Management of the Hyperosmolar Hyperglycemic Syndrome
Am Fam Physician. 1999 Oct 1;60(5):1468-1476.
This article exemplifies the AAFP 1999 Annual Clinical Focus on management and prevention of the complications of diabetes.
Hyperglycemic hyperosmolarity is part of a clinical spectrum of severe hyperglycemic disorders ranging from pure hyperglycemic hyperosmolarity without ketosis to diabetic ketoacidosis, with significant overlap in the middle. From 50 to 75 percent of hospitalizable patients who have uncontrolled diabetes present with significant hyperosmolarity. An altered state of consciousness attributable to uncontrolled diabetes is virtually always the result of severe hyperosmolar hyperglycemia. The linchpin of therapy is prompt, rapid administration of crystalloid solutions that have tonicity appropriate to the level of hyperosmolarity. A decrease in the plasma glucose concentration indicates the adequacy of therapy, especially rehydration; the goal is for the plasma glucose level to decline by at least 75 to 100 mg per dL (4.2 to 5.6 mmol per L) per hour. Patients with hyperosmolar hyperglycemic syndrome are often chronically ill, and they may have major total body deficits of potassium, phosphate and magnesium, as well as B-complex vitamins (especially thiamine). These deficits also require attention and correction during therapy.
Uncontrolled hyperglycemia remains a frequent complication in patients with diabetes mellitus. From 1989 through 1991 in the United States, the average annual number of hospital discharge diagnoses in patients with diabetes exceeded 100,000 for diabetic ketoacidosis, 4,500 for diabetic coma, 10,800 for diabetic hyperosmolar nonketotic coma and 18,800 for acidosis.1 The incidence of diabetic ketoacidosis ranges from 4.6 to 8.0 cases per 1,000 patients with diabetes per year.1 Reported mortality rates range from 9 to 14 percent for diabetic ketoacidosis and 10 to 50 percent for hyperosmolar nonketotic coma.1 Hyperglycemic hyperosmolarity occurs in 18 to 33 percent of patients with classic diabetic ketoacidosis.2,3
The pure hyperosmolar hyperglycemic syndrome, or hyperosmolar nonacidotic diabetes mellitus, is characterized by severe hyperglycemia, hyperosmolarity and dehydration in the absence of significant ketoacidosis. However, the presence of some ketonuria or mild ketonemia does not preclude the diagnosis (Table 1).4
Laboratory Parameters in Patients with Hyperosmolar Hyperglycemic Syndrome
The rightsholder did not grant rights to reproduce this item in electronic media. For the missing item, see the original print version of this publication.
The syndrome occurs more frequently in elderly patients with type 2 diabetes (formerly known as non–insulin-dependent diabetes). It develops more insidiously than diabetic ketoacidosis and is frequently associated with central nervous system signs and symptoms (the most profound of which is coma), as well as severe fluid depletion and impairment of renal function.
The clinical spectrum of severe hyperglycemic disorders ranges from hyperglycemic hyperosmolarity without ketosis to full-blown diabetic ketoacidosis, with considerable overlap in the middle (i.e., approximately one fifth to one third of patients with otherwise classic diabetic ketoacidosis also have hyperglycemic hyperosmolarity). Between one half and three fourths of hospitalizable patients with uncontrolled diabetes present with an effective osmolarity of 320 mOsm per L or greater.2
The initiating event in uncontrolled diabetes is the development of a glucosuric diuresis.4 The presence of glucose in the urine impairs the concentrating capacity of the kidney and thus exacerbates water losses.
If fluid intake is adequate and the glomerular filtration rate is maintained, renal loss of glucose functions as a “safety valve” by preventing the accumulation of nondiffusible osmotically active glucose in the extracellular fluid, along with the hyperosmolarity associated with this accumulation.5 A normally perfused kidney does not allow major hyperglycemia to persist, even over short periods. However, if adequate renal function is not maintained because of primary kidney disease or secondary to intravascular volume depletion with a fall in the glomerular filtration rate, plasma glucose levels increase markedly and hyperosmolarity develops.
