Several publications [5, 9, 21, 22] have demonstrated that pharmacist-acquired medication histories in an ED have led to reducing discrepancies; our results were consistent with previous studies. The effect of the intervention reduced discrepancies occurring by 33% (p < 0.0001; 0.1055 OR, 0.05-0.24 95%CI), despite patient recall bias.
Although every attempt was made to interview patients or family members and inspect prescription vials or medication bottles for all study subjects, medications could not be inspected in many cases; the interviewer/researcher relied on written medication lists provided by the patient, caregiver or family or calling his/her house in such situations.
In spite of the intervention, 71 patients (60.7%) in the intervention arm still had at least 1 ED admission medication discrepancy related to home medication. Emphasis was placed on patients reporting the use of at least 1 medication during ED admission. Error rates may have differed regarding services other than admission or among patients taking more than 1 medication. Our results were consistent with previous studies [21, 22] which have reported that medication discrepancies when being admitted to an ED are frequent and clinically significant. Omission regarding over-the-counter medications in our study was 100%.
Obtaining medication history can be challenging. One study has reported that unintentional medication discrepancies were more often due to errors in recording medication history than errors reconciling the history with medication orders. It also reported that relying on family members or caregivers as sources of medication information represented a risk factor .
Our study incorporated MedRec when being admitted to an ED. It has proved useful for improving patient safety by reducing medication discrepancies before harm can occur to patients . The reduction of medication discrepancies in the intervention arm reflected the importance of obtaining accurate and complete medication histories.
Some studies [5, 9, 24] have reported that pharmacists are ideal for supporting a multidisciplinary healthcare team because of their increased familiarity with medications. Our results agreed with previous studies which reported that pharmacists involved in ED admission could increase medication detection level and identify more medications per patient than doctors or nurses when noting their medication history [7, 24], i.e. such histories were more accurate and complete.
Pharmacists used multiple sources for gathering a complete medication history for proactively issuing medication orders to support doctors making prescriptions during ED admission. Dawson et al., have shown that doctors obtained an average of 79% of the complete drug history regarding prescription drug use and 45% for over-the-counter drug use, whereas pharmacists obtained a 100% complete history for both categories of drug use. Todd et al., has stated that pharmacists documented significantly more medication doses and dosage schedules than physicians did (614 cf 446 and 614 cf 404, respectively) (p ≤ 0.001 for both comparisons) and 614 medications were identified by pharmacists for the 55 patients interviewed, compared to 556 identified by doctors (p ≤ 0.001).
In some earlier non-randomised studies of pharmacist-conducted medication histories and assessment before admission to an ED , the most common type of discrepancy involved the omission of a medication which a patient was taking at home (57%); our findings confirmed such report (55.1%).
Hayes et al.,  found that allergy documentation was recorded for 62 patients in a control group (79%) compared to 60 patients in the study group (100%) (p = 0.001). Twenty-five patients were identified in our study suffering from allergies in the intervention arm whilst the standard care arm identified 23 (p < 0.001). Pharmacists identified an almost equal number of allergies as did the doctors.
Overall, a reduction in discrepancies was due to intervention, expected clinical conditions (number of medications being taken) and a particular ED setting. Although the hospitals participating in the study were selected as they had very similar general characteristics, it was revealed that an ED healthcare setting significantly affected the risk of discrepancies occurring.
The hospitals may have had different results because staff training/procedures were different and training admitting physicians and medical students could have had some impact. Significant barriers often occur regarding accurate and complete medication information being obtained during admission related to patients who go to hospitals due to the differing demographics of the populations which they serve. The amount and availability of resources for healthcare attention and the hospitals’ capacity for attending patients could have influenced the results. These observations were made during data collection.
Evaluating the potential clinical impact of the unintentional discrepancies identified during our study showed that 23.9% were judged to have had the potential to cause a patient severe discomfort. Potential harm was driven by the type of discrepancy (incorrect or omitted dose). An example would be a patient who was admitted with cardiac arrhythmia; 150 mg propafenone was ordered by the ED doctor (R2) because it had been noted during the first patient interview (R1) that he had been taking propafenone; however, the ED doctor did not know that a lower dose had already been ordered by a cardiologist 6 weeks earlier (i.e. based on later scrutiny of the patient’s record by MedRec staff – R3).
Zed et al.,  evaluated incidence, severity and preventability of medication-related visits to an ED, establishing that severity had been classified as moderate; of all patients having medication discrepancies at the time of ED admission in our study, more than two thirds of such discrepancies had the potential to cause moderate harm.
The number of patients having at least one interaction was very high (36.8%), compared to studies elsewhere . This could have been related to different definitions or prescribing practice. Further analysis is required for identifying what might have motivated the above differences (our group is currently working on an adverse-events-based paper to extend this line of research). Interactions do not always represent contraindications for use, but knowledge and an appreciation of them before prescription is essential for the safe use of medication. This should thus be considered when a doctor decides to continue with a set prescription after evaluating any possible interaction.
It was estimated that a pharmacist would have to spend 29.5 minutes per patient to provide this service for patients in a similar ED. However, such calculation was based on uninterrupted work-flow times and may have underestimated the actual time required outside a research setting. A consultation with just a pharmacist in an ED focusing on a medication list may be a solution. Models maximising the use of a pharmacist should be evaluated in future research.
This study had several limitations. Despite the study being conducted in teaching hospitals, the results may not be generalised to other settings because an ED setting was a factor regarding the risk of a discrepancy appearing. Future research could examine the effect of ED admission setting and blocking; a cluster study should thus be carried out.
It is acknowledged that the validity of any method for scoring medication discrepancy is difficult to assess because there is currently no gold standard for comparison. A reconciled history (F3) should provide the gold standard for identifying home medication use; however, this could have been a limitation of this study. Patient or caregiver reports were relied on, in conjunction with collateral information from medication vials whenever possible.
Error rates may differ regarding services other than an ED concerning admissions which are elective or involve a transfer from another healthcare facility, or concerning patients taking more than 1 medication. Our findings may not have been representative of other institutions using processes different to MedRec for admission. Eligible patients were not followed-up beyond the study; thus, we are not aware of the effect of such ED admission process on medical outcomes.
The rating method used for assessing the potential severity of the discrepancies and F1 questionnaire used during a medication history interview has not been validated. Inter-rater agreement was also not evaluated as interviewing the same patient twice could have led to recall bias.
The hospitals involved in this study are developing a MedRec pathway which will incorporate some strategies based on the present study’s findings. The next phase will involve an assessment of medication discrepancies once the new MedRec protocol is in place. The data presented herein has suggested that recording medication histories on admission could be improved. Future research could involve a cost-benefit analysis of the intervention for implementing such initiative.