Recent advances in the monitoring and management of diabetic ketoacidosis

/Home/Complications of Diabetes/Diabetic Ketoacidosis (DKA)/Diabetic Ketoacidosis Treatment

Recent advances in the monitoring and management of diabetic ketoacidosis
T.M. Wallace and D.R. Matthews

From the Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Hospital, Oxford, UK

Summary

Diabetic ketoacidosis (DKA) is still a major contributor to morbidity and mortality in diabetes. The triad of hyperglycaemia, ketosis and acidosis can be diagnosed within a few minutes of the patient presenting, by measuring blood glucose and ketones using a meter, and venous blood pH on a blood gas analyser. Quantifying ketosis allows accurate distinction between simple hyperglycaemia and metabolic decompensation. We review the management of DKA, and the emerging role of near-patient testing in diagnosing ketosis and monitoring its resolution.

Background

Diabetic ketoacidosis (DKA) is an acute and potentially fatal complication of diabetes typically characterized by hyperglycaemia, ketone body formation and metabolic acidosis. Over the past 20 years, there has been no reported reduction in mortality rates, which remain between 3.4 and 4.6%.1 Mortality is predominantly due to underlying morbidity, such as sepsis or acute myocardial infarction, but deaths also occur as a result of hypokalaemia-induced arrhythmias and cerebral oedema.

Diagnosis

The incidence of DKA is difficult to ascertain, as definitive criteria for the diagnosis have not been agreed, and there is no national register for recording cases. The diagnosis is based on the clinical features of uncontrolled diabetes in the presence of ketosis and acidosis (i.e. arterial blood pH <7.3), although in practice patients with a bicarbonate <15 mmol/l in the presence of a normal pH are often classified as having ketoacidosis. The presence of ketones can now be established by finger-prick testing, results being available within 30 s. We review the advantages of using such a system in the diagnosis and management of decompensated diabetes.

Pathophysiology

The pathology is invariably the result of relative or absolute insulin deficiency which, in combination with increased levels of stress hormones, stimulates lipolysis resulting in the production of acetylCoA from fatty acid. This acts as the substrate for hepatic synthesis of ketone bodies (acetoacetate, beta-hydroxybutyrate and acetone) (Figure 1). In addition to increased lipolysis, the relative lack of insulin in DKA results in decreased glucose utilization and increased gluconeogenesis.

Figure 1. (a) Increased lipolysis results in the production of acetylCoA from fatty acids which acts as the substrate for hepatic synthesis of ketone bodies (acetoacetate, beta-hydroxybutyrate and acetone). A lack of insulin results in decreased glucose utilization and a reduction in oxaloacetate production. (b) The amount of oxaloacetate available for condensation with acetylCoA is reduced; and (c) acetylCoA is thus diverted from entering the tricarboxylic acid cycle and (d) undergoes condensation to form acetoacetate followed by reduction to beta-hydroxybutyrate.

Measurement of ketones

The commonly used urine dipstick tests are based on the use of a nitroprusside reaction (Ketostix, Acetest) and give a semi-quantitative measure of acetoacetate, react weakly with acetone but do not register the presence of beta-hydroxybutyrate (?-OHB). Acetone is responsible for the characteristic Îpear-dropÌ smell present in DKA, but does not contribute to acidosis. In DKA, the ratio of ?-OHB to acetoacetate increases from 1:1 to as much as 5:1 and thus ?-OHB is the predominant ketone body contributing to the acidosis.2 When acidosis resolves with treatment, ?-OHB is oxidized to acetoacetate. Under these circumstances, urine tests may give the misleading impression that ketosis is not improving, and from a practical point of view there can be a problem obtaining urine samples from severely dehydrated patients at the time of presentation. Blood ketones can now be measured directly with a hand-held sensor (Optium meter, Medisense/Abbott) which measures ?-OHB in 30 s using blood from a fingerprick test.3 The meter has been shown to be reliable, with accuracy and precision that is well within acceptable clinical limits.3,4 The use of blood ketone tests based on the measurement of ?-OHB, rather than urine ketone tests, for diagnosis and monitoring of DKA, is now recommended by the American Diabetes Association.5