In patients who fail to respond to the stimulus of thirst because of incapacity or confusion, hyperosmolar hypergylcemia ensues. Often, these patients are unable to ingest or retain fluids because of restraints, sedation, coma, nausea, vomiting or diarrhea, or they are unable to make their needs known or recognized (e.g., certain patients in nursing homes or hospitals). They may be receiving inadequate free-water feedings, or they may have impaired renal function, such as an inability to concentrate urine or to respond adequately to anti-diuretic hormone by conserving water. Thus, they fail to ingest or retain sufficient free water to meet the demands of the glucosuric osmotic diuresis and lose water in excess of electrolytes.6 If these losses are not replaced, hypovolemia, intracellular and extracellular dehydration, and hyperosmolarity develop.
Hyperglycemia from gluconeogenesis causes a solute diuresis resulting in the manifestations of uncontrolled diabetes mellitus: polyuria, polydipsia and volume loss leading to hypovolemia, hypotension, organ hypoperfusion and tachycardia.
Although hyperosmolar hyperglycemic syndrome typically occurs in the elderly, it may occur at any age. Often, the patient is not known to have diabetes, or the disease has been managed by diet, oral hypoglycemic agents or small amounts of insulin. Patients with the syndrome may present with a depressed mental status. In one series including 275 uncontrolled diabetic patients,2 45 percent of patients with an effective osmolarity of greater than 350 mOsm per L were comatose on presentation.
Patients with the syndrome have a history of days to weeks of thirst, polyuria and, frequently in the background, a condition such as stroke or renal insufficiency. Weight loss, weakness, visual disturbances and leg cramps are common symptoms. The physical examination demonstrates profound dehydration, poor tissue turgor, soft, sunken eyeballs, cool extremities and, at times, a rapid thready pulse. On presentation, these patients have glucosuria and minimal or no ketonuria or ketonemia. Mild metabolic acidosis with an increased anion gap is present in up to one half of patients with hyperosmolar hyperglycemic syndrome.
Nausea, vomiting and abdominal pain occur less frequently in pataients with hyperosmolar hyperglycemia than in those with diabetic ketoacidosis. Occasionally, patients with the syndrome have constipation and anorexia. Gastric stasis and ileus occur less often than in patients with classic diabetic ketoacidosis.
Abdominal pain or tenderness, nausea and vomiting, lack of bowel sounds and ileus in patients with uncontrolled diabetes may obscure intra-abdominal pathologic processes that require urgent attention. Therefore, historical information and response to therapy are of critical importance. The development of findings secondary to uncontrolled diabetes follows the onset of symptoms rather than precedes it, and the symptoms usually improve markedly following the infusion of fluids and insulin.
Fatty infiltration of the liver, associated with abnormal liver function tests, may be another cause of abdominal pain and tenderness in patients with uncontrolled diabetes. Liver function tests are abnormal in up to one third of patients with uncontrolled diabetes.4
Approximately 20 to 25 percent of patients who have hyperosmolar hyperglycemic syndrome present in a coma.2,7 However, many patients have no clouding of consciousness.2,7 When present, coma is primarily the result of hyperglycemia and hyperosmolarity, not acidosis.2,3,7,8 Coma rarely occurs unless the measured osmolarity exceeds 340 mOsm per L or the effective osmolarity is greater than 320 mOsm per L. In one series,2 most of the comatose patients had an effective osmolarity of greater than 350 mOsm per L. The absence of hyperosmolarity (and hypoglycemia) in an obtunded patient with diabetes suggests a cause other than diabetic decompensation.4,8
Patients with diabetes and hyperosmolarity may present with neurologic abnormalities that rarely occur in patients with nonhyperosmolar diabetic ketoacidosis. These abnormalities include seizures, transient hemiparesis and other focal neurologic findings leading to the erroneous impression of stroke.4
Patients with hyperosmolarity who have adequate urine output and blood pressure on hospital admission may become oliguric and hypotensive after insulin lowers their glucose level (and osmolarity) to 250 to 350 mg per dL (13.9 to 19.4 mmol per L) unless total body fluid losses are concomitantly replaced. Once the glucosuric osmotic load dissipates, urine output falls to oliguric levels until total body water and intravascular volume are restored. These patients suffer from the “latent shock of dehydration.”9 This state may be made overt by the rapid correction of hyperglycemia with insulin without adequate repletion of free water.
