THERAPY

Criteria abstracted from The Users' Guides to the Medical Literature series in JAMA

Diabetes Care 2004 27(8):1873-8.[abstract]

Review posted July 06, 2007

1. What is being studied?
   
   
   1. The study objective:
      
      The goal of the study was to compare both the efficacy and safety of subcutaneous aspart insulin to an intravenous infusion of regular insulin in patients with uncomplicated diabetic ketoacidosis (DKA). With the introduction of fast-acting insulin, such as aspart, the administration of subcutaneous insulin during uncomplicated DKA may serve as an alternative to IV therapy. If subcutaneous aspart insulin therapy proves to be comparable to IV regular insulin, the subcutaneous route could provide a more cost-effective means of treatment.
      
      
   2. The study design:
      
      This is a prospective, randomized, open trial, consisting of uncomplicated DKA patients randomly assigned to 1 of 3 treatment groups with 15 patients in each group. The design is representative of an equivalent trial. If the effects of the two treatments differ in either direction by more than the equivalence margin, then equivalence does not hold
      
      
   3. The patients included:
      
      The study population consisted of 45 consecutive adult patients arriving in the emergency department (ED) with DKA. Criteria used to establish the diagnosis of DKA included a plasma glucose level > 250mg/dl, serum bicarbonate level < 15 mmol/l, venous pH < 7.30, and a positive serum ketone level at a dilution of at least 1:4 by the nitroprusside reaction and/or serum beta-hydroxybutyrate level > 3 mmol/l.
      
      
   4. The patients excluded:
      
      The patients who were excluded included anyone with one of the following:
      
      
      • Persistent hypotension after the administration of 1 liter of normal saline
      • Acute myocardial ischemia
      • End-stage renal or hepatic failure
      • Anasarca
      • Dementia
      • Pregnancy
        
        
   5. The interventions compared:
      
      The three therapy groups that were examined are as follows:

      1. SC-1hr Group: Initial dose of subcutaneous insulin at 0.3 units/kg followed by subcutaneous aspart at 0.1 units/kg every hour.
      2. SC-2hr Group: Initial dose of subcutaneous insulin at 0.3 units/kg followed by subcutaneous aspart at 0.2 units/kg every 2 hours.
      3. IV Group: Initial dose of IV insulin at 0.1 units/kg followed by continuous infusion of regular insulin at 0.1 units/kg every hour.
      
      

   6. The outcomes evaluated:
      
      Primary outcome

      Response to medical therapy was evaluated by assessing the time and amount of insulin required for resolution of DKA (bicarbonate level = 18 and venous pH = 7.3) and hyperglycemia (blood glucose levels = 13.8 mmol/l or 250 mg/dl).

      Secondary outcome

      Medical care data will also include the amount of fluid and insulin administered until the resolution of hyperglycemia and DKA, as well as length of hospitalization and the number of hypoglycemic events during therapy.
      
      

2. Are the results of the study valid?

Primary questions:

1. Was the assignment of patients to treatments randomized?
   
   Yes, 45 consecutive patients who met the DKA and study criteria were then randomly assigned in the ED to 1 of the 3 study groups.
   
   
2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?
   
   Was follow-up complete?
   
   Yes. All 45 patients were accounted for. No patients died, there was no recurrence of DKA during the study, and nobody was lost to follow up.
   
   Were patients analyzed in the groups to which they were randomized?
   
   Yes. Once assigned to a specific group, there was no cross over to other groups. Treatment was tolerated well, and all patients finished in the group that they started in.

Secondary questions:

3. Were patients, health workers, and study personnel "blind" to treatment?
   
   Due to hospital policy, patients receiving IV insulin infusion required ICU monitoring, while the subcutaneous groups were treated on the general medicine floor or step-down unit. Therefore, the ability to keep clinicians and patients blind was compromised. However, this should have had a limited effect on the study considering the standardized treatment protocol and the fact that most of the data collected consisted of objective measurements.
   
   
4. Were the groups similar at the start of the trial?
   
   The admission biochemical parameters were not significantly different among treatment groups. The mean age, duration of diabetes, and precipitating cause for DKA were also similar among the 3 groups. Each group had 4 patients with notably comorbidities. The comorbidites differed between the 3 groups (as shown below) with the exception of UTI’s, which were evenly distributed.

	Comorbidities
SC-1hr group	Leg abscess, Pneumonia, UTI, Pancreatitis
SC-2hr group	Cellulitis, UTI, Olanzapine overdose, Failure to take oral antidiabetic agent
IV Insulin group	Pneumonia, Cellulitis, UTI, Tooth abscess

5. Aside from the experimental intervention, were the groups treated equally?
   
   The most obvious difference would be ICU status for the IV group and floor status for the others. This did not impact treatment since treatment was based on protocol and objective lab values. All the serum labs were drawn at the same intervals with fluid and electrolyte replacement based on preset parameters. However, there were some slight differences.

• Fingersticks every 1 hr in SC-1 and every 2 hr in SC-2 & IV

• This could potentially allow for earlier detection of hyperglycemia resolution in the SC-1 group
  
  

• Initially SC received 0.3 units/kg and IV received 0.1 units/kg as a bolus upon admission

3. What were the results?
   
   
   1. How large was the treatment effect?
      
      The primary outcomes associated with the three treatment groups were compared using a difference of greater than 4 hours as the indicator of clinical significance (equivalent margin). It should be noted that the authors arbitrarily chose the 4-hour mark.
      
      Analysis showed no statistical difference between the 3 study groups for both the primary and secondary objectives. Each group had one episode of hypoglycemia.

