Stephanie L. Shimada, Amy K. Rosen, Amy L. Helwig, Dina K. Moss, Michael I. Harrison, Lucy A. Savitz, Haytham Kaafarani, Hillary J. Mull, Peter E. Rivard, Marlena H. Shin, Brian C. Sauer, Jonathan R. Nebeker
| Term |
General definition |
Clinical specificity |
Timing of trigger system firing and investigation |
Potential use of trigger system |
Evaluating trigger system performance |
| Overall Definitions |
| Trigger |
A trigger is a surveillance rule or algorithm that can be applied to a patient's clinical and/or administrative data to identify a time of high risk of one or more iatrogenic adverse events. A positive result of the surveillance rule "triggers" the further review of the patient record to confirm the occurrence of iatrogenic adverse events. |
Triggers may be designed at varying levels of specificity (i.e., to address specific and nonspecific causes and events). |
Triggers may contain logic that is time dependent. Time-dependent logic may determine whether the trigger can be investigated within a clinically relevant time period. |
Discussions of potential uses most appropriately focus on trigger systems, not on triggers alone. See potential uses of trigger systems in next entry. |
It is essential to evaluate the performance of trigger systems and not triggers alone because (1) by definition, it is impossible to implement a trigger without the rest of the investigation system and (2) the implementation of the trigger may have a large effect on the type and amount of information that is available for investigation in practice. |
| Trigger System |
A trigger system is the combination of the surveillance rule, the method of processing patient data according to trigger logic (firing), and the method of investigating a positive trigger to determine whether it is a true positive. |
Clinical specificity is largely a function of the trigger logic. However, a trigger system may increase the specificity of a trigger if the system applies the trigger only in a specific population. |
The method of applying the logic to patient data and investigating a positive (fired) trigger usually determines the investigation timing and the extent to which a positive trigger may be clinically relevant. |
Trigger systems can be useful (1) at the system level for surveillance of iatrogenic adverse events to guide improvement in systematic care processes and (2) at the patient level with subsequent primary prevention of a particular adverse event or secondary prevention (or mitigation) of its resultant harm in the particular patient. The implementation of the trigger system is usually more important than the logic in determining whether it can inform decision at the patient or system level. |
The most important performance characteristics of trigger systems are sensitivity and positive predictive value. The evaluation cohort (also known as the denominator) should be drawn from the union of two populations: that in which the trigger is desired to fire and that in which it can theoretically fire. The reference standard should be applicable iatrogenic adverse events identified through a suitably sensitive and specific method. The trigger should be counted as a true positive only if it fires within a prespecified time span around the occurrence of the adverse event. |
| Clinical Specificity |
| General Trigger System |
A general, or global, trigger system identifies a time of high risk for a broad variety of iatrogenic adverse events. The term "general" is preferred to "global," which misleadingly connotes a comprehensive capture of causes or effects. Examples of general triggers include readmission 30 days after a hospital discharge, filing a tort claim, and changing one's primary care physician. |
General triggers are designed to identify a spectrum of iatrogenic adverse events that is not specific in terms of either a set of manifestations or a set of causes of the events. Most general triggers or sets of triggers address a broad but not comprehensive set of causes and effects. |
The general category does not have inherent criteria for trigger timing. |
General triggers can be helpful for identification and surveillance of adverse events, interventions at the system level, and subsequent monitoring of the effects of these interventions on outcomes, patient safety, and quality of care. Since general triggers are not associated with specific treatments or specific outcomes, they are often less helpful for interventions at the patient level than are clinically specific trigger systems. |
For the evaluation of a single general trigger, an attempt should be made to identify the types of events, the cohort, and the time window that are most pertinent to the general trigger. For a set of general triggers, it is reasonable to evaluate the performance of the system in terms of all iatrogenic adverse events for the entire population of interest. |
| Specific Trigger |
A specific trigger system identifies an event occurring during a time of high risk for a specific type of iatrogenic adverse event or for a definable range of adverse events caused by a specific type of medical intervention. Examples of specific triggers include a positive C. difficile toxin assay and myocardial infarction within a week after surgery. |
Specific triggers are clinically specific regarding the cause of the adverse event, the manifestations of the event, or both. |
The specific category does not have inherent criteria for trigger timing. |
Specific triggers can be translated into both patient-and system-level interventions because they are clinically specific regarding the cause of the adverse event and/or the nature of the iatrogenic adverse event itself. |
Cause-specific triggers should be evaluated only in a cohort drawn from a population In the evaluation of event-specific, triggers, the denominator population should exclude subjects who are ineligible to experience both the effect and a positive trigger. |
| Trigger System Timing |
| Concurrent Trigger System |