In patients with uncontrolled diabetes, fluid losses range from 100 to 200 mL per kg. In patients with hyperosmolar hyperglycemia, the mean fluid loss is approximately 9 L.8
A calculation of effective osmolarity can serve as a guide in determining the type of fluid replacement to use.4,10 Normal serum osmolarity is 290 ± 5 mOsm per L. A rough approximation of the measured value can be obtained as follows:
In this equation, Na+ is the serum sodium level, K+ is the serum potassium level and BUN is the blood urea nitrogen level. In some calculations, the serum potassium level is omitted.
Because urea is freely diffusible across cell membranes, it contributes little to the “effective” serum osmolarity for the intracellular space. It is the effective osmolarity that is critical to the pathogenesis of the hyperosmolar state and the determination of the osmotic content of replacement solutions.
The distinction between osmolarity, which is the concentration of an osmolar solution, and tonicity, which is the osmotic pressure of a solution, must be made. Tonicity more appropriately reflects what is referred to as the effective osmolarity11 (Table 1).4 The initial fluid prescription in patients with uncontrolled diabetes should be isotonic until the effective osmolarity is known.
If the effective osmolarity exceeds 320 mOsm per L, significant hyperosmolarity exists, and this value guides the tonicity of the replacement solution used.4,10 Above an effective osmolarity of 320 mOsm per L, hypotonic crystalloids may be employed; below this value, isotonic electrolyte solutions are preferred.
The volume and rate of fluid replacement are determined by renal and cardiac function, as well as evidence of hypovolemia. Typically, replacement volumes are 1,500 mL in the first hour (15 to 30 mL per kg per hour), 1,000 mL in the second and possibly third hours, and 500 to 750 mL in the fourth hour and possibly beyond. For the physician, the key to fluid management is vigilant monitoring and adjustment of response on a continuous basis.
Volume replacement is critical in determining survival and in correcting hyperglycemic hyperosmolarity. One study12 of patients with diabetic ketoacidosis documented that hydration alone (without any insulin administration) reduced hyperglycemia, hyperosmolarity, acidosis and insulin counterregulatory hormone levels. Generally, however, insulin is also administered.
Early in treatment, a decrease in the plasma glucose level serves as an index of the adequacy of rehydration and the restoration of renal perfusion. Failure of the plasma glucose level to decline by 75 to 100 mg per dL (4.2 to 5.6 mmol per L) per hour usually implies either inadequate volume administration or impairment of renal function, rather than insulin resistance.4
The physician must be wary of uncontrolled diabetes in patients with renal failure. These patients may present with marked hyperglycemia and elevated BUN and creatinine levels. If renal failure is not recognized and rapid fluid replacement is instituted, the likely consequences are congestive heart failure and pulmonary edema.
In patients with hyperglycemia and renal failure, insulin administration may be all that is necessary. Insulin reduces elevated serum potassium levels. As the glucose level falls, water freed from its osmotic hold moves out of the extracellular fluid into the intracellular space (intracellular fluid), thereby decreasing the manifestations of circulatory congestion.
Sudden loss of diabetic control in the presence of renal failure may cause pulmonary edema and life-threatening hyperkalemia. Both of these conditions can be reversed by insulin alone.4
Electrolyte deficits in patients with uncontrolled diabetes are in the following ranges: sodium, 7 to 13 mEq per kg; chloride, 3 to 7 mEq per kg; potassium, 5 to 15 mEq per kg; phosphate (as phosphorus), 70 to 140 mmol; calcium, 50 to 100 mEq; and magnesium, 50 to 100 mEq.4,13
Chloride. A variety of physiologic multielectrolyte solutions (e.g., Plasmalyte, Isolyte, Normosol, lactated Ringer's solution) have been used in preference to normal saline solution2,4 to avoid the development of hyperchloremia, which is common when chloride is given in equivalent amounts with sodium in the treatment of ketoacidosis.14–16 These solutions also correct the “masked” hyperchloremic acidosis that sometimes occurs in patients with diabetic ketoacidosis.14–16
Sodium. Loss of sodium occurs because of osmotic diuresis and the absence of insulin, which is essential for distal tubular sodium reabsorption. Because sodium losses are proportionally less than water losses, hypernatremia may occur. In the absence of insulin, glucose is largely confined to the extracellular fluid compartment. The osmotic action of glucose causes water to flow out of the intra-cellular fluid into the extracellular fluid compartment, with consequent dilution of extra-cellular sodium.