Distinction between ketosis and acidosis

Ketosis and acidosis are not synonymous, because the increase in H+ concentration resulting from the production of ketone acids is initially buffered by bicarbonate. As the concentration of H+ exceeds the buffering capacity of bicarbonate, bicarbonate reserves become depleted and are no longer able to compensate for the excess production of H+ ions and acidosis results. During the initial compensated phase of a metabolic acidosis, the usual picture is of low bicarbonate and normal pH, the latter being maintained at the expense of bicarbonate loss. Thus it is essential to measure blood pH, and a venous sample is adequate for this purpose.6

Special situations

n a small percentage of cases of DKA, the glucose levels are not elevated at presentationÛÎeuglycaemic DKAÌ. This usually occurs in association with excessive vomiting and continued insulin administration. The role of monitoring ?-OHB is particularly valuable in euglycaemic (or near-euglycaemic) DKA, as the relatively low blood glucose levels in this state are a poor guide to the underlying metabolic derangement, and can result in inadequate insulin replacement. Euglycaemic DKA has been reported to occur in 10% of cases of DKA complicating pregnancy. Although there is no increase in maternal mortality in DKA during pregnancy, fetal mortality is high, and is reported to be in the region of 35%.7 African-Americans8 and black South Africans9 with type 2 diabetes may present with diabetic ketoacidosis, but following acute treatment with insulin they often revert to good control with oral hypoglycaemic agents.

Precipitants and clinical features of DKA

The most common identifiable causes of DKA are infection in 28Ò43% of cases and inadequate insulin doses in 13Ò45% (Table 1).10Ò13 The most frequently diagnosed infections are pneumonia and urinary tract infections.14 Other acute medical illnesses may precipitate DKA including pulmonary embolism, myocardial infarction, cerebrovascular accidents and protracted vomiting.

Patients typically present with a short history of symptoms developing over a few days. The cardinal symptoms are increasing polyuria and polydipsia, variable weight loss and weakness, followed by drowsiness, decreased level of consciousness and eventually coma (in 10%) (Table 2). Signs of dehydration and hypovolaemia (reduced skin turgor, hypotension and tachycardia) are common. Other features include the presence of Kussmaul ventilation (hyperventilation due to respiratory compensation for a metabolic acidosis) and a detectable odour of acetone on the patient's breath. DKA may be associated with nausea, vomiting or abdominal pain: an Îacute abdomenÌ may precipitate DKA, or the abdominal pain may be due to the metabolic acidosis; in the latter case, the pain will improve as the acidosis resolves.15

Investigations in DKA

The triad of hyperglycaemia, ketosis and acidosis can be diagnosed within a few minutes of the patient presenting by measuring blood glucose and ketones using a meter and venous blood pH on a blood gas analyser. Baseline investigations required in DKA are shown in Table 3. The differential diagnosis of marked hyperglycaemia with dehydration is the hyperosmolar non-ketotic state (HONK), which is characterized by hyperosmolarity in the absence of ketones.

Some blood results may be misleading in the presence of DKA: transaminases are raised in 25Ò50% of cases with no evidence of hepatocellular disease, and creatine kinase is elevated in 25Ò40% of cases in the absence of any evidence of myocardial infarction.16,17 Amylase and lipase levels are elevated in 16Ò25% of cases in the absence of any clinical evidence of pancreatitis; although acute pancreatitis may accompany or precipitate DKA, the diagnosis should not be based solely on the finding of elevated amylase or lipase levels, which may be increased by threefold.17 A full blood count often shows a non-specific leukocytosis. A raised creatinine may reflect dehydration and/or pre-renal renal failure, but may also be falsely elevated due to interference from acetoacetate with some automated creatinine assays.18

Management of DKA

Fluid replacement
The aim is to replace the fluid deficit (average 6 l) over approximately 24 h, but the rate of replacement depends on medical co-morbidity. Rehydration should initially be with intravenous 0.9% saline at a rate of the order of 1 litre over 30 min followed by 1 litre over 2 h, 1 litre over 4 h, 1 litre over 6 h and then 1 litre 6-hourly. An accurate record of fluid balance is essential. As the glucose falls, intravenous dextrose (5Ò10%) may be required. With prolonged use of dextrose, there is potential for hyponatraemia and if the sliding scale needs to be continued for several days after the biochemistry has corrected (e.g. if patient is not eating), then it may be appropriate to switch to using 5% dextrose alternating with 0.9% saline.

