who have altered consciousness to prevent this com-

plication. Cerebral edema and acute respiratory distress syn-

drome are other complications that will be discussed below.

One of the rare but significant conditions that should be

kept in mind in the management of diabetic ketoacidosis is

mucormycosis. This rare condition associated with ketoaci-

dosis is a treatable yet severe infection by the fungus

Rhizopus. Early diagnosis is essential so that therapy can be

instituted to prevent the severe morbidity associated with

this rapidly invasive infectious process. The hallmark of

mucormycosis is the finding of black necrotic debris in the

area of the eye, nose, or nasal cavity and histologic evidence

of vascular thrombosis or tissue infarction on biopsy. These

findings result from the propensity for invasion by

mucormycosis into the vascular system. This diagnosis is

considered to be a medical emergency that requires urgent

surgical and antifungal therapy.

Current Controversies and Unresolved Issues

A. Fluid Replacement—Fluid replacement remains a con-

troversial issue from two different points of view. The first is

volume replacement. A study comparing the effects of pro-

viding 3 L of volume replacement over the first 8-hour

period of treatment versus giving 6 L over the same period of

time showed very little difference in outcomes in the two

groups of patients. The conclusion is that smaller volumes of

fluid may be administered safely to some patients with dia-

betic ketoacidosis. It needs to be emphasized that patients

with severe volume deficits were excluded from this study,

but patients with moderate fluid deficits may do well with

slower replacement of their losses. One approach provides a

more flexible regimen for fluid replacement that recognizes

the unique requirements of each patient but consequently

requires careful monitoring. Fluid is replaced at a rate of

15–20 mL/kg per hour for 1 hour and then at 4–14 mL/kg

per hour thereafter according to need, as determined by care-

ful monitoring of fluid balance.

The type of fluid replacement also has been an issue of

controversy. Some investigators feel that colloid therapy is

better than crystalloid infusion in the management of dia-

betic ketoacidosis. Their concern is that insufficient fluid is

maintained in the intravascular space during therapy and

that much of the fluid, when given as crystalloid, finds its way

into the interstitial space, where edema is the result. It has

been suggested that the syndromes of cerebral edema and

pulmonary edema may be a consequence in part of the

sequestration of fluid outside the intravascular space.

However, no controlled studies of colloid infusion have been

performed. Colloid solutions should be considered in the

management of hypotension in patients who present with

diabetic ketoacidosis, but at this time, crystalloid infusion

remains the management of choice.

B. Cerebral Edema—This serious complication is fortu-

nately quite rare. Cerebral edema manifests as a deteriorating

level of consciousness at a time when other parameters of dia-

betic ketoacidosis are improving in response to treatment.

Clinical evidence for this complication generally appears

early in the course of treatment; almost two-thirds of

patients who develop neurologic deterioration begin to do

so within 12 hours of initiation of therapy. Cerebral edema

occurs mainly in children and young adults and is associ-

ated with a greater than 90% mortality. However, despite its

rarity, this complication has influenced one of the basic

aspects in our management of diabetic ketoacidosis because

of its catastrophic consequences. As mentioned earlier, it is

accepted practice to start glucose infusion to prevent serum

glucose concentrations from falling below 250–300 mg/dL.

This recommendation derives from studies of experimentally

induced cerebral edema in diabetic dogs that demonstrated

an association of cerebral edema with greater decrements in

plasma glucose concentrations during treatment.

In an observational study of children who developed cere-

bral edema with ketoacidosis, only treatment with bicarbon-

ate was associated with cerebral edema. Neither the initial

serum glucose concentration nor the rate of change in the

serum glucose concentration during therapy was associated

with the development of cerebral edema. Symptomatic cere-

bral edema has developed in a few children with diabetic

ketoacidosis before the initiation of therapy, suggesting that it

may not necessarily be caused by therapy. Children with dia-

betic ketoacidosis who have low partial pressures of arterial

carbon dioxide (relative risk 2.7 per decrease of 7.8 mm Hg)

and high serum urea nitrogen concentrations (relative risk 1.8

per increase of 9 mg/dL) at presentation were at increased risk

for cerebral edema. This supports the hypothesis that cerebral

edema in children with diabetic ketoacidosis is related to

decreased perfusion of the brain. Both hypocapnia, which

causes cerebral vasoconstriction, and extreme dehydration

would be expected to decrease brain perfusion. Furthermore,

because the developing brains of children are more sensitive

to hypoxia, this may explain the greater incidence of cerebral

edema in children than in adults.

