of limb weakness that mimic a cerebro-vascular accident (CVA), but the real diagnosis is severe HHS, and if care is delayed the risk for mortality is very high.
The routine checking of other providers’ work and conclusions in real-time is crucial in preventing such diagnostic errors. It is always useful for the diagnostician to ask whether there is some other explanation for what they see than the diagnosis they have decided on. It is also useful to provide feedback to all members of the diabetes care team. Often the feedback of relevant data will allow people to revise their initial impression and protect the patient from serious errors in insulin therapy.
It is surprising that rule-based errors are often difficult to correct. The presence of a rule that is easy to use, but incorrect, often generates resistance to change. Probably the best known example is sliding-scale insulin (SSI). It is an example of a simple, clear rule of giving insulin that is “strong, but wrong.” Because of the simplicity of such rules, there is often resistance to discontinue using these, even when people know that the rule does not work well. Ideally, SSI as monotherapy should not be allowed to be part of a computerized insulin order set, and basal-bolus insulin orders promoted as the alternative.
Types of Insulin Errors
In 2013, the ASHP convened a panel of experts to focus on the goal of enhancing the safety of insulin use in hospitals. They began by grouping the types of errors in insulin therapy into six categories: prescribing, transcribing, dispensing, storage, administering, and monitoring. Their nomenclature is a useful place to begin.4
Prescribing Errors
These errors are among the most common and the most important. Among the common examples is when the prescriber chooses an incorrect dosage or a method of insulin dosing that is irrational, as for example, SSI as monotherapy. More variables should affect the choice of insulin dosage than just the immediate glucose result. SSI monotherapy is both a rule-based error and also a knowledge-based error, because it indicates both an illogical belief and a lack of understanding of insulin therapy. The evidence shows that it is an inferior method of prescribing insulin at best, and at worst, it has resulted in severe morbidity and even mortality when prescribed to a patient with DKA or HHS.
During insulin prescribing in hospitals, the prescriber needs to provide for both basal and bolus insulin requirements, the bolus doses used to balance nutritional intake, and correction doses when the glucose is outside of the optimal glycemic range. The basal needs may be highly variable, affected by the underlying comorbid conditions and by concurrent medications, which may increase or decrease insulin resistance, or by other mechanisms increasing the risk for hyper- or hypoglycemia. Some examples are shown in Tables 4.2 and 4.3.
Table 4.2—Comorbid Conditions
| Increasing hyperglycemia risk | Increasing hypoglycemia risk |
| Infections | Weight loss |
| Myocardial infarction | Renal failure |
| Metabolic acidosis | Advanced age |
| Severe pain or anxiety | Adrenal insufficiency |
| Pregnancy | Liver failure |
| Surgery | Heart failure |
| Acute asthma | Alcoholism |
Table 4.3—Medications
| Increasing hyperglycemia risk | Increasing hypoglycemia risk |
| Corticosteroids | β-blockers |
| β-agonists | Incretins (when used with insulin, sulfonylureas) |
| Protease inhibitors | Sulfonylureas |
| Sirolimus | Haloperidol |
| L-asparagine | Pentamidine |
| Atypical antipsychotics | Tramadol |
The nutritional needs of the patient need to be coordinated with a coherent plan for insulin therapy, and the prescription for insulin therapy must reflect the current nutritional therapy and route, as well as provide for a change in insulin therapy if nutritional intake is reduced or stopped. Different methods of nutrition require different types of insulin, and sometimes different routes. For example, basal insulin dosing twice daily may be optimal for enteral feedings, but IV insulin is best when oral intake is uncertain and the expected insulin requirements may decrease rapidly. Orders for varying contingencies, such as what to do if oral intake is suddenly interrupted, will reduce errors.
Poor communication between the prescribing and treating health-care providers can result in an error of inadequate insulin dosage. For example, if a pulmonary intensivist adds β-agonists and corticosteroids for the respiratory needs of the patient, but the prescriber of insulin is unaware that the consultant, in effect, has increased the insulin resistance, the prescriber will choose inadequate insulin doses that will result in severe hyperglycemia.
The route of insulin administration is a key consideration. A patient with shock and hypotension or severe dehydration may be highly likely to have delayed and erratic absorption from subcutaneous (SQ) sites. An IV approach for insulin therapy would be much more effective. Alternately, in another scenario, the rapidity of the change in insulin resistance may be so fast that only an IV use of insulin therapy will be able to match the rapidity of the change in insulin needs.
The type of medical record, either paper-based or electronic, may affect the type and frequency of prescription errors. In paper-based hospital records, the errors may occur when prescribers write down what they believe to be the correct insulin type, but instead list an incorrect type of insulin because the insulin they chose had a similar sounding, but incorrect, type (short-acting versus long, or vice-versa). This kind of error is often termed an intentional error. In contrast, an example of a so-called unintentional error is a misspelling or error in the prescriber’s penmanship, which makes the result ambiguous in appearance. The most common error is using a “U” to depict units, which, if not written clearly, may be read as a zero. Another common error is to have a trailing zero after a decimal point, as for example, an IV insulin rate of 1.0 units/h, which, if the decimal point is not easily visible, may be seen as 10 units/h. Other examples of prescribing errors occur when the orders are verbal and may not be clearly understood by the transcriber, who is not familiar with the plan. These common errors can be reduced with the use of evidence-based order sets, preferably in an EHR.
Transcribing Errors
Transcription errors, while much more common when paper charts are used or a verbal order is transcribed, also may occur whenever there is a transition of care and a new set of orders is used. In many hospitals, a reconciliation of medication is done on a transfer from one care unit to another, and a new set of orders for insulin is generated. The person who performs the reconciliation is the one who may cause a transcription error and it is particularly common upon discharge in cases in which new orders are generated for post-hospital care. As an example, in one recent study, 18% of all patients who were discharged after an acute myocardial infarction (MI) did not receive their medications to control blood glucose levels upon discharge.36 Sixty-seven percent of the time, one of the omitted medications was insulin.36 Their retrospective review confirmed