Insulin replacement
The aim of therapy is to switch off ketogenesis and to correct the metabolic derangement smoothly and the evidence supporting the use of low-dose intravenous insulin infusion is well established.19Ò21 Treatment should aim to reduce glucose levels by 5Ò6 mmol/l per hour.

Sliding scale regimen: soluble insulin (e.g. 50 units Actrapid in 50 ml N saline) is usually given by continuous intravenous infusion according to a sliding scale (Table 4). Capillary glucose and ketones should be measured hourly and the insulin infusion rate adjusted accordingly. If the glucose and ketones do not fall by at least 5 mmol/l per hour and 0.5 mmol/l per hour, respectively, the fluid regimen and sliding scale should be reviewed.

If an infusion pump is not available for the administration of insulin, an intramuscular regimen22 can be used: 20 units soluble insulin (e.g. Actrapid) is given intramuscularly initially, followed by 5 units hourly until the blood glucose falls below 15 mmol/l. Once the glucose is <15 mmol/l, the route of administration is switched from intramuscular to subcutaneous.

Potassium replacement
Regular monitoring of electrolytes (at baseline, 2 h, 6 h and further as indicated) should guide potassium replacement. However plasma potassium is not a reliable indicator of total body potassium, which is usually depleted. No potassium is required in the first litre of fluid unless K+<3.5 mmol/l. A guide to potassium replacement thereafter is suggested in Table 5.

Other measures
Continuous ECG monitoring is recommended in view of the risk of hypo- or hyperkalaemia and consequent arrhythmias. A nasogastric tube should be inserted if the patient has an impaired level of consciousness, because of the risk of gastroparesis and aspiration. Urinary catheterization should be considered if there is impaired level of consciousness, or if the patient has not passed urine 4 h after treatment is commenced. The need for central venous monitoring should be assessed on an individual basis, but may be required in elderly patients or those with pre-existing cardiac failure. The administration of sodium bicarbonate is virtually never indicated.

Consideration should be given to antibiotic therapy if there is evidence of infection, but white cell count is often spuriously raised in DKA and does not confirm infection (history, examination, the presence of pyrexia and elevated C-reactive protein are more helpful markers).

Interpretation of capillary ketone levels
Under normal circumstances, ?-OHB concentrations do not exceed 1 mmol/l in type 1 diabetic subjects.4,23 In patients presenting with DKA, mean ?-OHB is about 7 mmol/l but can range between 4 and 12 mmol/l.2,4,24Ò26 With optimal treatment of DKA, ?-OHB would be expected to fall by 1 mmol/l per hour;4 failure to do so suggests inadequate treatment, and insulin and fluid infusion rates should be reviewed.

Monitoring the resolution of DKA
The aim of treatment in DKA is to switch off ketogenesis, with consequent resolution of acidosis. The measurement of capillary glucose levels have to some degree been used as surrogate markers of the resolution of metabolic derangement. However, glucose levels are often not particularly stable on a sliding scale, and if hyperglycaemia occurs, it is important to know whether this is accompanied by ketosis, implying metabolic decompensation, or not. Figure 2 shows data from two patients admitted with DKA (from reference 4) in whom the measurement of ?-OHB helped to distinguish between the recurrence of ketosis requiring the reinstatement of intravenous insulin therapy (Figure 2a) and simple hyperglycaemia (Figure 2b).

Figure 2. Glucose and ?-OHB profiles in two patients admitted with DKA. There is an initial reduction in ?-OHB and glucose with treatment. When glucose levels rise at 14Ò16 h, it is important to know whether this is accompanied by ketosis (a), implying metabolic decompensation, or not (b).