Subclinical cerebral edema appears to be quite common

in patients with diabetic ketoacidosis because compression of

the subarachnoid and ventricular spaces during treatment

has been observed by CT scan without any obvious clinical

manifestations. The high frequency of subclinical edema has

raised questions about the rate and composition of fluid

replacement discussed earlier in this section. Mannitol is

used in therapy of symptomatic cerebral edema.

C. Acute Respiratory Distress Syndrome (ARDS)—

Another unusual but serious complication of ketoacidosis is

ARDS. Patients present with progressive dyspnea and hypox-

emia; the earliest sign appears to be an increased P(

–a)

and pulmonary edema is found on chest x-ray. ARDS in dia-

betic ketoacidosis tends to occur most often in patients

under 50 years of age, and the mortality rate associated with



DIABETES MELLITUS, HYPERGLYCEMIA, & THE CRITICALLY ILL PATIENT

593

this syndrome is extremely high. The pathogenesis of ARDS

in diabetic ketoacidosis is unknown, but it also has been sug-

gested that the syndrome may be associated with crystalloid

infusion and perhaps with excess volume replacement. It is

unlikely that either of these factors is the sole cause of ARDS

because it is too unusual a complication of diabetic ketoaci-

dosis to be simply fluid-dependent given the wide range of

replacement regimens used. However, it is possible that there

is an association with excessive fluid replacement in patients

who have some underlying pulmonary disease process asso-

ciated with increased pulmonary epithelial permeability.

Glaser NS et al: Mechanism of cerebral edema in children with dia-

betic ketoacidosis. J Pediatr 2004;145:164–71. [PMID: 15289761]

Kamalakannan D et al: Diabetic ketoacidosis in pregnancy.

Postgrad Med J 2003;79:454–7. [PMID: 12954957]

Kitabchi AE et al: Hyperglycemic crises in adult patients with

diabetes: A consensus statement from the American Diabetes

Association. Diabetes Care 2006;29:2739–48. [PMID:

17130218]

Linfoot P, Bergstrom C, Ipp E: Pathophysiology of ketoacidosis in

type 2 diabetes mellitus. Diabet Med 2005;22:1414–9. [PMID:

16176205]



Hyperglycemic Hyperosmolar Nonketotic

Coma

This disorder is probably not a distinct entity but rather a

variant of the diabetic ketoacidosis syndrome because a large

overlap exists in the clinical presentation and pathogenesis of

diabetic ketoacidosis and nonketotic coma. Management of

the two conditions is essentially identical. However, there are

clearly major differences in clinical presentation between

typical patients with nonketotic coma and diabetic ketoaci-

dosis, who probably represent opposite extremes of this

metabolic syndrome.

The patient with nonketotic coma typically is an elderly

person who presents with diabetes for the first time, is

severely dehydrated, is more often in coma, and has severe

associated diseases and a poor outcome. In contrast, the

common first presentation of diabetic ketoacidosis is in a

teenager who is otherwise healthy, not often in coma, and has

an excellent chance of recovery. Biochemically, the typical

patient with hyperglycemic hyperosmolar nonketotic coma

has more severe hyperglycemia (eg, serum glucose often >

1000 mg/dL) and hyperosmolarity, more pronounced elec-

trolyte abnormalities, and no ketosis. Between these two

extremes are a large number of patients in whom the clinical

syndromes overlap, so some degree of ketosis is often present

in a patient who otherwise would be classified as having typ-

ical nonketotic coma; on the other hand, hyperosmolarity

and severe hyperglycemia occur in what otherwise appears to

be typical diabetic ketoacidosis.

The pathogenesis of hyperglycemic hyperosmolar nonke-

totic coma is not well understood. The severity of the hyper-

osmolarity can be ascribed to a greater degree of volume

depletion. Indeed, severe volume depletion is the most

important common feature and can best be explained by the

degree of hyperglycemia. Figure 26–1 provides an explana-

tion for this severe hyperglycemia. Since volume depletion is

a characteristic feature, renal blood flow and the glomerular

filtration rate are reduced, thus diminishing the urinary

escape of glucose. However, the hormonal setting ensures a

continuing high rate of hepatic glucose production (see

above for further discussion of Figure 26–1), and this has the

capacity to rapidly increase blood glucose levels. If this

sequence of events is not halted, severe hyperglycemia and

hyperosmolarity ensue.

