The EQUIP study done in the UK provides some interesting insights into prescribing errors. Though originally established to look at prescribing errors made by first year residents, the study demonstrated that prescribing errors were both common and made by physicians at all levels. Looking at over 100,000 medication orders across 20 hospital sites, they found an average error rate of 8.9 errors per 100 medication orders. The error rate for first year residents, responsible for about half the orders, was 8.4% - actually lower than that for the entire group. All physician levels, including attendings, made prescribing errors. The highest rate (10.3%) was actually seen for second year residents. Interestingly, nurses and pharmacists who were allowed to order medications, had much lower error rates.
Potentially serious errors occurred in about 5% of cases and potentially lethal errors in about 2%. The good news, however, is that almost all errors were caught by pharmacists before they could affect patients. Errors were made most often at the time of admission. In fact, orders at admission were 70% more likely to contain an error. Analgesics, antibacterials, bronchodilators, antianginals and corticosteroids were the classes of medication most frequently involved.
Of the error types, the commonest was omission of a drug on admission, accounting for almost 30% of the errors. Wrong dose or missing dose accounted for almost another 30% of the errors. Other errors involved times, routes of administration, formulations, duplications, clinical contraindications, and a variety of others.
Medications that were order via electronic order entry were 12% less likely to contain errors than those in handwriting.
They also performed a systematic review of the literature. In that review, overall error rate was slightly lower (median 7%) but occurred in over 50% of hospital admissions and, again, most errors were intercepted before harm came to patients. Errors were commonest with antimicrobials and were more common in adults than children. The most common type of error was incorrect dosage.
The most important part of the study, however, was to capture insight into factors contributing to prescribing errors. They did both a systematic review of the literature and interviewed residents and medical school curriculum directors to get further insight.
In the interviews with residents, one of the most striking features was that residents often did not remember making errors (though they did suspect they made more errors than they knew about), nor did they seem overly anxious about them. Interviewees often distinguished between “silly” errors and more serious ones. An example of a “silly” error might be ordering a medication at the wrong time of the day.
They grouped errors according to Reason’s model of accident causation (active failures, error-provoking conditions, and latent conditions). Active failures were identified and classified as either knowledge-based errors, rule-based errors or skill-based memory lapses and slips. Of 85 errors analyzed, 18 were mainly due to a knowledge-based mistake, 34 to a rule-based mistake, and 23 were mainly due to slips or lapses and three were direct violations. Seven others were related to receipt of incorrect information and therefore were not the active failure of the respondent but of another individual. They provide a nice diagram and verbatim comments from the respondents illustrating how slips interacted with contributing factors such as workload or time pressures to result in errors. They provide a similar diagram and verbatim comments for skill-based memory errors. Examples such as forgetting the dates or times for medications often elicited comments about inadequate design of the prescription “chart” or the electronic order entry screens. Both slips and memory lapses were types of errors often captured by the “safety net” (nurses and pharmacists) and seldom resulted in patient harm.
A similar diagram is used to illustrate 34 rules-based mistakes. Here, lack of expertise comes into play. That is often lack of expertise in framing the clinical situation and, as a result, applying the wrong rules. One issue raised here and elsewhere is that the residents felt so comfortable with the “safety net” that they expected a nurse or pharmacist would identify any errors.
The same process was applied for knowledge-based mistakes and for communication errors. Often in knowledge-based incidents residents did not seek out help (due to a variety of factors such as workload, time pressures, fear that others would look upon the lack of knowledge as a weakness, etc.). Some of the comments, however, showed that they felt better once they realized it was okay to have to look something up. Especially seeing one of their more senior physicians look up a dose was reinforcing in that respect. Regarding communication errors, the most salient finding may be showing how easily some errors may get perpetuated throughout the healthcare system.
The section on violations is excellent. Examples are given where computer-generated alerts are willfully ignored or where intentional violations are done just to “get the job done”.
They did find some overarching themes in prescribing errors:
A section with extensive comments about prior educational experiences of residents is included, as well as those about transitioning from medical school to the first year of residency. Perhaps the most significant comments relate to the difference between pharmacology and prescribing. The former is well taught in medical schools, the latter poorly covered. Thus, there is a big gap between the theoretical and the practical. During the first year of residency, training in prescribing was appreciated but most wanted more. Lack of feedback about prescribing errors was also noted as a system problem.
They also performed a systematic review of those studies which reported causative or contributing factors for medication prescribing errors and grouped them according to Reason’s model of accident causation (active failures, error-provoking conditions, and latent conditions). The most common active failure was a mistake due to inadequate knowledge of the drug or the patient but skills-based slips and memory lapses were also common. Error-provoking conditions included lack of training or experience, fatigue, stress, high workload, inadequate communication, etc. Latent conditions included reluctance to question senior colleagues and inadequate provision of training.
Just as we typically see when we do root cause analyses of adverse events, prescribing errors are often multifactorial, with several active failures and error-provoking conditions often acting together (the error cascade) to result in the ultimate outcome. Because of that, the authors caution that solutions addressing a single cause are likely to have only limited benefit and that multifactorial interventions will likely be necessary. In particular, the authors suggest that education, in the traditional sense, is not likely to significantly impact prescribing errors. If education is to play a significant role, it would have to be more along the lines of “just in time” education.
Amongst their recommendations are:
This document is 215 pages long but it is fairly easy reading. It is also an excellent introduction to the Reason model of accident causation for those who need some introduction into the science of “human factors”. It also has some excellent diagrams that illustrate both the complexities and interactions of numerous contributory factors. And it has a great bibliography and review of the literature on prescribing errors. Read it – you’ll see that all the contributing factors in play are also in play at your facility whether you are a teaching hospital or not.