One group of investigators17 demonstrated that the measured serum sodium concentration in the presence of hyperglycemia may be corrected by adding 2.4 mmol per L to the measured serum sodium concentration for each 100 mg per dL (5.6 mmol per L) that the plasma glucose level exceeds the normal value of 100 mg per dL.
Potassium. Total body potassium depletion is the rule in uncontrolled diabetes and related chronic catabolic states (e.g., starvation, chronic poorly controlled diabetes), although the presenting serum potassium concentration may be normal or high.18,19
As with phosphate and magnesium, the degree of total body potassium depletion is often unrecognized or underestimated (because of the initially elevated serum concentrations) until correction of the underlying catabolic state is undertaken. At this time, as potassium returns to its intracellular site (along with magnesium and phosphate) under the influence of insulin and renewed protein synthesis, the deficiency may become chemically and, occasionally, clinically overt unless supplements are administered during the early phases of therapy.
Recommendations for potassium administration in adults are provided in Table 2.4 Potassium replacement should begin in the first hour of treatment if urine output is adequate, the electrocardiogram (ECG) shows no evidence of hyperkalemia and the serum potassium level is less than 5 mEq per L (5 mmol per L). The use of potassium acetate, potassium phosphate or a mixture of the two avoids administration of excess chloride.2,4
Suggested Treatment Protocol for Patients with Hyperosmolar Hyperglycemia
The rightsholder did not grant rights to reproduce this item in electronic media. For the missing item, see the original print version of this publication.
The objective of potassium replacement is to maintain normokalemia. The total body deficits of potassium cannot and should not be replaced acutely. Full correction of the potassium level requires days to weeks of steady anabolism.
Even in the presence of renal insufficiency, early potassium replacement may be indicated because, with the exception of urinary loss, all other factors that drive the serum potassium level down in the treated patient are occurring. Caution must be exercised in patients with diabetic nephropathy who may have renal tubular acidosis associated with hyperaldosteronism and hyporeninemia (and resultant hyperkalemia).
Once insulin is administered, potassium moves intracellularly because of either insulin stimulation of sodium-potassium adenosine triphosphatase or insulin-induced synthesis of phosphate esters intracellularly. These anions attract potassium into the cells.19 Chloride is predominantly an extracellular anion. Thus, to replace intracellular losses, potassium should be given with an anion that distributes in the intracellular fluid (i.e., phosphate instead of chloride).
Phosphate. The benefits and risks of phosphate repletion in patients with uncontrolled diabetes have been investigated primarily in patients with acute diabetic ketoacidosis.18,20,21 In patients who were not given supplemental phosphate, the mean nadir of the phosphate concentration was only in the moderately hypophosphatemic range.22 It has been suggested23 that when diabetic decompensation occurs over a period of days or weeks in association with a prolonged osmotic diuresis (as is typical in hyperglycemic hyperosmolar states), total body stores are more likely to be depleted than in an acute illness (i.e., one with a duration of hours to several days) such as diabetic ketoacidosis.
Chronic catabolic states, including hyperosmolar diabetes, are likely to be associated with more severe total body phosphate depletion and its unmasking during treatment, as originally noted in the refeeding syndrome.24 This is not surprising, because phosphate is the predominant intracellular anion.
Although phosphate replacement makes physiologic sense, no controlled data demonstrate that it alters the outcome or contributes to survival in patients with uncontrolled diabetes unless severe reductions in the serum phosphate level (i.e., to less than 1.0 to 1.5 mEq per L [0.32 to 0.48 mmol per L]) are evident. The only risk of giving phosphate is the occasional development of hypocalcemic tetany described in diabetic ketoacidosis,22 but this condition does not occur when magnesium is supplemented.4
Magnesium. As many as 40 percent of outpatients with diabetes and 90 percent of patients with uncontrolled diabetes after 12 hours of therapy are hypomagnesemic.18 Manifestations previously ascribed to hypokalemia and hypophosphatemia may also be due to hypomagnesemia. These manifestations include ECG changes, arrhythmias, muscle weakness, convulsions, stupor, confusion and agitation.25
As is true of the serum potassium level, the serum magnesium concentration is an unreliable marker of total body stores of this predominantly intracellular cation. Many patients have elevated serum magnesium levels on presentation, and hypomagnesemia may not be evident for hours. Serum magnesium levels and body stores parallel and mirror those of potassium in hyperosmolar hyperglycemic syndrome. Unless precluded by renal failure of hypermagnesemia, the routine administration of magnesium to patients with uncontrolled diabetes is safe and physiologically appropriate.