	SC-1hr group (mean ± SD)	SC-2hr group (mean ± SD)	IV group (mean ± SD)
Resolution of hyperglycemia (hr)	6.9 ± 4	6.1 ± 4	7.1 ± 5
Resolution of ketoacidosis (hr)	10 ± 3	10.7 ± 4	11 ± 3

When data reveals no statistical difference, the results should be interpreted with caution. A well-executed clinical trial that correctly demonstrates the treatments to be similar can not be distinguished, on the basis of the data alone, from a poorly executed trial that fails to find a true difference. The small sample size and biases from non-blinded investigators could possibly affect the study results. Researchers must be cognizant of these issues and account for them in either the design or discussion.

For this particular research question, it would be more appropriate to use a non-inferiority trial approach. The investigators should have set a priori what would be the acceptable level of uncertainty or non-inferiority margin (i.e. 95% confidence interval of the difference in the duration of DKA resolution). However, when determining no statistical difference, the 95% CI must not be too wide. A wide CI may prove problematic when trying to conclude that the SC insulin is not inferior to the IV insulin.

If we assume the data is normally distributed, the 95% CI = mean ± 1.96 × SD. Given the SD of ± 3, the 95% CI = mean ± 1.96 × 3.This results in a 95% CI of the mean ± 5.88. Therefore, there is a 95% chance that the time of DKA resolution would range from 10 ± 5.88 in the SC-1 hr group and from 11 ± 5.88 in the IV group. Thus, there is the possibility that the difference in time to resolution could be as much as 10.76 hr (15.88-5.12) in the SC group. This difference is too large to justify that SC insulin is not inferior to IV insulin.

2. How precise was the estimate of the treatment effect?
   
   The estimate was not precise. The wide 95% CI of (-10.76 , 12.76) for time to resolution of ketoacidosis is actually greater than the arbitrarily duration of 4 hours judged by the investigators as the clinically significant duration. Based on this imprecision, we cannot justify whether SC insulin therapy is superior or inferior to the IV insulin therapy for the patients with DKA.
   
   This imprecision was the consequence of assuming a large effect size (f = 0.5, using SD of 3) for sample size calculation. For a non-inferiority study, the small effect size (f = 0.1-0.25) should have been used. The ANOVA for this 3-group comparison will require at least 80 subjects per group to achieve the same power.

4. Will the results help me in caring for my patients?
   
   
   1. Can the results be applied to my patient care?
      
      The biggest difference between the study population and our patient population is age. The mean ages of the patients in the study were 36, 38, and 40 years. However, the data should still be applicable as shown by a similar study that looked at a pediatric population (1). It consisted of 64 DKA patients with an average age of 11.3 yrs for the subcutaneous group and 12.1 yrs for the IV insulin group. While the overall results were consistent between the two studies, the pediatric study differed from the adult study both in its protocol and primary outcomes.
      
      Another difference is the presence of comorbidities in the older population, which may not play as large of a role in pediatrics. Since the presence of comorbidities correlated more with increasing the length of hospital stay, it is unlikely that this difference would deter implementation in our patient population. Since the authors excluded DKA patients complicated by hypovolemic shock, comatose state, MI, CHF, and renal or hepatic failure, we can only apply this study to our patients with uncomplicated DKA.
      
      
   2. Were all clinically important outcomes considered?
      
      The outcomes that related to the resolution of DKA and hyperglycemia were well covered. However, there was no subjective component to this study. The patients’ satisfaction with the different treatment methods was not evaluated, but may be an important factor. Some patients, especially children, may have strong objections to receiving a subcutaneous shot every hour or two as opposed to the placement of one IV line.
      
      
   3. Are the likely treatment benefits worth the potential harms and costs?
      
      Although, the international consensus statement on DKA in children and adolescents (2) indicates that low-dose intravenous insulin administration should be the standard of care, the alternative subcutaneous insulin apart offers a simplified and cost-effective option that is worth exploring. Since there was no significant statistical difference found between the three groups, the potential harm of the subcutaneous group is not in its efficacy. The authors concluded that subcutaneous treatment is a safe and effective alternative to IV insulin therapy. Given the uncertainty of the results described, we suggest the study be replicated with larger sample size. Another important question is whether or not the benefit of being more cost effective and freeing up ICU beds negates the possible discomfort of multiple subcutaneous injections. The decision to use it is best made in light of the severity of the DKA, the patient’s values, and the availability of appropriate supervision, nursing staff and medical resources.

References

1. Della MT, Steinmetz L, Campos PR. Subcutaneous use of a fast-acting insulin analog: an alternative treatment for pediatric patients with diabetic ketoacidosis. Diabetes Care. 2005;28:1856-61.
   
   
2. Dunger DB, Sperling MA, Acerini CL, Bohn DJ, Daneman D, Danne TPA, Glaser NS, Hanas R, Hintz RL, Levitsky LL, Savage MO, Tasker RC, Wolsdorf JI, European Society for Pediatric Endocrinology, Lawson Wilkins Pediatric Endocrine Society: European Society for Pediatric Endocrinology/Lawson Wilkins Pediatric Endocrine Society consensus statement on diabetic ketoacidosis in children and adolescents. Pediatrics 113:e133– e140, 2004
   
   
3. Snapinn, Steven M, Merck Research Laboratories, West Point, Pennsylvania, USA, Commentary on Noninferiority Trials. Curr Control Trials Cardiovasc Med 2000, 1:19-21.

Last Updated: July 06, 2007

Fill out our Feedback Form | Editorial Notes | About Us