Serum beta-hydroxybutyrate

One study (total 129 participants) showed that serum beta-hydroxybutyrate of 3 mmol/litre or more (defined by serum bicarbonate 18 mEq/litre or less) has a high sensitivity and moderate specificity for the diagnosis of DKA in children and young people. The positive likelihood ratio indicated that the test was moderately useful and the negative likelihood ratio indicated that it was very useful. The evidence for these findings was of very low quality.

One study (total 90 participants) showed that serum beta-hydroxybutyrate of 3 mmol/litre or more (defined by venous pH of less than 7.3) has a moderate sensitivity and low specificity for the diagnosis of DKA in children and young people. The positive and negative likelihood ratios indicated that the test was not useful. The evidence for these findings was of moderate quality.

One study (total 110 participants) showed that serum beta-hydroxybutyrate of 3 mmol/litre or more (defined by blood glucose more than 13.9 mmol/litre) has a low sensitivity and moderate specificity for the diagnosis of DKA in children and young people. The positive and negative likelihood ratios indicated that the test was not useful. The evidence for these findings was of moderate quality.

End-tidal carbon dioxide

Two studies (total 107 participants) showed that end-tidal carbon dioxide has a moderate sensitivity and high specificity for the diagnosis of DKA in children and young people (using a cut-point of 29 mmHg or less). The positive likelihood ratio indicated that this test was very useful. The negative likelihood ratio suggested that the test was moderately useful. The evidence for these findings was of low quality.

One study (total 58 participants) reported the diagnostic test accuracy of end-tidal carbon dioxide (using a cut-point of 30 mmHg or less) for the diagnosis of DKA in children and young people. The study reported a high sensitivity, moderate specificity, moderately useful positive likelihood ratio and very useful negative likelihood ratio. The evidence for these findings was of low quality.

One study (total 42 participants) reported the diagnostic test accuracy of end-tidal carbon dioxide (using a cut-point of less than 36 mmHg) for the diagnosis of DKA in children and young people. The study reported a high sensitivity, low specificity, not useful positive likelihood ratio and very useful negative likelihood ratio. The evidence for these findings was of high quality.

18.2.1.7.1. Relative value placed on the outcomes considered

Children and young people (with or without a prior diagnosis of diabetes) presenting with DKA may have symptoms consistent with diabetes (for example polyuria, polydipsia, weight loss and tiredness) and also symptoms which are recognised as occurring in ketoacidosis (for example nausea, vomiting, abdominal pain, respiratory distress, dehydration and reduced level of consciousness). Many of these clinical manifestations are non-specific and might be reported in other clinical contexts. The guideline development group considered that it was important to review available evidence on the predictive value of these various symptoms and signs. In addition, they wished to consider evidence on the predictive value of biochemical abnormalities associated with hyperglycaemia, acidosis and ketosis.

18.2.1.7.2. Consideration of clinical benefits and harms

DKA is a life-threatening complication of diabetes and so the harm associated with missing the diagnosis far outweighs the minimal harm from over-investigation. The tests to be undertaken to identify DKA or to rule it out consist of minimally invasive procedures such as blood testing for hyperglycaemia, pH and ketone levels. The guideline development group's view was that it was particularly important that the tests employed for the diagnosis of DKA should have a high sensitivity (90% or higher).

18.2.1.7.3. Consideration of health benefits and resource use

The cost of delayed or missed diagnosis of DKA is considerable and could result in death. The longer a diagnosis of ketosis is delayed, the sicker the child or young person becomes and the more intensive and protracted treatment is likely to be. For example, early ketosis causes nausea and vomiting and is easily mistaken for food poisoning or gastro-enteritis. This often leads to inadequate insulin being given for fear of causing hypoglycaemia. The individual then becomes much sicker. Near-patient beta-hydroxybutyrate, using ketone strips, is cheaper than a laboratory measurement and a more rapid diagnosis may bring important benefits for this life-threatening condition.

18.2.1.7.4. Quality of evidence

Although there was no available evidence on the predictive value of clinical symptoms or signs individually or in combination, the guideline development group made recommendations based on their clinical knowledge and experience regarding the clinical manifestations that should lead to investigation of DKA.

In the first instance they considered the clinical features that would raise the possibility of DKA in a child or young person not previously known to have diabetes. They agreed that if there was increased thirst, polyuria, recent unexplained weight loss or excessive tiredness (these are recognised clinical manifestations of diabetes mellitus) together with any of a number of other features (nausea or vomiting, abdominal pain, hyperventilation, dehydration or a reduced level of consciousness) then a capillary blood glucose measurement should be obtained. All of these other features are accepted in clinical practice as suggestive of DKA in a person in whom diabetes is suspected or confirmed. If the blood glucose was normal then DKA would be ruled out as an explanation for the symptoms. If the blood glucose was greater than 11 mmol/litre then DKA should be suspected as the likely explanation and the child or young person should immediately be sent to a hospital with acute paediatric facilities.

When children and young people with known diabetes develop DKA they would often have a raised blood glucose but this is not universal. The guideline development group noted that it was well recognised that it is possible for those with a prior diagnosis of diabetes who are receiving insulin therapy to have DKA with a normal glucose level. Some young children with type 1 diabetes and gastroenteritis can develop ketoacidosis with hypo- or normo-glycaemia because of relative insulin deficiency and starvation. This important point was therefore highlighted in a recommendation. The group recommended that DKA should be suspected in any child or young person with diabetes irrespective of the blood glucose level if they had any of the suggestive clinical features (nausea or vomiting, abdominal pain, hyperventilation, dehydration or a reduced level of consciousness). When DKA is suspected in a child or young person with known diabetes the blood ketones (beta-hydroxybutyrate) should be measured using a near-patient method if this is available. Those presenting to their GP, for example, might well have the testing equipment to hand. If the level of beta-hydroxybutyrate is elevated, then they should be sent immediately to a hospital with acute paediatric facilities because the diagnosis of DKA is likely. If it is not possible in this setting to measure the beta-hydroxybutyrate level then DKA should be suspected and they should be sent immediately to a hospital with acute paediatric facilities.