Because many patients with diabetes and hyperosmolar hyperglycemic syndrome fit the criteria for catabolic malnourishment, they are at risk for development of the refeeding syndrome.26 Thus, the administration of a B-complex vitamin supplement, especially thiamine, is prudent.4,20
Concern, primarily focused on the potential for cerebral edema, has been expressed about the effects of rapid fluid administration, the use of hypotonic fluids to treat hyperglycemic hyperosmolarity and the dangers of rapid reduction of the effective osmolarity to 320 mOsm per L or the glucose level to 250 to 300 mg per dL (13.9 to 16.7 mmol per L). Cerebral edema occurs almost exclusively in young patients who have nonhyperosmolar diabetes with diabetic ketoacidosis. This complication is exceedingly rare in adult patients with hyperosmolar hyperglycemic syndrome.3,4,10
In the United States, diabetic ketoacidosis and hyperosmolar diabetes are responsible for thousands of deaths every year, whereas fatal cerebral edema in adults almost never occurs.10 Therefore, in light of the potentially dire consequences of undertreatment, it is indefensible to base therapeutic recommendations on the theoretic prevention of this rarely encountered complication, especially because no study has demonstrated that adherence to recommendations for slower correction of hyperglycemic hyperosmolarity produces the desired benefits.
Many more patients with uncontrolled diabetes die as a result of undertreatment than overtreatment. Any recommendation to reduce the rate of correction of hyperglycemia risks increasing the mortality rate and should be viewed with alarm. Rapid correction of hyperosmolarity to an effective osmolarity of 320 mOsm per L and the plasma glucose level to 250 to 300 mg per dL (13.9 to 16.7 mmol per L) is the goal. Thereafter, a much slower rate of correction toward normal is warranted.2,4,10
Insulin has a direct antinatriuretic effect on the kidney. Following treatment with crystalloid solutions and insulin, patients may develop significant edema, termed “insulin edema.” Occasionally, patients with preexisting cardiovascular disease may develop frank congestive heart failure, pulmonary edema or hypertension.26,27
Vascular occlusions (i.e., mesenteric artery occlusion, myocardial infarction, low flow syndrome and disseminated intravascular coagulation) are important complications of hyperosmolar hyperglycemic syndrome. With aggressive rehydration, the incidence of these complications may be reduced to 2 percent.2,4
Improving the management results for hyperosmolar hyperglycemic syndrome requires an effective preventive strategy. Patients with hyperosmolar diabetes tend to be elderly and to have type 2 diabetes. Often, these patients live alone, and social isolation is a precipitant in one third to one half of hyperosmolar diabetes episodes.
A knowledgeable “significant other” who maintains daily contact with the elderly diabetic patient is essential. If the patient displays any change in mental status or symptoms of loss of diabetic control, this person is instructed to contact the patient's physician immediately.
Nursing home residents are prone to develop dehydration, hyperosmolarity and hyperglycemia. Thus, education of nursing home staff in the prevention and detection of diabetes-related problems is also essential.4
REFERENCESshow all references
1. Fishbein H, Palumbo PJ. Acute metabolic complications in diabetes. In: National Diabetes Data Group, eds. Diabetes in America. 2d ed. Washington, D.C.: National Institutes of Health, 1995:283–91....
2. Carroll P, Matz R. Uncontrolled diabetes mellitus in adults: experience in treating diabetic ketoacidosis and hyperosmolar nonketotic coma with low-dose insulin and a uniform treatment regimen. Diabetes Care. 1983;6:579–85.
3. Ennis ED, Stahl EJ, Kreisberg RA. The hyperosmolar hyperglycemic syndrome. Diabetes Rev. 1994;2:115–26.
4. Matz R. Hyperosmolar nonacidotic diabetes (HNAD). In: Porte D, Sherwin RS, Ellenberg M, Rifkin H, eds. Ellenberg and Rifkin's Diabetes mellitus. 5th ed. Stamford, Conn.: Appleton & Lange, 1997:845–60.
5. Matz R, Drapkin A. Hyperosmolar coma in diabetes [Letter]. Lancet. 1966;1:1101.
6. Brodsky WA, Rapaport S, West CD. The mechanism of glycosuric diuresis in diabetic man. J Clin Invest. 1950;29:1021–32.