The recent introduction of a blood ketone meter allows insulin administration to be titrated against falling ketone levels rather than glucose levels. Obviously, it remains important to continue to monitor glucose levels in order to minimize the risk of hypoglycaemia. But if hypoglycaemia is a problem in the face of continuing ketosis, insulin administration should be continued with the addition of glucose infusion, e.g. 10Ò20% dextrose. Monitoring capillary ketone levels has the advantage of being much less invasive than arterial blood gas monitoring, and is a more appropriate measure than plasma bicarbonate levels as the replenishment of bicarbonate stores lags behind the resolution of the metabolic acidosis. The time course of bicarbonate and ?-OHB in a patient admitted with DKA are illustrated in Figure 3 (from reference 4).

Figure 3. The time course for blood glucose (squares), ?-hydroxybutyrate (?-OHB, diamonds) and bicarbonate (triangles) in hours from initiation of treatment. The lower dotted line represents the upper limit of normal for ?-OHB, and the higher dotted line represents the lower limit of normal for bicarbonate.


Long-term management
The patient should be switched to subcutaneous insulin when the ketone reading is <1.0 mmol/l, the blood glucose is reasonably controlled, and the patient is clinically well and eating normally. Intravenous insulin should be discontinued 1 h after the first subcutaneous dose. Patients admitted with DKA should be referred to a diabetes team for education and follow-up.

Prevention of DKA

Education aimed at raising the awareness of the symptoms of diabetes is essential in order to prevent DKA in newly presenting type 1 diabetes.27 For prevention of DKA in established type 1 diabetes, the primary issue is that of early detection and prompt intervention in cases of metabolic decompensation. Although home blood glucose monitoring is well established in diabetic patients, the development of high glucose levels (>15 mmol/l) can create a dilemma for patients and healthcare professionals managing type 1 diabetes, as rising levels are associated with an increasing risk of developing ketoacidosis. The availability of point-of-contact testing by patients and healthcare professionals should reduce unnecessary visits to hospital and avoid the need for more invasive tests. ?-OHB levels do not normally exceed 1 mmol/l in patients with type 1 diabetes who have no evidence of metabolic decompensation, irrespective of the prevailing glucose concentration.4 Figure 4 shows guidelines for the use of the meter in out-patients.

Financial cost of DKA

The average cost of treating a single episode of DKA in the US is $6500Ò$7500, which represents about 25% of the total spent on the care of patients with type 1 diabetes.28,29 These figures do not take into account any costs arising from time off work. Monitoring ?-OHB levels in children with DKA reduces the length of stay on the intensive care unit, with an average saving of $16330 or {euro}184 per patient.31 We suggest that the introduction of routine ?-OHB monitoring could not only reduce the duration of hospital admission, but may also help to prevent episodes of DKA, with a consequent reduction in healthcare expenditure.

Conclusion

In summary, blood ketone testing is a useful adjunct to glucose monitoring and allows clinicians and patients to distinguish between simple hyperglycaemia and a potentially life-threatening ketotic state. The facility to measure blood ketones should be available in A&E departments, GP surgeries, diabetes centres, and for self-monitoring in patients with type 1 diabetes. Monitoring ketones is also clinically useful in the management of DKA in children.32 Further studies on the impact of measuring ketones on the management and prevention of DKA are needed; in particular, there are no randomized controlled trials of the use of routine ketone monitoring in DKA, although the logistics and ethics of establishing such a trial would be problematic.

Although there has been no reported reduction in mortality rates in DKA over the past 20 years, it is difficult to ascertain the incidence of DKA and its mortality, due to the lack of a national reporting system. The introduction of a national register would allow accurate statistics on incidence, complications, mortality and treatment to be collected. This would allow a structured audit of the management to be undertaken, which could form the basis for multi-centre or national prospective studies to determine the optimal management strategy.


Acknowledgments

TMW is the Oxford Servier Clinical Diabetes Research Fellow.


Footnotes

Address correspondence to Professor D.R. Matthews, Oxford Centre for Diabetes, Endocrinology and Metabolism, The Churchill Hospital, Old Road, Oxford OX3 7LJ. e-mail: david.matthews@ocdem.ox.ac.uk