The guideline development group noted that 2 studies that evaluated diagnostic test accuracy of blood beta-hydroxybutyrate found evidence to support its use in diagnosis, but this was of very low to moderate quality. These studies investigated diagnostic test accuracy of beta-hydroxybutyrate only in terms of its ability to predict DKA based on a single parameter (blood glucose, blood pH or blood ketones), not the combination of these parameters (see below) required to definitively diagnose DKA. Thus the reference test in these studies was not a true ‘gold standard’ diagnosis of DKA. This limited the applicability of serum beta-hydroxybutyrate as a definitive test for diagnosing DKA. Nevertheless, if the level were elevated the guideline development group considered this made a diagnosis of DKA likely in this setting and if it were normal then DKA would be ruled out (ketosis being an essential component of DKA).

The guideline development group was concerned that whenever DKA was suspected or confirmed, the child or young person and their family members or carers should be fully aware of the serious nature of the concern and that urgent hospital assessment was mandatory – and they made a specific recommendation in this regard. Any delay in attending hospital carried a significant risk. The group recommended that when a child or young person did arrive at the hospital with suspected or confirmed DKA, investigations should be carried out to confirm the suspicion (or rule it out): these would include a capillary blood glucose, capillary blood ketones using a near-patient method (if near-patient testing for capillary blood ketones is not available, use a urine ketone test instead) and lastly a capillary or venous pH and bicarbonate test. These were essential to confirm the diagnosis. The group made a recommendation stating the diagnostic criteria for DKA based on the demonstration of acidosis and ketosis or ketonaemia. They made a recommendation that for the purposes of this guideline and its recommendations on management, DKA should be categorised as severe based on the finding of a blood pH below 7.1.

The group was aware that laboratory measurement of beta-hydroxybutyrate takes time and may delay the diagnosis of DKA. Rapid diagnosis is important in this setting as DKA is potentially life-threatening and the group agreed that near-patient testing on arrival at the hospital would facilitate rapid diagnosis and was advisable.

The only other included studies on diagnostic tests for DKA examined end-tidal nasal carbon dioxide measurement. Although 1 of these studies provided high-quality evidence suggesting that end-tidal nasal carbon dioxide had a high sensitivity and useful negative likelihood ratio, the group was not persuaded that it was of practical value. Moreover, this form of test was not in general use in this setting. There were concerns about the diagnostic test accuracy of measurements with this technique and, importantly, it was often poorly tolerated. The guideline development group did not recommend it.

18.2.1.7.5. Other considerations

The guideline development group was aware of evidence that there may be an increased risk of DKA in certain ethnic groups and children and young people living in deprived areas (Khare 2013). The group made recommendations about the importance of considering contributory factors in those who present with an episode of DKA with a view to reducing the risk of future episodes (see Section 18.7).

18.2.1.7.6. Key conclusions

The guideline development group recommended that the biochemical criteria required for the diagnosis of DKA should be as discussed above. The group specifically recommended that healthcare professionals should measure capillary blood glucose at presentation in children and young people without known diabetes who have increased thirst, polyuria, recent unexplained weight loss or excessive tiredness and any of the following:

• nausea or vomiting
• abdominal pain
• hyperventilation
• dehydration
• reduced level of consciousness.

The group also recommended that if the blood glucose level is above 11 mmol/litre in a child or young person without known diabetes and they have symptoms that suggest DKA, then DKA should be suspected and the child or young person should be sent immediately to a hospital with acute paediatric facilities. A blood glucose of this level would be consistent with a diagnosis of diabetes, and when there are also symptoms suggestive of DKA immediate referral to hospital is essential because DKA is a life-threatening condition requiring urgent management.

The guideline development group emphasised that healthcare professionals should be aware that children and young people taking insulin for diabetes may develop DKA with normal blood glucose levels, and that they should suspect DKA (even if the blood glucose is normal) in children and young people with known diabetes and also in children and young people without known diabetes if any symptoms and signs are present that would trigger capillary blood glucose measurement.

The guideline development group further recommended that when DKA is suspected in a child or young person with known diabetes, blood ketones (beta-hydroxybutyrate) should be measured using a near-patient method if available, and if the level is elevated, the child or young person should be sent immediately to a hospital with acute paediatric facilities. This will identify those with DKA even if the blood glucose is in the normal range, as may happen in children and young people with type 1 diabetes who are using insulin therapy. The group considered whether a specific threshold for blood ketones should be stated in the recommendation about sending the child or young person to a hospital with acute paediatric facilities. The group concluded that a threshold should not be stated because the evidence reviewed for the guideline did not support ketone testing as being a specific test for DKA, and the recommendation should not risk preventing the child or young person being sent to hospital by including an arbitrary threshold that may not quite be met in individual circumstances. The recommendation is not about diagnosing DKA (this will be done in the hospital) and a child or young person with known type 1 diabetes should already have ketone testing equipment and advice about seeking help plus an individualised sick-day management plan so they will be able to detect elevated ketones. Children and young people with type 1 diabetes and their family members or carers (as appropriate) may have the necessary ketone testing strips with them when they present. When DKA is suspected in a child or young person with known diabetes and it is not possible to measure the blood ketones (beta-hydroxybutyrate) using a near-patient method, the child or young person should be sent immediately to a hospital with acute paediatric facilities for further investigation.

