We thank Dr. Roulet and colleagues for their astute comments regarding our recently published clinical decision rules (CDRs). The outcomes that the CDRs are designed to predict include moderate and severe adverse drug events (ADEs; or reactions) regardless of their association with the chief complaint. We have previously shown that emergency physicians commonly do not attribute the symptoms of incidentally found ADEs to medication use.[2, 3] Therefore, it is important for the CDRs to identify these events.
We agree with Roulet et al. that the CDRs may challenge the manpower requirements for clinical pharmacists in emergency departments (EDs). However, presently there are no other evidence-based rules or algorithms to help pharmacists select high-risk patients for medication review in this setting. At the present time, patient selection for pharmacist-led medication review is neither evidence-based nor standardized.
Until recently, Accreditation Canada, a national organization that evaluates health care institutions based on the quality of care they provide, required health care institutions to conduct medication reconciliation on all patients at all transitions in care, including those that occur in EDs, regardless of the risk status of the patient and irrespective of existing pharmacist manpower restrictions. Our research is intended to direct this resource-intensive intervention to the highest-risk patients, while allowing us to forego the same intervention in the majority of patients who are at low risk for an adverse outcome. Finally, because sensitivity and specificity are always a tradeoff, rules with greater specificity that are expected to generate a smaller number of high-risk patients will miss a larger number of events. Prior to determining the desired sensitivity cutoff for the CDRs, we surveyed Canadian emergency physicians to determine the lowest sensitivity that would be acceptable by physicians so that the rule would be implemented.
Finally, we agree with Roulet et al. that deriving rules for unattributed or unrecognized ADEs as a primary outcome may help to further rationalize the use of our pharmacist resources, if it is possible to derive more specific rules for this outcome. We foresee two challenges: if ADE recognition is related to physician or institutional factors it is unlikely that a rule can be derived that will be generalizable. Second, the sample size for such a study would need to be larger given that unrecognized ADEs are a subset of ADEs.