I honestly don’t know where I got this (it’s a little too eloquent for me to have written it). If anyone recognizes where it came from let me know. Its an excellent analysis of why errors are so common in clinical care.
“Human beings are complex organisms with extremely intricate interactions. In the verbiage of Normal Accident theory, human medicine is an industry with high complexity and loose coupling. The complexity is self-evident, the coupling implies that there are few direct, inviolate interactions in the human system. It also means that there are opportunities for recovery of the organism after failure. As an analogy, a nuclear missile in flight is a tightly coupled system of high complexity. Once in flight, in the absence of a self-destruct mechanism or a competent anti-missile system, the launch will invariably lead to a landing of the missile and detonation. Shooting a gun at a target is a system of low complexity and tight coupling (once the trigger is pulled there are few ways to prevent the outcome), a sailor in a rowboat trying to arrive at a dock is a system of moderate complexity and loose coupling in that there are many variables, and many approaches and recovery strategies.
Thus, when we look at the delivery of clinical medicine, and the high error rate, several paradigms emerge. First, mistakes often result from incorrect modeling. In examinations of medical errors, the root causes often center on incorrect assumptions about the physical state of the patient, and the necessity for further intervention. Also, due to the high complexity of the delivery system, errors abound in communication, accuracy of results, and finally in efficacious decision making and delivery of therapy. So, when we try to ascertain remedies for incorrect modeling, where do we start?
Models are created from several sources: internal biases, knowledge base and experience; available quantitative data (labs, etc.), qualitative data (patient complaints, physical examination, etc.), and finally external sources (communication, time pressure, etc.). In creating an accurate model you would desire: accurate quantitative that reflects the true nature of the physical condition and performance of the organ system, confirmatory qualitative data to reinforce the hypotheses created from the quantitative data, sufficient knowledge and experience to integrate this data into an effective diagnostic and treatment plan, and visualization of therapy and its effect and possible complications.
What has been previously stated is not without controversy. Our belief is that in the acute and critical care environment where quantitative data is readily available and of reasonable accuracy, where the risk of serious consequences is high, and where many treatments are readily available better care can be provided by using the following process:
• setting up qualitative and quantitative filters to indicate abnormal system performance,
• initial quantitative confirmation of the triggered filter ( but with a very low index of suspicion for dismissing the triggered filter since it is based on tenuous data and is not meant to exclude system problems)
• Targeted quantitative examination through labs and studies
• Qualitative confirmation of the hypotheses created on the basis of the quantitative data (with the emphasis placed on the validity of the quantitative data, and not on the ability of the exam and history to dismiss quantitative evidence of poor system performance).
• Creation and implementation of a treatment plan the best incorporates the known data
• Follow-up to insure that the model was correct and that system performance is improving.
The reasons for the dependence on quantitative data are several. First, there is tighter coupling between laboratory data and system performance, slightly looser coupling for radiological or other studies that depend on interpretations, and extremely loose coupling for physical examination and medical history (meaning that if a lab was run 10 times on the same sample, one would expect little variation, however if a history and physical examination was performed by ten different clinicians, there is likely to be significant variability). Remember that when problem solving in any system, one wants to decrease the complexity of the problem and increase the accuracy of the measurement you are using. Possibly the most complex interaction with the loosest coupling is to walk into a patient’s room who is ill and attempt to diagnose and treat their condition solely through the use of qualitative measures. The tightest coupling would be to measure hourly urine output as a determination of renal function.”