The other major feature of nonketotic coma—the rela-

tively low level of serum keto acids—also may play an impor-

tant role in the development of severe hyperglycemia.

Ketoacidosis makes the patients feel more ill and thus pre-

vents the syndrome from going undiagnosed for prolonged

periods. The lack of ketosis in hyperglycemic hyperosmolar

nonketotic coma may delay presentation, resulting in ongo-

ing osmotic diuresis that results in more severe volume

depletion. The mechanism for the lower levels of ketosis is

unexplained. The hormonal picture at presentation is not

helpful because concentrations of circulating counterregula-

tory hormones do not appear to be different from those seen

in diabetic ketoacidosis, nor do insulin levels seem to differ.

Nevertheless, it is thought that a restraining effect on lipoly-

sis by small increases of circulating insulin—and, as a result,

less availability of substrate for ketogenesis—may explain

low ketone concentrations in this syndrome. It also appears

that hyperosmolarity may itself play a role in diminishing

ketosis; this is based on in vitro studies in which hyperosmo-

larity was shown to suppress lipolysis.

Management is not different from that of diabetic

ketoacidosis. Plasma glucose concentrations should be

allowed to decrease at a rate of about 100 mg/dL per hour,

and fluid replacement should begin with normal saline as

discussed earlier. Early aggressive volume replacement is

important to prevent hypotension that may result from

fluid shifts that accompany insulin-mediated glucose

transport into the intracellular compartment. Many of

these patients are elderly, and monitoring fluid balance

with central venous catheters therefore is advised. Because

of the massive osmotic diuresis that precedes presentation

in these patients, electrolyte losses may be severe. It is par-

ticularly important to monitor serum potassium concen-

tration carefully during early management and in

subsequent days so as to provide potassium supplements if

they become necessary. The mortality rate is high, with

deaths often due to concurrent illnesses or thrombotic and

infectious complications.

Kitabchi AE et al: Hyperglycemic crises in adult patients with

diabetes: A consensus statement from the American Diabetes

Association. Diabetes Care 2006;29:2739–48. [PMID:

17130218]



CHAPTER 26

594

MANAGEMENT OF THE ACUTELY ILL PATIENT

WITH HYPERGLYCEMIA OR DIABETES MELLITUS

Management of patients in the ICU setting has undergone a

strategic change in a matter of a few years since the publica-

tion of a landmark work known as the Leuven study in 2001.

This study demonstrated unequivocally that management of

hyperglycemia is of fundamental importance in critically ill

patients and has changed the way in which hyperglycemia

and diabetes are managed in the ICU.

For many years it was thought that moderate control of

blood glucose in inpatients and the critically ill was appro-

priate care. This was a result of the prevailing interpretation

of the hyperglycemia of injury as a benign adaptive process

that contributes to survival. Mild or moderate hyperglycemia

is common in critically ill patients, even in those not previ-

ously known to have diabetes. Hyperglycemia is due to a

combination of physiologic changes including insulin resist-

ance and increased hepatic glucose production despite

increased secretion of insulin and was seen as necessary to

provide glucose as fuel during severe injury. The Leuven

study challenged these concepts after demonstrating that

tight glucose control (<110 mg/dL) improved mortality and

morbidity in critically ill patients and thereby changed not

only the practice of medicine in the ICU but also how we

think about glucose regulation in acute illness.



Hyperglycemia

Pathophysiology

The most frequent pathophysiologic abnormality of glucose

metabolism induced by acute illness is an increase in insulin

resistance. The molecular mechanisms are not well under-

stood, but a number of associated abnormalities contribute to

the development of insulin resistance. Increased secretion of

glucose counterregulatory hormones occurs in the acutely ill

patient. Increased concentrations of cortisol, catecholamines,

growth hormone, and glucagon lead to increased hepatic glu-

cose production by gluconeogenesis and glycogenolysis, as well

as insulin resistance at the level of muscle and adipose tissue.

Although insulin secretion increases in response to rising glu-

cose concentrations, it is insufficient to prevent hyperglycemia,

which is usually mild in nondiabetic patients. Catecholamines

may play a role in restraining an adequate insulin secretory

response by their action to suppress beta cell function.

