The dynamic TIMI risk score for STEMI is a prospectively validated means of prognostication that incorporates an accepted tool, the TIMI risk score, to allow a continuous assessment of risk. Risk stratification is a dynamic process, altered by a patient's changing condition and informed by each preceding complication.7,23
Therefore, a more complete scoring system should deliver an initial estimate of disease severity and then incorporate in‐hospital events prior to rendering a long‐term mortality estimate.
Our results have generated a classification system that can be easily calculated at the bedside, has been validated in a large unrelated cohort, and takes into account the dynamic nature of risk assessment. The TIMI risk score can be used in its original form to evaluate historical information, physical examination, and details of the presentation to provide clinicians with a measurement of initial disease severity for patients entering the hospital, through prediction of their potential in‐hospital and 30‐day mortality.3
The 6 elements of the dynamic score occurring during the index hospitalization can then be added to the admission score to produce an estimate of 1‐year mortality on discharge (). This information may then be used by the discharging physician, outpatient clinician, and patient to help steer subsequent care.
A number of helpful stratification schemes are currently in use, many of them focused on supplying prognostic information at presentation.3–6,8
Although some validated risk scores include in‐hospital events, none allow a physician to assess risk at admission and then easily reclassify on discharge while retaining the simplicity of an adjunctive scoring system.4,10–12
The dynamic TIMI risk score for STEMI retains all the benefits of the original TRS including the absence of weighted terms, the simplicity of integer values, and the ability to produce a final score through simple addition. Furthermore, the dynamic score is fully compatible with the original, allowing a risk assessment on both admission and discharge. The dynamic risk stratification described here provides the benefit of a highly accurate estimation of 1‐year mortality and an excellent discriminatory capacity without the need for computer calculation. It can easily be calculated by either physicians or physician extenders. With the needs of underresourced medical settings in mind, the dynamic risk score has been developed without novel markers or laboratory values. All the score's elements are easily recognizable clinical events that do not require prognostic tests.
A dynamic risk score that can provide both an initial risk assessment and subsequent discharge reclassification could help clinicians to make decisions about the postdischarge care of STEMI patients. Physicians caring for patients post‐STEMI could use the dynamic TRS to inform frequency of follow‐up and decide on the threshold for a monitored trial of treatment withdrawal. The risk‐benefit ratio for the use of therapeutic devices and drugs varies with a patient's estimated mortality and morbidity.28–29
Without a detailed understanding of the patient's risk at discharge, it is difficult to evaluate the efficacy of new treatments. Also, treatments such as prasugrel or ticagrelor may have enhanced absolute and relative benefits when there is a higher estimated mortality.29
The dynamic TRS can better define risk, perhaps guiding tailored cardiopulmonary rehabilitation, frequency of follow‐up visits, and consideration of potent but expensive therapeutics for targeted populations.30–33
In addition, the dynamic TRS could be used by investigators planning future randomized controlled trials to target appropriate populations for hypothesis testing. It could help to define a higher‐risk cohort for whom more intense therapy can reduce mortality (eg, adding a new oral anticoagulant on top of current therapy). Alternatively, the dynamic TRS may be useful in targeting lower‐risk patients who would be good candidates for withdrawal of therapies after a finite post‐MI period. Because innovation in post‐STEMI care has been additive, most commonly used medications have not been tested in the setting of modern treatments. By identifying low‐risk patients, we can begin a more contemporary assessment of post‐STEMI treatments. Such analyses would be particularly relevant in our current environment of restrictive economics.
As with any attempt to balance the need for simplicity versus the desire for accuracy, this study has its limitations. The proportionality assumption was examined for the new in‐hospital variables. The only new in‐hospital variable that violated the proportional hazard assumption was renal failure. We believe that in this case the proportional hazard test is probably unstable because of the small number of subjects with renal failure (n=17). It is widely accepted that the Cox proportional hazards model is robust to small proportionality assumption violations.34
The overall hazard ratio for renal failure can be interpreted as an average effect over time.
In addition, it should be noted that the derivation and validation of the dynamic risk score has been conducted in phase 3 studies. The score was derived in a majority fibrinolysis‐treated population from ExTRACT‐TIMI 25 and found to be translatable to an entirely PCI‐treated population in TRITON‐TIMI 38. Because the score was developed using patients who qualified for enrollment, application to a variety of medical settings must still be assessed.35–36
Although absolute risk prediction based on the derivation data set may not be applicable to all populations, the score's ability to risk‐stratify has been reliably demonstrated in the TRITON‐TIMI 38 external population. The score has been validated only for use with STEMI patients and may not be able to reliably predict risk at alternate times or for populations with unstable angina and NSTEMI. The decision to exclude novel markers and laboratory studies while focusing on in‐hospital clinical events was consciously made to allow for wide employment of the score without the need for further testing. Although the exclusion was deliberate, factors such as troponin elevation and lack of ST‐elevation resolution have been shown to confer risk and may have a significant independent effect on mortality.37–40
It is also important to note that we have used the original TRS as a comparator when assessing the performance of the new score. Although the original TIMI risk score for STEMI was derived for short‐term mortality, the dynamic TRS has only been derived and validated for 1‐year outcomes. The dynamic TRS updates the risk for patients who survive to discharge and allows for reclassification based on in‐hospital events. In essence, the new model is predictive of a new outcome not assessed by the original TRS. The decision was made to use the original TRS as the basis for comparison because it shares a common derivation and continues to have a strong discriminatory capacity for prediction of 1‐year mortality.8