A further recommendation emphasised that if DKA is suspected or confirmed in a child or young person, health professionals should explain to them and to their family members or carers (as appropriate) that DKA is a serious matter that needs urgent hospital assessment.

Mortality

One RCT (total 33 participants) reported no events in either group (blood ketone testing or urine ketone testing) during treatment of DKA. The evidence for this finding was of high quality.

Duration of treatment

One RCT (total 33 participants) reported that children and young people who received blood ketone monitoring during treatment of DKA had a quicker resolution of ketosis compared with those who underwent urine ketone monitoring. The evidence for this finding was of high quality.

One case series (total 28 participants) reported that the time to resolution of ketosis was quicker when blood ketones were monitored than when urine ketones were monitored. The evidence for this finding was of low quality.

One case series (total 99 participants) reported that the time to resolution of ketosis did not differ according to whether blood ketones or urine ketones were monitored. The evidence for this finding was of very low quality.

18.2.2.7.1. Relative value placed on the outcomes considered

The guideline development group considered which clinical factors should be recognised and which investigations should be recommended in children and young people with DKA at the time of presentation and subsequently during treatment. These would be important in determining treatment required at the outset and in guiding subsequent management. For these review questions the group set up the review protocols to focus predominantly on comparative studies as these would best inform clinical practice. Many different strategies for clinical assessment and monitoring could be considered and an appropriate strategy for laboratory and other investigations could both guide treatment and identify complications.

Particular concern related to the development of cerebral oedema and hypokalaemia, both of which are potentially life-threatening complications of DKA.

18.2.2.7.2. Consideration of clinical benefits and harms

The guideline development group agreed that close monitoring is essential for all children and young people presenting with DKA. They made recommendations regarding the observations and clinical assessment that should be performed both to determine the appropriate treatment strategy and subsequently during treatment. These were consistent with current practice and with existing guidance (for example guidance from the British Society for Paediatric Endocrinology and Diabetes [BSPED]). The group was especially concerned that signs of cerebral oedema should not be overlooked and they made a recommendation on the clinical findings that should be recognised as possible early signs of cerebral oedema, as well as signs that should be assumed to indicate cerebral oedema and require urgent specialist advice and treatment (see Section 18.6.1).

With regard to investigations, the guideline development group recommended that at presentation with suspected or known DKA, the following should be measured: capillary blood glucose, urea, electrolytes and levels of bicarbonate, blood gas and beta-hydroxybutyrate. The group noted that DKA should be diagnosed in children and young people with diabetes who have acidosis (indicated by blood pH below 7.3 or plasma bicarbonate below 18 mmol/litre) and either ketonaemia (indicated by blood beta-hydroxybutyrate above 3 mmol/litre) or ketonuria (++ and above on the standard strip marking scale). The group considered that it is not essential that the child or young person should also have hyperglycaemia because DKA can occur with normal blood glucose levels in those using insulin therapy. Moreover, a blood pH below 7.1 was considered indicative of severe DKA.

The group made recommendations regarding the frequency of repeat measurements following commencement of treatment and they recommended that there should be continuous ECG monitoring to provide evidence of developing hypokalaemia. Recognition of resolution of ketosis and acidosis was important. The evidence indicated that blood monitoring of beta-hydroxybutyric acid was preferable to urine monitoring because the latter persisted after resolution of blood ketosis. Monitoring of urea and electrolyte were essential to patient safety and were needed to guide the intravenous fluid regimen.

18.2.2.7.3. Consideration of health benefits and resource use

A single US study (Vanelli 2003) identified in the global search for economic evidence suggested that blood ketone testing reduced the time spent in intensive care by 6.5 hours compared with urine ketone testing, giving a total cost saving of €184 per patient. Another study demonstrated that blood ketone testing resulted in an earlier resolution of DKA (Noyes 2007). Therefore, the guideline development group considered that blood ketone testing was likely to be cost effective as the higher cost of blood ketone testing would be more than offset by a reduction in hospital costs associated with a quicker recovery time. This led the group to recommend that the use of a near-patient blood ketone (beta-hydroxybutyrate) testing method should be considered for rapid diagnosis and monitoring of DKA in children and young people in hospital. A strong recommendation for universal implementation of this method could not be justified on the basis of available evidence.

18.2.2.7.4. Quality of evidence

The studies identified for inclusion in the guideline review all focused on near-patient ketone monitoring. However, the guideline development group recognised that it was essential to make recommendations on the key aspects of clinical monitoring and investigation. These recommendations were therefore based on the group's clinical expertise and were consistent with current clinical practice and in keeping with existing guidelines.

18.2.2.7.5. Other considerations

There were no other considerations.

18.4.1.7.1. Relative value placed on the outcomes considered

The outcomes prioritised by the guideline development group for this question reflected the serious nature of potential outcomes of DKA, including mortality and incidence of cerebral oedema.

18.4.1.7.2. Consideration of clinical benefits and harms

The guideline development group agreed that as a general principle oral fluid administration is preferable to intravenous fluids in children and young people if this is tolerated. The group recognised that some children and young people presenting with DKA would not yet have become seriously ill. In such cases the child or young person would appear alert and well. Provided children and young people appeared alert, did not have nausea or vomiting (which would impair their ability to tolerate oral fluids) and were no more than minimally dehydrated then intravenous fluid administration might not be required and insulin could be given subcutaneously rather than intravenously. This was in keeping with clinical practice and clinical experience showed this to be safe and effective. However, those who were seriously ill at presentation, with severe nausea, vomiting, hyperventilation or evidence of poor peripheral perfusion, would require intravenous fluids for rehydration.