The ability to continue to secrete increasing (although

inadequate) amounts of insulin in response to elevated serum

glucose is probably what prevents the development of ketoaci-

dosis in most patients with type 2 diabetes when acute illness

supervenes, although insufficient to prevent hyperglycemia.

The presence of even low levels of insulin may be sufficient to

inhibit lipolysis in adipose tissue and thereby suppress ketoge-

nesis in the liver. Indeed, it is unusual for patients with type 2

diabetes to develop diabetic ketoacidosis even though they

have the same pattern of glucose counter-regulatory hormone

release. If diabetic ketoacidosis occurs in type 2 diabetes, it is

due to more severe insulin deficiency than in equally ill type

2 patients without ketosis.

Clinical Benefits of Intensive Insulin Therapy

in the Critically Ill

A brief outline of the Leuven study will be provided to illustrate

the benefits of the use of insulin-mediated tight glycemic con-

trol in the ICU. Critically ill patients, almost all post–cardiac

surgery and most not previously known to have diabetes, were

studied with the intention of testing the role of maintaining

strict normoglycemia (<110 mg/dL) in the ICU. Those patients

randomized to the control group also were treated with insulin

if the blood glucose level exceeded 200 mg/dL, so average blood

glucose concentrations were 150–160 mg/dL, previously

thought to be adequate glycemic control in this setting.

Mortality was reduced dramatically in the intensive insulin ther-

apy group, especially among patients with prolonged critical ill-

ness in the ICU. In those requiring intensive care for more than

5 days, mortality was reduced from 20% to about 10%. Time in

the ICU was decreased significantly. Morbidity also was reduced

significantly, including prevention of severe complications such

as acute renal failure, nosocomial infections, liver dysfunction,

critical illness neuropathy, muscle weakness, and anemia. Other

recent reports confirm the findings in this landmark study,

including medical ICU patients, although less strikingly in this

population. Carefully monitored intravenous insulin titration

regimens are recommended to induce normoglycemia without

the risk of significant hypoglycemia.

This approach is also applied to patients with diabetes,

although in known diabetic patients hyperglycemia may be

more severe to begin with. Patients who have inadequate

insulin secretory reserve but who are not frankly diabetic are

at great risk for severe metabolic decompensation owing to

intercurrent disease. These patients are unable to respond to

increasing insulin needs, but unlike in previously diagnosed

diabetes, they may not necessarily recognize the symptom

complex associated with hyperglycemia. This can be exacer-

bated if large quantities of glucose are infused or ingested.

For example, in patients who ingest a large amount of

sucrose from sugar-containing soft drinks because of

extreme thirst induced by diuresis or diarrhea, ensuing

osmotic diuresis can make this a self-perpetuating disorder.

Similarly, peritoneal dialysis using solutions with high glu-

cose concentrations, used in an attempt to remove excess

extracellular fluid, may be associated with hyperglycemia and

even hyperosmolar diabetic coma.

An example of how control of diabetes influences the

management of another disorder is when diabetes insipidus

and diabetes mellitus coexist. In diabetes insipidus, large

volumes of free water are lost in the urine because of an

inability to concentrate urine. During treatment, equally

large quantities of electrolyte-poor solutions—usually 5%

dextrose—are infused to match urine output. If the urine

output is large, 6–10 L of 5% dextrose may be given within a



DIABETES MELLITUS, HYPERGLYCEMIA, & THE CRITICALLY ILL PATIENT

595

24-hour period, representing a load of 300–500 g of glucose

into the extracellular pool (compared with about 20 g in the

extracellular pool in a patient with normal serum glucose

concentration; see Figure 26–1). Nondiabetic subjects handle

this glucose load by increasing the secretion of insulin.

However, in patients who have underlying or overt diabetes

and a poor insulin secretory response, serum glucose

increases, and glycosuria and an osmotic diuresis develop

unresponsive to the effect of vasopressin. Continuing urine

volume losses result in the administration of even larger vol-

umes of dextrose, perpetuating a vicious circle. Proper man-

agement consists of preventing glycosuria by tightly

controlling blood glucose concentration. This can be achieved

using an intravenous insulin infusion in the ICU while simul-

taneously treating the diabetes insipidus with vasopressin.



Hypoglycemia

Some diabetic patients appear to have decreased insulin

requirements or a decreasing need for oral hypoglycemic

agents when acute illness supervenes. A decrease in insulin

requirements may result from a number of different

mechanisms.

