Twelve-Month Review of Infusion Pump Near-Miss Medication and Dose Selection Errors and User-Initiated “Good Save” Corrections: Retrospective Study

Abstract

Background There is a paucity of quantitative evidence in the current literature on the incidence of wrong medication and wrong dose administration of intravenous medications by clinicians. The difficulties of obtaining reliable data are related to the fact that at this stage of the medication administration chain, detection of errors is extremely difficult. Smart pump medication library logs and their reporting software record medication and dose selections made by users, as well as cancellations of selections and the time between these actions. Analysis of these data adds quantitative data to the detection of these kinds of errors. Objective We aimed to establish, in a reproducible and reliable study, baseline data to show how metrics in the set-up and programming phase of intravenous medication administration can be produced from medication library near-miss error reports from infusion pumps. Methods We performed a 12-month retrospective review of medication library reports from infusion pumps from across a facility to obtain metrics on the set-up phase of intravenous medication administration. Cancelled infusions and resolutions of all infusion alerts by users were analyzed. Decision times of clinicians were calculated from the time-date stamps of the pumps’ logs. Results Incorrect medication selections represented 3.45% (10,017/290,807) of all medication library alerts and 22.40% (10,017/44,721) of all cancelled infusions. Of these cancelled medications, all high-risk medications, oncology medications, and all intravenous medications delivered to pediatric patients and neonates required a two-nurse check according to the local policy. Wrong dose selection was responsible for 2.93% (8533/290,807) of all alarms and 19.08% (8533/44,721) of infusion cancellations. Average error recognition to cancellation and correction times were 27.00 s (SD 22.25) for medication error correction and 26.52 s (SD 24.71) for dose correction. The mean character count of medications corrected from initial lookalike-soundalike selection errors was 13.04, with a heavier distribution toward higher character counts. The position of the word/phrase error was spread among name beginning (6991/10,017, 69.79%), middle (2144/10,017, 21.40%), and end (882/10,017, 8.80%). Conclusions The study identified a high number of lookalike-soundalike near miss errors, with cancellation of one medication being rapidly followed by the programming of a second. This phenomenon was largely centered on initial misreadings of the beginning of the medication name, with some incidences of misreading in the middle and end portions of medication nomenclature. The value of an infusion pump showing the entire medication name complete with TALLman lettering on the interface matching that of medication labeling is supported by these findings. The study provides a quantitative appraisal of an area that has been resistant to study and measurement, which is the number of intravenous medication administration errors of wrong medication and wrong dose that occur in clinical settings.

The impact of drug error reduction software on preventing harmful adverse drug events in England: a retrospective database study

“…Context is also important in gauging potential severity as patient comorbidities would impact on dose-related or dose–rate-related response and many DERS studies have examined pump operation around HMX alerts. 29 30 Furthermore, as a study covering 1 year of data, we have not allowed for any changes to the device programming parameters being made in response to pump data reviews at each site. It cannot be excluded that the incidence or nature of events may have changed over the study period.…”

“…Thus, assumptions are implicitly made about each HMX event based on the personal and professional opinions of the rater. Context is also important in gauging potential severity as patient comorbidities would impact on dose-related or dose–rate-related response and many DERS studies have examined pump operation around HMX alerts 29 30. Furthermore, as a study covering 1 year of data, we have not allowed for any changes to the device programming parameters being made in response to pump data reviews at each site.…”

Critical Care Nurses’ Knowledge of Correct Line Types for Administration of Common Intravenous Medications: Assessment and Intervention Study

“…The user cannot proceed with programming without confirming that they have read the advisory. In one study [ 12 ], a medication safety strategy that used centrally applied medication labels from the central compounding service to match the dose, dilution, advisories, and warnings as presented on the patient’s electronic medication administration record (aligned with the medication information available on the automated dispensing cabinet and integrated refrigeration unit and on the intravenous smart pump) was described. The drawback of this approach is that the nurse has often primed the intravenous administration line manually (either by gravity for a volume pump or by manual purge in the case of a syringe driver); therefore, the clinical advisory may be ignored in the interest of convenience, either because a considerable amount of pump programming has already taken place to get to the clinical advisory or in order to avoid wasting the medication that has already travelled through the intravenous administration set (this can be around 20 mL with most volume pump intravenous administration sets, thus, not an inconsiderable amount, particularly in the case of pediatric and neonatal patients).…”

“…Many intravenous smart pumps are not connected to a central server, particularly when they are deployed to dispersed high-dependency beds or to infection-isolation units. Wireless connectivity can facilitate updates, but with older intravenous pumps, updates are dependent on biomedical engineers uploading new libraries manually to each intravenous smart pump [ 12 , 13 ].…”

Evaluating An Automated Compounding Workflow Software for Safety and Efficiency: Implementation Study

“…It is likely, however, that pharmacists, given their workload, the reduced numbers of qualified staff available against a backdrop of increasing demand on health care services, and the closed nature of the sterile compounding unit, can only be present for “key stages” of the compounding process. For example, the key parts of the compounding (diluent, medication vials, closed system transfer devices, final administration IV bag, and recipe) may be shown to the pharmacist as a “guarantee” of correct constituents for compounding, but given that lookalike–soundalike errors remain prevalent in pharmacies (estimated at 25.9% of all errors) [ 9 ] and that in hurried checks the “4-eyes” process may only reinforce error rather than avert it [ 10 , 11 ], this is far from optimal. To this end, a compounding process that has other monitoring processes outside of the assumed presence and infallibility of a second human check is desirable.…”