18.4.1.7.3. Consideration of health benefits and resource use

The guideline development group accepted that for some children and young people with DKA, oral fluid administration – a cheaper method of fluid administration – would be appropriate and safe. However, seriously ill children and young people would require intravenous fluids, justifying the relatively small increased administration costs.

18.4.1.7.4. Quality of evidence

No evidence was identified for inclusion for this review question, but the guideline development group did not regard this topic as a priority for future research.

18.4.1.7.5. Other considerations

Although intra-osseous fluid administration was included in the protocol for this question, no relevant evidence was identified. The guideline development group recognised that this was an established technique in clinical practice in occasional cases where intravenous access proves impossible. They did not therefore make a specific recommendation about the use of intra-osseous fluids.

18.4.1.7.6. Key conclusions

Owing to the lack of evidence identified for inclusion the guideline development group's recommendations were based on their consensus view of good clinical practice.

Cerebral oedema

One study (total 212 participants) found evidence of an increased risk of cerebral oedema for the second tertile of total fluids administered in the first 4 hours of treatment versus the referent tertile. The quality of evidence for this outcome was moderate. The same study found no evidence for an increased risk of cerebral oedema for volume of fluids of between the third tertile of total fluids administered in the first 4 hours of treatment versus the referent tertile. The quality of evidence for this outcome was low. This study identified an overall trend of increasing risk of cerebral oedema with increasing tertile of fluid administration.

One study (total 63 participants) found evidence of an increased risk of cerebral oedema per 1 ml/kg/hour increase in fluids. The quality of the evidence for this outcome was low. Another study (total 244 participants) found no evidence for an increased risk of cerebral oedema with a 5 ml/kg/hour increase in rate of fluid rehydration. The quality of the evidence for this outcome was low.

One study (total 195 participants) found a difference in fluid administration rates within the first 4 hours of treatment in children and young people with brain herniation compared with those without brain herniation. The quality of the evidence for this outcome was very low.

One study (total 36 participants) found no difference in the risk of brain swelling on MRI scan in children and young people who received a slower rate of fluid therapy compared with those who received a faster rate of fluid therapy. The quality of the evidence for this outcome was very low.

Time to resolution of acidosis

One study (total 90 participants) found a reduction in the time to resolution of acidosis in children and young people who received a slower rate of fluid therapy compared with those who received a faster rate of fluid therapy. The quality of the evidence for this outcome was very low.

Change in serum sodium

One study (total 90 participants) found no difference in the change in serum sodium concentration in children and young people who received a slower rate of fluid therapy compared with those who received a faster rate of fluid therapy. The quality of the evidence for this outcome was very low.

Change in serum chloride

One study (total 90 participants) found no difference in the change in serum chloride concentration in children and young people who received a slower rate of fluid therapy compared with those who received a faster rate of fluid therapy. The quality of the evidence for this outcome was very low.

Admission to ICU

One study (total 90 participants) found no difference in the change in rates of admission to ICU in children and young people who received a slower rate of fluid therapy compared with those who received a faster rate of fluid therapy. The quality of the evidence for this outcome was very low.

18.4.2.7.1. Relative value placed on the outcomes considered

The guideline development group identified mortality and incidence of cerebral oedema as the most important outcomes for this question (because the serious nature of DKA carries a risk of severe adverse consequences). The group also regarded the following as important outcomes (because quicker recovery or return to normal biochemical values would be beneficial): time to resolution of dehydration; rate of change in blood glucose concentration; resolution of acidosis; serum chloride concentration; and serum sodium concentration. Healthcare utilisation was also considered as a proxy for severity of illness.

18.4.2.7.2. Consideration of clinical benefits and harms

The guideline development group was cognisant of the fact that in children and young people the severity of fluid deficit is difficult to determine accurately. Studies have shown that agreement between clinical symptoms and signs and actual percentage dehydration is poor. The NICE clinical guideline on diarrhoea and vomiting in children under 5 looked at this in some detail and concluded that it was possible only to distinguish between those with no clinical evidence of dehydration, those with some evidence and those with very pronounced findings suggesting imminent or actual hypovolaemic shock. The clinical manifestations of dehydration may be even more difficult to interpret in DKA. For example, urine output is maintained even in severe dehydration because of osmotic diuresis, the oral mucous membranes may tend to be dry due to hyperventilation associated with acidosis rather than as a result of dehydration, and acidosis leads to peripheral vasoconstriction resulting in prolonged capillary refill time which is normally considered to indicate the presence of severe dehydration. In principle, if a recent weight record is available for comparison then a child or young person's weight on presentation can be used to estimate the degree of dehydration (based on percentage weight loss). This is much less reliable, however, in children and young people presenting with DKA, because in that setting tissue-wasting as a consequence of poorly controlled diabetes may have contributed to any observed weight loss. Insulin deficiency has a direct tissue catabolic effect. This factor may be more significant in those who are not known to have diabetes prior to developing DKA because they may have experienced insulin deficiency for a longer period of time prior to presentation. In this context the guideline development group noted that a study considered in the evidence reviews for assessment, monitoring and investigation at presentation with DKA and during management of DKA (Koves 2004) relied on weight to determine the severity of dehydration and may therefore have overestimated dehydration. Reliance on weight loss to calculate the percentage of dehydration will always overestimate fluid loss in children and young people with DKA because these children and young people will have lost body weight associated with catabolism due to insulin deficiency.