7. Umpierrez GE, Kelly JP, Navarette JE, Casals MM, Kitabchi AE. Hyperglycemic crises in urban blacks. Arch Intern Med. 1997;157:669–75.
8. Siperstein M. Diabetic ketoacidosis and hyperosmolar coma. Endocrinol Metab Clin North Am. 1992;21:415–32.
9. Haapenen E. Hyperosmolar coma in diabetes [Letter]. Lancet. 1966;1:1154.
10. Matz R. How big is the risk of cerebral edema in adults with DKA? J Crit Illn. 1996;11:768–72.
11. Arieff AI, Carroll HJ. Hyperosmolar nonketotic coma with hyperglycemia: abnormalities of lipid and carbohydrate metabolism. Metabolism. 1971;20:529–38.
12. Waldhausl W, Kleinberger G, Korn A, Dudczak R, Bratusch-Marrain P, Nowotny P. Severe hyperglycemia: effects of rehydration on endocrine derangements and blood glucose concentration. Diabetes. 1979;28:577–84.
13. Ennis ED, Stahl EJ, Kreisberg RA. Diabetic ketoacidosis. In: Porte D, Sherwin RS, Ellenberg M, Rifkin H, eds. Ellenberg and Rifkin's Diabetes mellitus: theory and practice. 5th ed. Stamford, Conn.: Appleton & Lange, 1997:827–44.
14. Androgue HJ, Wilson H, Boyd AE 3d, Suki WN, Eknoyan G. Plasma acid-base patterns in diabetic ketoacidosis. N Engl J Med. 1982;307:1603–10.
15. Androgue HJ, Eknoyan G, Suki WK. Diabetic keto-acidosis: role of the kidney in acid-base homeostasis re-evaluated. Kidney Int. 1984;25:591–8.
16. Androgue HJ, Barrero J, Eknoyan G. Salutary effects of modest fluid replacement in the treatment of adults with diabetic ketoacidosis. Use in patients without extreme volume deficit. JAMA. 1989;262:2108–13.
17. Hiller TA, Abbot D, Barrett EJ. Hyponatremia: evaluating the correction factor for hyperglycemia. Am J Med. 1999;106:399–403.
18. Matz R. Parallels between treated uncontrolled diabetes and the refeeding syndrome with emphasis on fluid and electrolyte abnormalities. Diabetes Care. 1994;17:1209–13.
19. Halperin ML, Goldstein MB. Fluid, electrolyte, and acid-base emergencies. Philadelphia: Saunders, 1988:212–4.
20. Fisher JN, Kitabchi AE. A randomized study of phosphate therapy in the treatment of diabetic ketoacidosis. J Clin Endocrinol Metab. 1983;57:177–80.
21. Zipf WB, Bacon GE, Spencer ML, Kelch RP, Hop-wood NJ, Hawker CD. Hypocalcemia, hypomagnesemia, and transient hypoparathyroidism during therapy with potassium phosphate in diabetic ketoacidosis. Diabetes Care. 1979;2:265–8.
22. Keller U, Berger W. Prevention of hypophosphatemia by phosphate infusion during treatment of diabetic ketoacidosis and hyperosmolar coma. Diabetes. 1980;29:87–95.
23. Bohannon NJ. Large phosphate shifts with treatment for hyperglycemia. Arch Intern Med. 1989;149:1423–5.
24. Solomon SM, Kirby DF. The refeeding syndrome: a review. J Parenter Enteral Nutr. 1990;14:90–7.
25. Matz R. Magnesium: deficiencies and therapeutic uses. Hosp Pract. [Off Ed] 1993;28(4A):79–82.
26. Saudek CD, Boulter PR, Knopp RH, Arky RA. Sodium retention accompanying insulin treatment of diabetes mellitus. Diabetes. 1974;23:240–6.
27. Blumenthal SA. Observation on sodium retention related to insulin treatment of experimental diabetes. Diabetes. 1975;24:645–9.
This article is one in a series developed in collaboration with the American Diabetes Association. Guest editors of the series are Bruce Zimmerman, M.D., and Richard Kahn, Ph.D., American Diabetes Association, Alexandria, Va.
Copyright © 1999 by the American Academy of Family Physicians.
This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP. Contact firstname.lastname@example.org for copyright questions and/or permission requests.
Want to use this article elsewhere? Get Permissions