Pathophysiology

A. Decreased Insulin Resistance—Weight loss may occur

prior to or during the acute illness. It is well known that

weight loss reduces insulin resistance in diabetes mellitus,

and small decreases in weight often are sufficient to produce

a substantially decreased requirement for insulin or other

hypoglycemic agents. Another cause of decreased insulin

resistance is a deficiency of one of the glucose counterregu-

latory hormones. Adrenal or pituitary insufficiency may

occur as part of an intercurrent disease process. Decreased

growth hormone or cortisol can diminish the need for hypo-

glycemic agents by enhancing peripheral insulin sensitivity

and decreasing gluconeogenesis.

B. Decreased Clearance of Insulin or Oral Hypoglycemic

Agents—Because the liver and the kidneys are the primary

organs involved in metabolism of insulin and the sulfony-

lurea drugs, development of renal or hepatic failure may

delay drug clearance and result in hypoglycemia. In the case

of the sulfonylureas, this may be associated with severe, pro-

longed hypoglycemia and coma. Concomitant use of other

drugs also may influence the metabolism of sulfonylurea

agents (Table 26–11). A relatively common cause of unex-

pected hypoglycemia occurs during acute episodes of con-

gestive heart failure in insulin-treated patients. As a result of

liver congestion and possibly decreased insulin clearance

with reduced hepatic blood flow, insulin requirements are

transiently diminished. If the insulin dose is not reduced,

hypoglycemia may occur.

C. Drug Interactions—Drug interactions are of particular

importance with the first generation of oral hypoglycemic

agents that are tightly protein-bound and may be displaced

by other protein-bound drugs. Table 26–11 summarizes pos-

sible interactions between drugs used in management of dia-

betes and therapeutic agents that may affect glucose control.

Because multiple drugs are used often in acutely ill patients,

particularly the elderly, the possibility of drug interactions

giving rise to hypoglycemia always needs to be considered.

Clinical Features

Hypoglycemia in hospitalized patients occurs most com-

monly in patients with diabetes, and this is usually due to a

decrease in caloric intake related to illness or hospital routine

with continued administration of hypoglycemic drugs. Some

of the causes of hypoglycemia observed in hospitalized

patients with diabetes are presented in Table 26–12.

Concomitant diseases play an important role in increasing

the risk of hypoglycemia in patients with diabetes, for exam-

ple, renal insufficiency, malnutrition, and sepsis—three dis-

orders known to be associated with hypoglycemia even in

nondiabetic hospitalized patients—as well as alcohol inges-

tion, liver disease, shock, pregnancy, and malignancy. A high

mortality rate has been reported in nondiabetic critically ill

patients who develop hypoglycemia, and in some, this may

represent a terminal phenomenon associated with malnutri-

tion or kidney and liver disease.

Pentamidine, used for treatment for Pneumocystis carinii

infection, is associated with hypoglycemia probably owing to

destruction of pancreatic beta cells with acute insulin release.

Treatment

Several factors determine appropriate management for

patients with diabetes mellitus and acute illness to avoid the

consequences of hyperglycemia and hypoglycemia.

Interaction Drug

Displacement of sulfonylureas

from plasma proteins

Clofibrate, phenylbutazone,

salicylates, sulfonamides

Reduced hepatic sulfonylurea

metabolism

Dicumarol, chloramphenicol,

monoamine oxidase inhibitors,

phenylbutazone

Decreased urinary excretion of

sulfonylureas or metabolites

Allopurinol, probenecid, phenylbu-

tazone, salicylates, sulfonamides

Intrinsic hypoglycemic activity Insulin, alcohol, beta-adrenergic

agonists, salicylates, guanethidine,

monoamine oxidase inhibitors

Table 26–11. Pharmacokinetic and pharmacodynamic

interactions that augment the hypoglycemic actions of

sulfonylurea drugs.



CHAPTER 26

596

A. Goal for Serum Glucose—Based on the data of the

Leuven study, target glucose concentrations in acutely ill

patients have been changed to attempt to achieve normo-

glycemia, defined as concentrations of less than 110 mg/dL.

In medical patients, a target of 130–140 mg/dL may be more

appropriate. This is best achieved using continuous intra-

venous insulin infusion with a titration algorithm. Once

patients are out of the critical care setting, other regimens

may be appropriate.