The guideline on diarrhoea and vomiting in children under 5 recommended that in children with possible or suspected dehydration due to diarrhoea and vomiting the degree of dehydration should, therefore, be assessed as being 0%, 5% or 10% dehydrated as a pragmatic approach to fluid management. Using these estimates as a starting point rehydration would be given and the response to the fluid replacement given would be kept under clinical review. This was considered a safe approach in that even if this sometimes overestimated the required fluid volume, giving more fluid than is essential was not associated with significant risks and reduced the risk of under-hydration and delayed recovery. The volume of fluid replacement would be increased only if subsequent clinical assessment showed persistent symptoms or signs of dehydration. The guideline development group for this guideline considered this, but concluded that a somewhat different approach was required in those presenting with dehydration due to DKA. Existing guidelines in the UK (BSPED) at the time this guideline was being developed advised that based on the clinical assessment, children and young people should be categorised as having mild (equivalent to 3%) dehydration, moderate (5%) dehydration or severe (8%) dehydration. The group favoured coming into line with other guidance, recognising that clinical assessment of dehydration is not possible. Although there were concerns associated with rapid or excessive administration of fluid during DKA, the increase in replacement fluid is likely to be offset by the recommended reduction in maintenance fluid.

In simple dehydration (for example that associated with gastroenteritis), the principle is to attempt quite rapid rehydration to expedite recovery. The guideline development group concluded that the evidence suggests that when this approach is taken in children and young people with DKA, there is an increased risk of cerebral oedema. Moreover, these children and young people will usually require intravenous therapy for a significant period of time because the acidosis resolves gradually with insulin therapy. The group also recommended avoidance of fluid bolus administration unless there are signs of shock associated with poor urine output or hypotension. If a bolus is to be given, the group recommended 10 ml/kg rather than 20 ml/kg for the same reason. In keeping with current international practice, the group believes that aiming to give intravenous fluid replacement of the deficit evenly over 48 hours is an appropriate approach.

The guideline development group recommended using a pH of less than 7.1 as a threshold for treating the child or young person as if there is 10% dehydration because this is a marker for severe DKA. Severity of DKA is predominantly defined by the level of acidosis. A pH of less than 7.1 is generally recognised as a threshold for severe DKA and is likely to be associated with more severe dehydration.

18.4.2.7.3. Consideration of health benefits and resource use

The recommendations made by the guideline development group are largely in line with current practice and will have minimal cost impact in the context of the overall management of the condition. This practice has evolved over time and clinical experience indicates that it is safe and effective.

18.4.2.7.4. Quality of evidence

The guideline development group noted a lack of evidence related to several of their prioritised outcomes, namely mortality, time to resolution of dehydration and rate of change in blood glucose concentration. Also, the evidence that was identified for inclusion in the guideline review was of very low or low quality.

18.4.2.7.5. Other considerations

The guideline development group was aware that a clinical guideline on intravenous fluids therapy in children was being developed for NICE contemporaneously with the development of this guideline. The group acknowledged that recommendations in this guideline related to the rate of fluid administration for rehydration in children and young people with DKA should be distinguished from recommendations related to other indications for fluid therapy. For this reason, and noting that incorrect rates of rehydration in children and young people with DKA could have serious adverse consequences, the guideline development group was careful to specify in the guideline recommendations the population of children and young people to which they referred.

18.4.2.7.6. Key conclusions

The guideline development group's recommendations took account of the evidence and their clinical expertise and experience. The group recommended the following assumptions related to the fluid deficit (degree of dehydration) in children and young people with DKA:

• assume a 5% fluid deficit in mild to moderate DKA (indicated by a blood pH of 7.1 or above)
• assume a 10% fluid deficit in severe DKA (indicated by a blood pH below 7.1).

Sodium

Bicarbonate

Bicarbonate

One study (total 244 participants) found a statistically significant increase in the risk of cerebral oedema in children and young people with DKA who received treatment with bicarbonate compared with those who did not receive bicarbonate. The quality for the evidence was moderate.

Phosphate

18.4.3.7.1. Relative value placed on the outcomes considered

The most important outcomes prioritised for this review question were mortality and incidence of cerebral oedema (because DKA and associated conditions such as cerebral oedema can be life-threatening). Time to resolution of dehydration, rate of change of blood glucose concentration, incidence of hypoglycaemia, resolution of acidosis and resolution of blood ketosis, serum chloride concentration, hypokalaemia, serum sodium concentration, serum calcium concentration and carbon dioxide concentration were also considered important (because a faster recovery and normalisation of biochemical parameters is beneficial). Healthcare utilisation was also selected as an outcome for consideration (as a proxy for severity of DKA or presence of cerebral oedema).

18.4.3.7.2. Consideration of clinical benefits and harms

The guideline development group considered that there is little published evidence regarding the use of 0.9% saline (sodium chloride) or other intravenous solutions in DKA. The International Society for Pediatric and Adolescent Diabetes (ISPAD) guidance recommends giving 0.9% saline for the first 4 to 6 hours of treatment. The British Society for Paediatric Endocrinology and Diabetes (BSPED) guidance recommends giving 0.9% saline for the first 12 hours of treatment. The guideline development group believed that risk of cerebral oedema was greatest during the first 12 hours of treatment which is why this timeframe had been chosen. The group was aware, however, of published studies showing an association between falling sodium concentrations (corrected for glucose) and the risk of cerebral oedema (Durward 2011; Fiordalisi 2007). The recommendation to use 0.9% saline for both rehydration and maintenance fluid in children and young people with DKA is consistent with guidance issued by the National Patient Safety Agency (NPSA).

