A key driver for safety, medication reconciliation (MedRec) is the process for verifying a medication list at a transition of care. Implicated in 20% of fatal medical errors, there is a paucity of literature about its accuracy. Limited data suggest 22-72% of pediatric admissions contain a MedRec discrepancy. With multiple caregivers involved during an admission, the ability to understand MedRec is a true opportunity. We aimed to characterize MedRec at our hospital. We conducted a cross-sectional observational study of 50 consecutive admissions to the general pediatrics inpatient service. We completed an independent admission MedRec with the patient and/or guardian. Demographic data, admission MedRec completed by the nurse, and by the admitting physician were abstracted from the EMR. The patient’s outpatient pharmacy provided a list of active medications. A best medication list (BML) was determined by the authors. For each information source compared to the BML, the percentage of correct medications was calculated. Binomial tests assessed whether each of the %-correct values differed from 50% (the value expected by chance). Absolute and chance-corrected (kappa) agreement assessed congruence among the four information sources. Substantive results were similar when mixed effects models were used to cluster medications by patient. 3 patients met exclusion criteria (n=47). 7 of 47 (15%) admissions occurred during daytime hours. To limit bias, analysis was restricted to patients with 1 to 5 medications (study n=38). This left 9 patients with 1 medication (24%), 13 patients with 2 medications (34%), 6 patients with 3 medications (16%), 5 patients with 4 medications (13%), and 5 patients with 5 medications (13%). Average age was 5.9 years. MedRec errors were identified in 16 admissions (42%) ranging in severity by NCC MERP Index from A to E. Figure 1 shows the percentage correct medication by source. Table 1 shows the chance-corrected agreement between sources. Overall, the family reported the correct medication most often, while the pharmacy had the medication correct least often. With the exception of nurse, MD, and pharmacy reports for OTC medications, all %-correct were significantly greater than chance; pharmacy reports were significantly worse than chance. Assessing pairwise agreement, for OTC medications nearly all the overall agreement comes from that between nurses and physicians with little agreement between any of the other pairs. For prescription medications, overall agreement and pairwise congruence were very low but for all pairs the percent of the time when there was agreement on the correct answer was very high. Error pervades MedRec. It is a dynamic list that is not owned by any one person/entity. In our hospital, the available medication list in the EMR is a better starting point for an accurate MedRec than what patient/guardian reports.

Figure 1

Percent of medications reported correctly by each information source, broken down by whether the medication was over the counter (blue bars) or prescription (orange bars)

Figure 1

Percent of medications reported correctly by each information source, broken down by whether the medication was over the counter (blue bars) or prescription (orange bars)

Close modal
Tables 1a-b

show the kappa (chance-corrected agreement) values overall and between each information source. Because agreement between two sources could potentially be high if both sources agree on an incorrect answer (or artificially low if there is very little agreement on an incorrect answer), agreement is also broken down by the % of the time when the sources agree on a correct medication and the % of the time when they agree on the incorrect medication.

Tables 1a-b

show the kappa (chance-corrected agreement) values overall and between each information source. Because agreement between two sources could potentially be high if both sources agree on an incorrect answer (or artificially low if there is very little agreement on an incorrect answer), agreement is also broken down by the % of the time when the sources agree on a correct medication and the % of the time when they agree on the incorrect medication.

Close modal