B. Fasting Patients—Patients who for any reason are not

being fed do not need bolus injections of rapid-acting insulin

to maintain glucose homeostasis postprandially; a continu-

ous intravenous insulin infusion is more appropriate.

Intravenous insulin is given routinely if a patient is receiving

total parenteral nutrition because of the large load of calories

given to patients during this therapy. However, even in the

absence of an exogenous source of glucose, such as total par-

enteral nutrition, it is important to recognize that hyper-

glycemia in patients with diabetes is the result of a

continuous delivery of large amounts of glucose from an

endogenous source, that is, the liver. It is logical, therefore,

that serum glucose concentration be controlled by continu-

ous levels of insulin, and this can be done with intravenous

insulin infusion. The response should be evaluated by moni-

toring blood glucose level at the bedside at 2–3-hour intervals.

An alternative to the use of intravenous infusion of insulin for

patients who are no longer critical is the use of long-acting sub-

cutaneous insulin (eg, glargine insulin or insulin-detemir).

These insulins analogues provide the equivalent of a basal

infusion rate, but because they do not provide the possibility

of hour-to-hour titration, they are unsuitable for anyone but

very stable patients, in whom day-to-day titration of insulin

is often sufficient.

C. Patients with Intermittent Food Intake—Patients who

are acutely ill, are vomiting, or have decreased food intake

can be managed somewhat differently. Under these circum-

stances, longer-acting insulins may be inappropriate because

it is difficult to predict insulin needs for more than a few

hours in advance. Thus, even if NPH insulin were adminis-

tered in an appropriate dose, inability to eat a meal hours

later could put the patient at risk for hypoglycemia. Rapid-

acting insulin given before each meal and at midnight is

preferable.

Rapid-acting insulin can be given to patients four times a

day using a variable insulin dosage (2–10 units) based on the

bedside blood glucose determination (sliding scale). Glucose

is measured preprandially, immediately before each insulin

dose is to be administered, and at midnight. This provides

insulin coverage for the expected glucose excursion with each

meal, and the dose can be reduced if the patient is not going

to eat the next meal. As soon as the patient begins to stabilize

and is able to eat consistently, an attempt should be made to

convert the patient to morning and evening split-dose

insulin injections using a combination of NPH and regular

insulin or multiple-dose insulin using a combination of pre-

meal rapid-acting analogues and a basal insulin (eg, glargine

or detemir).

D. Patients Eating Regularly—Patients who are eating

meals or receiving feeding in bolus fashion in reasonably

consistent manner can be treated with a split-dose regimen

of NPH and rapid-acting insulin. Alternatively, basal insulin

(eg, glargine or detimir) with premeal rapid-acting ana-

logues can be used. If necessary, a sliding scale of supplemen-

tary preprandial rapid-acting insulin doses can be added to

provide good glycemic control. There is no need for pro-

longed use of sliding scales in this situation because with the

sliding-scale approach the doses of insulin are variable from

day to day, and more stable patients will do better if managed

with a more constant regimen.

Once patients have stabilized, in addition to preprandial

glucose measurements, which are the basis for the sliding

scale, blood glucose measurements also should be performed

2 hours after meals to evaluate whether the dose of rapid-

acting insulin given prior to those meals is appropriate for

adequate glucose control. These 2-hour postprandial glucose

measurements are most useful for determining the required

dose of rapid-acting insulin. Rapid-acting insulin dosages

can be increased or decreased depending on the change in

glucose concentration from that obtained immediately prior

to the meal (keeping in mind that there is a normal rise in

blood glucose levels of about 40–50 mg/dL after a meal).

Table 26–12. Proximate causes of hypoglycemia in 42

diabetic patients receiving insulin (60 episodes) or oral

hypoglycemia drugs (four episodes).

Cause % Episodes

Decreased intake of calories

Nausea, vomiting, anorexia, lethargy

NPO because of diagnostic test or surgical

procedures

Enteral tube feedings held for measurement

of residual

Meals not delivered

Adjustment of insulin dosage

Treatment of diabetic ketoacidosis or nonketotic

coma

Attempt at tighter metabolic control

Sliding scale insulin doses too high for patient with

renal insufficiency

Failure to reduce insulin dose after one hypoglycemia

episode

Less insulin needed as infection resolved

Incorrect dose of insulin given

No cause identified