Children and young people may present with normal or even elevated plasma potassium concentrations but in DKA there is always a significant total body potassium deficit. Furthermore, once insulin is administered potassium concentrations in the blood fall precipitously. The guideline development group therefore recommended that potassium should be added to intravenous fluid infusion at the start of treatment. Hypokalaemia is an important cause of death in children and young people with DKA.

It is known that hypophosphataemia occurs in children and young people presenting with DKA. However, clinical experience and current practice indicate that intravenous phosphate administration is not necessary. The guideline development group was aware of the possibility that phosphate administration could induce hypocalcaemia. For these reasons the group did not make a recommendation to add phosphate to the intravenous fluid regimen. This was in keeping with existing guidelines such as those from the BSPED.

The guideline development group recommended that bicarbonate should not normally be given as part of the treatment for DKA. There was evidence that bicarbonate administration is associated with an increased risk of cerebral oedema. Moreover, in DKA the acidosis is primarily due to ketoacidosis and it responds to treatment with insulin.

The guideline development group recommended adding 5% glucose to the intravenous fluid infusion once blood glucose falls below 14 mmol/litre. This level was chosen based on established international practice. In the event that blood glucose falls below 6 mmol/litre the group recommended increasing the glucose concentration of intravenous fluid infusions above 5% to prevent hypoglycaemia.

18.4.3.7.3. Consideration of health benefits and resource use

DKA and its consequences, including the possibility of death, provides an overriding rationale for optimising fluid management, and thus resource use is considered to be appropriate if such serious adverse outcomes can be prevented.

18.4.3.7.4. Quality of evidence

Evidence was lacking for most outcomes except bicarbonate and the evidence that was identified for inclusion was graded mainly as very low quality. This did not, however, prevent the guideline development group from formulating recommendations based on the available evidence for bicarbonate along with their clinical experience and practice relevant to the other fluids and additives considered in the review protocol (sodium, glucose, potassium and phosphate).

18.4.3.7.5. Other considerations

There were no other considerations.

18.4.3.7.6. Key conclusions

The guideline development group's recommendations took account of the evidence and their clinical expertise and experience. The group recommended the use of 0.9% sodium chloride without added glucose for both rehydration and maintenance fluid in children and young people with DKA until the blood glucose concentration is below 14 mmol/litre.

The group recommended that healthcare professionals ensure that all fluids (except any initial bolus) administered to children and young people with DKA contain 40 mmol/litre potassium chloride, unless they have renal failure The group further recommended changing fluids to 0.9% sodium chloride with 5% glucose and 40 mmol/litre potassium chloride once the blood glucose concentration falls below 14 mmol/litre.

The group recommended that if, during treatment for DKA, a child or young person's blood glucose falls below 6 mmol/litre the glucose concentration of the intravenous fluid infusion should be increased.

The group recommended that intravenous sodium bicarbonate should not be given to children and young people with DKA.

18.4.4.1.1. Review question

When should intravenous insulin therapy be started and stopped in children and young people with diabetic ketoacidosis?

18.4.4.1.2. Introduction

The purpose of this review question is to establish when intravenous insulin therapy should be started and stopped in children and young people with DKA. The review allowed the inclusion of observational studies in addition to RCTs and systematic reviews.

The guideline development group identified several key outcomes for this review:

• mortality
• rate of change of blood glucose
• incidence of hypoglycaemia
• incidence of cerebral oedema
• resolution of acidosis
• hypokalaemia
• resolution of blood ketosis
• healthcare utilisation.

For the part of the question regarding when to start intravenous insulin therapy the intervention was delayed insulin and the comparator immediate insulin. For the part of the question about stopping intravenous insulin outcomes were to be compared according to the blood ketone concentration at which insulin was stopped. Subgroup analyses by type of diabetes and age group were to be undertaken where possible.

The only study identified for inclusion for this review question (Edge 2006) included the Chair of the DKA subgroup as the primary author. To avoid any conflict of interest in the recommendations arising from the review question the NCC-WCH Clinical Director for Children's Health chaired the discussions. Although the Chair of the DKA subgroup participated in discussion of the evidence and formulation of the recommendations, she did not have a casting vote on the agreement of recommendations.

18.4.4.1.3. Description of included studies

A single study was identified for inclusion in this review (Edge 2006). This was a case-control study which addressed the timing at which to start insulin therapy in children and young people with type 1 diabetes and DKA. The study included 43 cases and 169 controls from England, Scotland and Wales. The mean ages of the participants were 8.5 years and 8.9 years for cases and controls, respectively.

Of the guideline development group priority outcomes, only risk of cerebral oedema was reported. Cases were defined as having a diagnosis of DKA-related cerebral oedema. Controls were defined as having a diagnosis of DKA without cerebral oedema. Controls were matched to cases based on age, sex, whether or not diabetes was newly diagnosed and the month of admission. Adjustments were also made for baseline biochemical measures to take account of the severity of acidosis. No subgroup analyses were possible.

No evidence was identified for: mortality, rate of change of blood glucose, hypoglycaemia incidence, resolution of acidosis, hypokalaemia or healthcare utilisation. No evidence was identified which addressed the time at which insulin therapy should be discontinued in children and young people with DKA.

18.4.4.1.4. Evidence profile

The evidence profile for this review question (timing of insulin therapy in children and young people with DKA) is presented in Table 75.

Table 75

Evidence profile for the effect of insulin administered within 1 hour of fluid replacement compared with insulin administered at least 1 hour after fluid administration on the risk of cerebral oedema.

18.4.4.1.5. Evidence statements

18.4.4.1.6. Health economics profile

A systematic literature search did not identify any relevant published economic evidence relating to starting and stopping intravenous insulin therapy.

