Parsimonious Practice: How to Treat DKA with 5 Lines of Orders in Half a Day

Few processes in medicine are as simple as the treatment of DKA (diabetic ketoacidosis) or have been as gratuitously complicated by anal retentive micromanagers of physiology.  Here is a departure from that custom that I have refined iteratively during the past four years.  It is guided by the goal of reversing ketoacidosis (and associated dehydration) and getting patients eating and back on subcutaneous insulin as expeditiously as possible, while reducing waste and burdens of care and without compromising safety.  It does not have as a goal to rigidly govern lab values or usurp control of physiology during the process.  The caveat to be aware of is that I have refined it in young(ish), adult, non-compliant Type I diabetics without insulin resistance who have moderate to severe acidosis and hyperglycemia.  (I do not treat "DKA" with a serum HCO3- greater than or equal to 14 with an insulin infusion - I treat it with fluids and reinstitution of subcutaneous insulin.)  It also presumes that there was no trigger other than non-compliance, or that the trigger (e.g. UTI) has been investigated and addressed.  I will briefly discuss the 5 orders, their benefits and potential drawbacks.  First, the orders:

1.)  Bolus with 5 liters of Lactated Ringers Solution

2.)  Begin insulin infusion at 5 units per hour (FIXED DOSE, NO BOLUS, NO TITRATION).

3.)  Check blood sugar every 2 hours; When blood sugar less than 200 mg/dL, reduce insulin drip to 1 unit per hour.

4.)  If blood sugar is less than 100 mg/dL (on any insulin dose) or greater than 300 mg/dL on 1 unit per hour, call MD.

5.)  Check serum K+ and HCO3- 12 hours after the start of treatment.

Elaboration:

1.)  Bolus with 5 liters of Lactated Ringers (LR) Solution

The average fluid deficit in DKA is said to be 5 liters (reference:  Harrison's Principles of Internal Medicine, 13th edition; years of experience.)  Moreover, dehydration in my estimation contributes a lot  to the symptoms of DKA and indeed I have come to believe that getting dehydrated by itself, with no noncompliance and no other perturbations can tip a diabetic into DKA.   I have found that rather than drips and titrations and guesswork about the actual fluid deficit, 5 liters works pretty good the vast majority of the time.  It should be infused rapidly, 1-2 liters per hour.  (Part of the initial symptomatic response to DKA treatment comes from the osmotic diuresis [volume restored, hyperglycemia still present] which causes ketones to be excreted in the urine.)  I don't use Normal Saline (NS) because of emerging (but not conclusive) evidence that high chloride containing solutions are less benign than resuscitation solutions with less chloride, and because use of NS frequently results in a hyperchloremic metabolic acidosis which prematurely closes the anion gap in the treatment of DKA.  This is also why I follow the serum bicarbonate rather than the anion gap as a gauge of response to treatment.

2.)  Insulin infusion at 5 units per hour (FIXED DOSE, NO BOLUS, NO TITRATION).

Boluses don't make sense with a drug with a short half-life which reaches steady state quickly with an infusion.  The usual insulin infusion rate is 0.1 units/hr/kg body weight, which would be 7 units per hour in a 70 kg (average) person.  Seven, five, what's the difference?  The main reason that I have settled on 5 (besides its being conveniently congruent with the dose of LR and the number of orders) is that I don't want to drive the blood glucose down too rapidly.  If that happens, I will have to prematurely turn down the insulin rate before the ketosis is corrected, or - Dread! - I will have to start a D5 or D10 infusion.  Since I have begun using 5 units/hour, I have noted that I rarely have to use D5/D10, and that I have fewer problems with hypokalemia.  (Kaliuresis [loss of potassium in the urine] is caused by high urine flows.  If you're running D5/D10 at a high rate (>200cc/hour), you're going to waste potassium in the urine and have to replace it.)  Plus, it's so complicated - if the glucose climbs to 450 on 4 units of insulin and D5 at 250 cc/hour, what do you do?  Reduce the D5?  Increase the insulin?  It is VERY hard if not impossible to protocolize and simplify DKA treatment if you insist upon titrating insulin especially if you add D5/D10 titration to the mix.

3.)  Check blood glucose every 2 hours; When blood glucose less than 200 mg/dL, reduce insulin drip to 1 unit per hour.

On 5 units per hour of insulin, the blood sugar is unlikely to fall precipitously such that it needs to be checked every hour.  (Hourly checks are arbitrary anyway - why not every 50 minutes or 80 minutes?)  This saves the patient 6 fingersticks in 12 hours, and allows them to sleep in 2 hour intervals.  Whether it's day or night, it is important to recognize that many of these patients have not slept a wink for several days, and an increased interval between fingersticks will be welcomed.  I also order a REGULAR DIET on admission and allow patients to eat and drink whatever they want whenever they have the inclination.  This prolongs the amount of time that the infusion is at 5 units per hour and makes it less likely that glucose will decline precipitously - without the use of D5 and D10. When the blood sugar is below 200, usually 6-12 hours have elapsed and the ketosis has been arrested and mostly reversed.  One (1) unit per hour is the basal insulin requirement for most people in the fasting state.  At this rate, almost all patients will have blood sugars between 100 (safe) and 300 (creeping up and indicating inadequate insulin dosing and possible recrudescent ketosis).

4.)  If blood sugar less than 100 mg/dL or greater than 300 mg/dL on 1 unit per hour, call MD.

If, on 1 unit per hour, the blood sugar falls below 100 or climbs greater than 300, some fine tuning is going to be needed.  I do not try to put this part on auto-pilot for two reasons: 1.) I think that decisions on what to do next would be too nuanced to make protocolized instructions to guide them; and 2.) this so rarely happens that it would be a waste of ink and an unnecessary complication of an otherwise beautifully parsimonious protocol!

5.)  Check serum K+ and HCO3- 12 hours after the start of treatment.

After about 12 hours, the ketosis will usually be reversed as evidenced by a serum HCO3- level of greater than or equal to 14.  At that point, the patient can be transitioned to long acting subcutaneous insulin if symptoms are resolved.  If K+ has been depleted during the process, it can now be corrected.  Rarely (if ever) does it fall into a range where sequelae are apparent with the use of this orderset, so I do not order chemistries at frequent intervals.

If the ketosis has not abated, the insulin infusion can be continued (I usually continue it at 1 unit per hour) or if the rate of correction of ketosis is deemed too slow, the protocol can be modified, insulin increased, and/or D5 or D10 instituted.  If this is the case, it is not a failure of the protocol per se.  The simplified protocol gave the patient the opportunity to improve quickly and with minimal interventions and blood draws.  If s/he did not, little is lost.  The gains, besides the unrealized opportunity, will come to the patients who do improve with the simple and low burden protocol.

If it makes you uncomfortable to treat DKA in such a parsimonious fashion, you can develop your own simplified protocol the way I have - iteratively over several years by "backing off" and backing away from the anal retentive micromanager approach.  While it may not be identical (maybe you prefer hourly fingersticks and are reassured by chemistries every 6 hours), I think that you will find that your protocol converges on this orderset - reduced aggressiveness of insulin dosing, no titration of insulin or fluids, avoidance of intravenous dextrose solutions, fewer labs and fingersticks, tolerance for a wider range of glucose values, and feeding patients during treatment.  I wager also that patients will get better faster, with fewer burdens and resource expenditures, and that you will have more satisfied patients and less bewildered nurses.