to a hypoglycemic patient. Similar errors can occur when the line from which the specimen is being drawn has “dead space” and that line cannot be flushed adequately, diluting the sample and giving an erroneous result.31
Each hospital must be aware of the strengths and the weaknesses of the POC meters used in their institution. POC meters that use a glucose-oxidase method will tend to give falsely elevated glucose values when the patient is hypoxic, and lower glucose levels when high levels of acetaminophen are present. Glucose-dehydrogenase meters will give more accurate levels in the presence of hypoxia, but falsely lower results in the presence of dopamine infusions and falsely higher glucose levels when higher levels of acetaminophen are present.
In addition, the storing and the care of the glucose monitoring strips are crucial. If the bottle containing the strips is left open, the strips may deteriorate rapidly. If the strips are stored at a temperature ≥39.2°C, the strips may give falsely low blood glucose values, and when the strips are stored <20°C, the strips may give falsely high values. Strips also may be destroyed by extremes in temperature or humidity.
Some of the strips hospitals now use do not have good corrections for hematocrit levels. This is particularly crucial in intensive care and operating room suites, where wide variations of hematocrit are common. Falsely elevated glucose levels in POC meters when patients are anemic occur and most meters do not have a specific method to deal with hematocrit variation. Cembrowski, a NICE-SUGAR investigator,32 found that some of the batches of his POC glucose meter strips did not correct for hematocrit variations as claimed by the manufacturer. He concluded that the error due to the faulty strips may have misled his team and caused the diagnosis of hypoglycemia to be either delayed or missed entirely.32
In addition, proper training of staff on the proper technique of performing POC glucose monitoring is crucial. Too often, POC glucose monitoring is done in hospitals by the most poorly trained and poorly supervised staff. The resultant errors can be quite large. For example, if a person fails to clean the finger properly before obtaining a glucose sample, the result may be as much as 35% or more above the true glucose level, while failure to adequately dry the finger can dilute the sample and give a falsely low value that may be as much as 25% below the true glucose level. Such large deviations from the true blood glucose may cause an outlier that can mislead the prescribing health-care provider.33
Glucometrics
Evidence-based metrics can be used to evaluate not only the harmful errors but also noninjurious errors and latent errors. A latent error occurs when a deviation from optimal care occurs that, although not directly causing harm, increases the probability that an important error will occur later. Some latent errors include systemic issues that create recurrent vulnerability of the health-care system, such as excessively high patient-to-nurse ratios, a situation in which the nurses have more to do than they can safely perform. Another common example of a latent error is leaving a vial of U-500 amid a number of U-100 vials. This can be reliably expected to increase the risk of vial confusion, potentially resulting in an overdose of insulin. Another example of a latent error that is a systemic error at the so-called blunt end of care is allowing inadequate lighting in an area of the hospital where staff must read labels on vials, syringes, and infusion sets.
Both latent errors and noninjurious errors are important to identify and correct, as these ultimately may lead to patient injury. It is imperative to educate the caregivers and to give them feedback, if we are to reduce the total number of errors. Moreover, because injurious errors usually involve multiple errors, often by multiple caregivers, the chain of undetected errors and the sequence of errors may result in an injurious error. Medicine is a high-risk endeavor and errors are frequent. Although most errors are corrected by the person who made them, multiple errors often include some that escape detection and cause harm. As an example, the development of DKA de novo in a hospital setting, still a far-too-common event, usually involves multiple people and multiple errors of omission and may result in an injurious or even lethal event.
Diagnostic Errors
No discussion of errors in insulin therapy is complete without discussing the role of the individual and the human factor in errors in insulin therapy. Although defective systems may create potential for “error traps” or situations in which different individuals make the same error, the role of the individual at the POC is key to understanding how to prevent injurious errors.
Humans frequently make errors and, when candid, most people will admit that they frequently have slips and lapses that, for the most part, they correct themselves. Not only do people have different cognitive abilities, but most people have two distinct mental processes active at any time, often termed by cognitive psychologists as type 1 and type 2 thinking.34 Type 1 is intuitive, rapid, and almost automatic, relying upon a vast warehouse of experience and knowledge. Many people use this method of thinking much of the time because of its ease and speed. Examples of this type of thinking include what we do when driving a car or when a jazz musician is playing with their fellow musicians. This type of thinking, however, although blindingly fast and intuitive, is still error prone, subject to cognitive biases, and not at all quantitative. Type 2 thinking, by contrast, is slow, deliberative, and analytic. This form of thinking is more likely to be useful in quantitating risk, but may be just as prone to some cognitive biases as type 1 thinking. We would prefer our accountant to use type 2 thinking. But when it comes to the formulation of a medical diagnosis, particularly when the diagnostician feels confident or is working quickly, pattern recognition, which is usually a function of type 1 mental processes, is the predominant pattern. Although the experienced diagnostician may be correct, the pattern of thinking they most often use is particularly prone to overconfidence and a premature closing of the possibility that the diagnosis may either be incorrect or incomplete.
In a seminal study of the causes and frequency of diagnostic errors, Graber et al.35 looked at 100 cases of diagnostic error. These resulted in 90 injuries and 33 deaths. Of the 100 cases, seven were due to no-fault errors alone. Of the remaining 93 cases, many had system-related factors (63%) and cognitive errors. The system-related factors were most often organizational problems (94.3%). Only 5.6% of the system-related factors were the result of technical and equipment problems.
In 74 cases of diagnostic errors resulting from faulty cognition, cognitive factors were noted 320 times (4.3/case). The most common problems were faulty synthesis (82.8%) or faulty data gathering (14.6%). Surprisingly, inadequate knowledge or skill accounted for only 3.4% of the diagnostic errors resulting from faulty cognition. Put another way, it wasn’t common that the errors occurred because of a lack of knowledge, but rather because of how the clinician collected and put together the data to formulate a diagnosis.
Certain cognitive and system-related factors co-occur commonly, such as an inadequate history leading to misinterpretation of lab results. In general, faulty information gathering greatly increased the risk that there would be a faulty synthesis of data and premature closure, as for example, “it can only be this.” Faulty data gathering was identified in 45 instances by Graber et al.,35 but they identified inadequate or faulty knowledge or skills least often, in only 11 cases overall in the study. The researchers also identified failure to consult an expert as a significant cause of diagnostic error (15 cases), as well as failure to periodically review the situation (10 cases) or failure to gather other important information to verify the diagnosis.
Medical diagnostic errors are common when the presentation of the patient is atypical, as for example if a young woman appears in the emergency room with right lower quadrant pain that in actuality is due to undiagnosed DKA and not appendicitis. If therapy for DKA is delayed or even not done, the results may be catastrophic. It is likely that many of the so-called never events noted by CMS were due to diagnostic errors early in the development of DKA, HHS, or severe hypoglycemia.
Another not uncommon example of a diagnostic error that may occur is in the elderly patient with diabetes with HHS. The patient may present