This question was not prioritised for health economic analysis as the timing of starting or stopping intravenous insulin therapy does not have resource implications.

18.4.4.1.7. Evidence to recommendations

18.4.4.2.1. Review question

How should the dosage of insulin be calculated for children and young people with diabetic ketoacidosis (DKA)?

18.4.4.2.2. Introduction

The objective of this review question is to determine the most appropriate dose of insulin to treat DKA in children and young people with either type 1 diabetes or type 2 diabetes. Specifically, the question addresses whether a low dosage of insulin (0.025 U/kg/hour or 0.05 U/kg/hour) may result in fewer adverse outcomes compared with the current ‘standard’ dosage of 0.1 U/kg/hour.

The guideline development group identified priority outcomes as being:

• mortality
• rate of change of blood glucose
• incidence of hypoglycaemia
• resolution of acidosis
• resolution of blood ketosis
• incidence of cerebral oedema
• hypokalaemia
• healthcare utilisation.

Subgroup analyses were to be undertaken for type 1 and type 2 diabetes and by age group where possible. The review includes observational studies because no RCTs met the inclusion criteria.

18.4.4.2.3. Description of included studies

Three retrospective cohort studies were identified for inclusion in this review (Al Hanshi 2011; Kapellen 2012; Puttha 2010). The studies were carried out in Australia, Germany and the UK, respectively. All 3 studies assessed DKA in children and young people with type 1 diabetes. As no studies addressing DKA associated with type 2 diabetes were identified for inclusion in the review no subgroup analyses by diabetes type were possible.

The number of participants ranged from 64 to 93. The age range of participants was 1.25 to 17.7 years across the 3 studies as a whole. Two of the studies included participants with narrower age ranges of approximately 6 years (Kapellen 2012; Puttha 2010). One study compared the standard dosage of insulin with a lower dose of 0.025 U/kg/hour (Kapellen 2012). The other studies compared the standard dosage of insulin with a lower dose of 0.05 U/kg/hour (Al Hanshi 2011; Puttha 2010).

Data were reported for 4 of the guideline development group's priority outcomes: change in blood glucose from admission, change in blood pH from admission, incidence of hypoglycaemia and incidence of hypokalaemia. One study reported results for children under the age of 5 years in a subgroup analysis (Puttha 2010). Four further outcomes were not reported in sufficient detail to be included in GRADE profiles:

• 1 study reported a case of cerebral oedema using a case-report approach (Kapellen 2012)
• 1 study reported time to normalise acidosis, but no confidence intervals or p-values for between-group differences were reported (Kapellen 2012)
• 1 study reported duration of hospital stay, but it was not clear whether the reported values were means or medians and confidence intervals were not reported (Puttha 2010)
• 1 study reported the time to normalise blood glucose, but no confidence intervals were provided and it was not clear whether the values were means or medians (Kapellen 2012).

No studies reported results for mortality or resolution of ketosis.

18.4.4.2.4. Evidence profile

The evidence profile for this review question (dosage of intravenous insulin) is presented in Table 76.

Table 76

Comparison of insulin dosage of 0.025 U/kg/hour or 0.05 U/kg/hour with a dosage of 0.1 U/kg/hour in children and young people with type 1 diabetes and DKA.

18.4.4.2.5. Evidence statements

18.4.4.2.6. Health economics profile

A systematic literature search did not identify any relevant published economic evidence related to the dosage of insulin for children and young people with DKA.

This question was not prioritised for health economic analysis as the guideline development group considered there were more important priorities for health economic analysis.

18.4.4.2.7. Evidence to recommendations

Mortality

One study (total number of participants is not calculable) found that use of hypertonic saline alone was associated with higher mortality than use of mannitol alone, adjusted for discharge year, hospital clustering, gender, predictors of severity (mechanical ventilation, brain imaging with CT scan and ICD-9 code [250.2 and 250.3]) after non-significant predictors of mortality (age, race and ICU admission) were sequentially removed. Participants treated with both agents were excluded from odds ratio (OR) analysis in the study. The quality of the evidence for this outcome was very low.

Healthcare utilisation

The study (total 1501 participants) was not able to determine whether healthcare utilisation (which the study authors expressed as severity of illness) was a consequence of the treatment as it could not ascertain what clinical criteria were used to warrant treatment. The study found more children and young people treated with hypertonic saline alone were admitted to ICU than those treated with mannitol alone. The same study found a few more children and young people treated with mannitol alone had mechanical ventilation and brain imaging with CT scan than hypertonic saline alone. The quality of the evidence for these outcomes was very low.

18.6.1.7.1. Relative value placed on the outcomes considered

The guideline development group specified mortality as the highest-priority outcome and this was reported in the included study. Long-term neurological problems were also selected as a priority outcome but these were not reported in the included study. However, the absence of evidence for this outcome did not prevent the group making recommendations.

18.6.1.7.2. Consideration of clinical benefits and harms

The guideline development group noted that cerebral oedema is potentially life threatening and that prompt action to treat cerebral oedema is essential once the condition is detected. The guideline development group was aware that mannitol has been the standard treatment for cerebral oedema but that increasingly hypertonic saline (sodium chloride) is being used in intensive care units (ICUs) for treatment of non-DKA related cerebral oedema and in some settings hypertonic saline is now recommended as first-line treatment.

A possible benefit of hypertonic saline over mannitol is that it can be given repeatedly with persisting benefit whereas mannitol becomes less effective with repeated administration.

The group also noted that mannitol is more readily available on paediatric wards (although this does not hold true for ICUs).