A concurrent trigger system allows identification of the time of high risk during the clinical episode in which the adverse event originates. Usually this will occur within minutes or days of the beginning of the period of heightened risk, but it can take longer. How quickly this occurs depends on how long the underlying episode lasts. |
The concurrent category does not have inherent criteria for clinical specificity. |
Concurrency relates to the timing of trigger investigation. A concurrent trigger system provides the ability to identify a problem during the clinical episode in which the problem originates. For example, a concurrent trigger system would alert a provider to hyperkalemia 1 to 3 weeks after starting an angiotensin receptor blocker. |
Some concurrent trigger systems allow productive intervention at the patient level to prevent or mitigate an iatrogenic adverse event. Other concurrent trigger systems identify events for which appropriate treatment is either underway or not possible. |
To evaluate a concurrent trigger system, it is essential to use criteria that define the clinical episode of interest. |
| Real-Time Trigger System |
A real-time trigger system is a type of concurrent system in which the trigger investigation is substantially completed within hours of the trigger firing. |
The real-time category does not have inherent criteria for clinical specificity. |
The trigger investigation is substantially completed within hours of the trigger firing. |
Uses are the same as for concurrent trigger systems. |
Real-time triggers should be investigated within hours of the time the adverse events are about to happen or become manifest. The window of interest should be measured in hours. |
| Retrospective Trigger System |
A retrospective trigger system allows identification of the risky situation only after the clinical episode has occurred. |
The retrospective category does not have inherent criteria for clinical specificity. |
A retrospective trigger system allows identification of the risky situation after the clinical episode has terminated. |
Retrospective trigger systems are most suitable for event rate measurement and for designing and evaluating system-level interventions. |
The requirements for trigger-system evaluation (see definition for trigger systems) apply to retrospective trigger systems. |
| Designed Purpose of Trigger System |
| Interventionist Trigger System |
An interventionist trigger system reliably provides information necessary to guide new action on a specific iatrogenic adverse event in a specific patient. An actionable notification allows identification of unrecognized iatrogenic adverse events that are imminent for the purpose of primary prevention, evolving for the purpose of mitigation, or mature for the purpose of treatment and prevention of an imminent recurrence. |
Although both specific and general trigger systems can provide actionable notifications, medical providers may find it easier to take action on notifications from clinically specific triggers. It is easier for them to intervene when the trigger gives them direction about the cause, the iatrogenic adverse event, or both. |
Interventionist trigger systems are almost always concurrent. Real-time systems are more likely to provide notifications about events that have not already been treated. In some cases an actionable notification may not be concurrent; for example, trigger systems may target delayed adverse events such as osteoporosis from prior drug therapy. |
Interventionist trigger systems are used for primary prevention, mitigation, and secondary prevention of adverse events. Because these systems are designed to identify a fraction of all iatrogenic adverse events, they will not accurately reflect rates of all iatrogenic adverse events. |
There are two alternative metrics of an actionable notification: (1) whether the notification resulted in an event-relevant change in the care plan and (2) whether the notification could have resulted in actions to prevent or mitigate a relevant iatrogenic adverse event. Positive predictive value is possible to assess for the first metric. A counterfactual approach is necessary to estimate sensitivity for either metric and positive predictive value for the second metric. |
| Rate-Estimation Trigger Systems |
Rate-estimation trigger systems are designed primarily to gather information to estimate the burden or rate of iatrogenic adverse events. |
Both clinically specific and nonspecific trigger systems can be used to collect information on counts of iatrogenic adverse events. |
Rate-estimation trigger systems may be retrospective or concurrent. Retrospective investigation has the advantage of being able to more definitively characterize the nature of detected events. |
Rate-estimation trigger systems may provide instructional feedback to clinicians and help managers design system changes. High-performance systems may be used for benchmarking. |
High sensitivity and specificity are essential performance characteristics of rate-estimation trigger systems and are necessary for accurately estimating changes in rates across time or across settings. |
| Awareness-Raising Trigger System |
Awareness-raising trigger systems are designed to increase the salience of certain types of iatrogenic adverse events. |
The need for clinical specificity is a function of the clinical specificity of the intended effect of the system. |
The timing may vary to match the intended data presentation and behavior changes. |
Salience of events may be increased to increase likelihood of patient-level action (learning) or system redesign. |
Positive predictive value as a measure of efficiency is relevant. To accurately measure reductions of all targeted events, a rate-estimation trigger system must be used. |
| Other Purposes |
Trigger systems may have many other uses. |
Clinical specificity must match the system. |
Timing should match the purpose of the system. |
There may be many other purposes. |
The requirements vary according to the purpose. |
540 Gaither Road Rockville, MD 20850 