18.6.1.7.3. Consideration of health benefits and resource use

The guideline development group emphasised that effective treatment of cerebral oedema will save lives and noted that it may improve neurological outcomes. The treatment options considered by the group – mannitol and hypertonic saline – were both noted to be low cost, and the group noted that the intention was not to recommend either mannitol or hypertonic saline in preference to the other, but to recommend the use of whichever of the 2 treatments would be readily already available, and therefore it was expected that there would be no uplift in cost associated with the recommendation.

Illustrative costs for mannitol and hypertonic saline are illustrated below, based on the unit costs reported in Table 78. They suggest that for the pack sizes listed the costs would be similar.

Table 78

Unit costs of osmotic agents.

Following the upper limit of the guideline recommendations a girl aged 12 years weighing 40 kg would require:

• Mannitol 20% – 40 g or 200 ml which could be supplied from 1 bag of 250 ml solution
• Hypertonic saline 2.7% – 200 ml which could be supplied from 1 bag of solution.

18.6.1.7.4. Quality of evidence

The guideline development group noted that the evidence for their priority outcome of healthcare utilisation was slightly indirect in that the study authors reported severity of illness rather than a direct measure of healthcare resource use. The group also noted that a record of administering mannitol or hypertonic saline would provide a reasonable marker for presence of cerebral oedema, although this could not be ascertained with certainty.

The group noted that the evidence was seriously indirect due to the methods used in the study. In particular, the group felt that there was strong risk of bias because hypertonic saline was reported to be used more frequently in ICU settings and it was plausible that the relevant participants would have been more unwell than the other participants. This meant that no conclusions could be drawn in terms of comparing the effectiveness of mannitol and hypertonic saline.

Where both treatments (mannitol and hypertonic saline) were used it was impossible to tell which was used first and again the guideline development group thought it was likely that participants who received both treatments would have been more unwell than the other participants. Therefore nothing could be concluded about the potential benefit of using both treatments sequentially, nor the order in which the treatments were used.

18.6.1.7.5. Other considerations

The dosages for mannitol and hypertonic saline recommended by the guideline development group are broadly in keeping with general statements in the summaries of product characteristics (SPCs) for these products and with ISPAD guidance which reflects current practice in the UK. The guideline development group wished to include the dosages in the recommendations so that the information would be to hand when management of cerebral oedema was necessary, rather than healthcare professionals having to check the dosages separately and thus delay potentially life-saving treatment.

18.6.1.7.6. Key conclusions

The guideline development group concluded that cerebral oedema in children and young people with DKA should be treated promptly using mannitol or hypertonic saline (sodium chloride), whichever is most readily available in the non-ICU setting. Specifically, the group recommended that if cerebral oedema is suspected in a child or young person with DKA, they should be treated immediately with the most readily available of mannitol (20%; 0.5 to 1 g/kg over 10 to 15 minutes) or hypertonic sodium cholride (2.7% or 3%; 2.5 to 5 ml/kg over 10 to 15 minutes). The same treatment should be given to any child or young person with DKA who develops any of the following signs: deterioration in level of consciousness; abnormalities of breathing pattern, for example respiratory pauses; oculomotor palsies; pupillary inequality or dilatation.

The group further recommended that after starting treatment for cerebral oedema with mannitol or hypertonic sodium chloride in a child or young person with DKA, specialist advice on further management, including which care setting would be best, should be sought immediately.

18.6.2.7.1. Relative value placed on the outcomes considered

The guideline development group prioritised the following physical outcomes for consideration in this review question:

• mortality
• incidence of venous thrombosis (of any type, including deep vein thrombosis, visceral thrombosis and cerebral thrombosis)
• incidence of pulmonary embolism
• healthcare utilisation (for example duration of admission, admission to intensive care)
• adverse events, including bleeding and thrombocytopaenia.

The group also prioritised the satisfaction of children, young people and families with the intervention as an outcome for consideration.

The group's priorities reflected the serious nature of potential outcomes associated with venous thrombosis and pulmonary embolism, including the possibility of death, and their selection of priority outcomes reflects this alongside the importance of offering treatments that are acceptable to children and young people with DKA and their families.

18.6.2.7.2. Consideration of clinical benefits and harms

The guideline development group recognised that there is a risk of venous thrombosis in children and young people with DKA (this was based on the group's knowledge of relevant case reports). The extent of this risk has, however, not been accurately quantified.

The group's view was that the use of central venous catheters increases the risk of thrombosis (again this is based on the group's knowledge of case reports).

The group noted that the risk of venous thromboembolism (VTE) in children and young people with DKA compared with other children in an intensive therapy unit (ITU) environment is unknown, and there are potential harms associated with anticoagulant prophylaxis and treatment (the main harm being bleeding).

Overall, the guideline development group noted the lack of evidence for this review question and concluded that there was no general consensus regarding the role of anticoagulant prophylaxis for children and young people with DKA, either in terms of benefits or harms.

18.6.2.7.3. Consideration of health benefits and resource use

The guideline development group did not enter into detailed consideration of the cost effectiveness of alternative management strategies based on anticoagulant prophylaxis because they did not wish to recommend this form of treatment due to a lack of evidence related to its clinical effectiveness.

18.6.2.7.4. Quality of evidence

No evidence was identified for inclusion for this review question, but the guideline development group did not view this area as a priority for future research.

18.6.2.7.5. Other considerations

There were no other considerations.

18.6.2.7.6. Key conclusions

The guideline development group concluded that anticoagulant prophylaxis was not to be recommended for children and young people with DKA and they agreed not to make any recommendations on this topic. The group did, however, recommend that healthcare professionals should be aware of the increased risk of venous thromboembolism in children and young people with DKA, especially those with central venous catheters.