A medication administration error (MAE) is defined as a medication error occurring during the bedside administration of a medication.1 Studies have shown that MAEs cannot be prevented by any combination of computerized physician order entry (CPOE), electronic health record (EHR), clinical decision support system, or an automated dispensing system but can only be mitigated by a closed-loop medication administration system.2,3,4,5 The closed-loop medication administration system includes automated identification technology such as radio-frequency identification (RFID) and/or barcodes.6,7 It is used to verify the so-called “five rights” (5R; right patient, right drug, right dose, right route, and right time) of medication administration at the bedside by cross-checking patients’ identification, prescription information, and dispensed drugs using a hand-held point-of-care device with RFID and/or a barcode reader.8,9
In 1999, the Institute of Medicine10 reported that medical errors result in 44,000–98,000 preventable deaths and more than 1,000,000 injuries each year in US hospitals. Medication errors can occur at any stage of the medication process, including prescribing, transcribing, dispensing, and administration.11
Previous research has indicated that closed-loop medication administration systems reduce non-intravenous MAEs to 39% and also reduce the administration of the wrong dose as well as omission errors.12 However, personal digital assistant (PDA) or hand-held point-of-care devices have been adopted in only 27.19% hospitals in China due to difficulties with application, high implementation costs, and maintenance fees.13
The present study was conducted at the University of Hongkong-Shenzhen Hospital (HKU-SZH) from August 1, 2016, to December 31, 2016. The hospital is a general teaching hospital affiliated with the University of Hongkong and owned by Shenzhen government. It opened in 2012 and has 2,000 beds and an out-patient capacity of 8,000–10,000 patients. The HKU-SZH has set up a wireless environment since its beginning and has adopted a clinical information system including CPOE, EHR, a clinical decision support system, and picture archiving and communication systems.
In 2015, the HKU-SZH reported 105 medication errors out of 515 incidents (20.39%). The aim of this study was to determine whether the development and implementation of a closed-loop medication administration system could reduce MAEs in the hospital.
Prior to commencing the study, a literature review was performed in order to analyze the workflow of medication administration in clinical settings and a diverse project team was set up, which consisted of an information technology (IT) manager, IT engineers, pharmacists, doctors, head nurses, and registered nurses (RNs). The project plan had four steps: (a) preparation period: develop the system and finish PDA procurement; (b) pilot study: apply the system in pilot wards and modify the system accordingly; (c) nurse training and application of the system to whole hospital; and (d) data collection and analysis and summarization of the findings in a paper.
The closed-loop medication administration system was developed using Microsoft Visual Studio Net 2005. The system adopted an object-oriented design, programmed with Microsoft Foundation Classes; repacked the controls; and perfected the system function. The client and server system consisted of three layers of logically separated frameworks: the user interface (UI) layer, the Buss’ rule layer, and the data access layer.
The closed-loop medication administration system requires technical support to obtain clinical information, which mainly includes the following: (a) in-patients wear barcoded wristbands, from which patient information can be read; (b) the closed-loop medication system is based on a CPOE system in order to achieve intelligent and structured prescription records; (c) an automatic drug dispensing system is applied to the pharmacy in order to integrate drug and patient information; and (d) nurses use PDAs and mobile workstations to identify patients, drug and prescription information, and match information and automatically obtain execution confirmation.
In the traditional paper-driven process of medication administration, two nurses manually double-check the medication information including name, dose, time, route of medication, and the patient’s identity before the medication is administered.14
With the closed-loop medication administration system, the process of medication administration requires nurses to scan the barcodes on the patient’s wristband as well as the barcodes on the medication before it is administered.15,16 If the dose being scanned corresponds with the approved medication order of a pharmacist and the patient is due to receive this medication, the administration is automatically documented in real time. However, if the medication does not correspond to a valid order, the system issues a warning.
Since the closed-loop medication administration system provides an additional layer of safety with the real-time scanning of barcodes,17 we changed the process of medication administration after application of the system from requiring two nurses at the bedside to perform the double-check to the following two methods: a single nurse (a) manually checked as routine at the bedside and (b) scanned the barcodes on both patient’s wristband and the medication before medication administration.
The study was implemented in four pilot general wards. We used a before-and-after design and collected data on oral medication administration times before and after the system implementation. We also evaluated the MAE alert logs after the implementation and surveyed the nurses’ satisfaction with the system.
Self-reported satisfaction questionnaires were used for the survey. A brainstorming meeting of the clinical head nurses was used to design the first draft of the questionnaire. A pretest of the questionnaire was conducted in the pilot wards; based on the results of the pretest, we further modified the questionnaire to develop the formal version. The formal questionnaire contained questions from five perspectives, including the quality of nursing, patient safety, nurse workload, work efficiency, and nurses’ views of the PDA. Each question was answered according to five degrees: strongly disagree, disagree, neutral, agree, and strongly agree.
Four nursing students observed and recorded the oral medication administration times in the four pilot wards respectively before and after intervention (December 12–16, 2016, and January 2–6, 2017). The satisfaction questionnaire survey was conducted in the four pilot wards on January 6, 2017, using the Wenjuanxing Internet platform (
IBM SPSS Statistics for Windows, version 20.0, was used for analysis. The nursing time required for medication administration was presented as the mean ± standard deviation. We applied two independent sample
The nursing times required for oral medication administration before and after implementation of the closed-loop medication administration system are shown in Table 1. The average nursing time of the four wards before the new system was 31.56 ± 10.88 minutes, which was reduced to 18.74 ± 5.60 minutes after the implementation of the system.
Average nursing times for oral medication administration before and after implementation of the system in the pilot wards. Note: Two independent samples t-tests showed a significant difference between the two groups. (Ward Average nursing time (M ± SD), min Before system adoption After system adoption Medical Ward 1 32.57 ± 8.72 15.73 ± 4.57 Medical Ward 2 41.89 ± 5.44 20.07 ± 2.25 Surgical Ward 1 29.33 ± 9.46 18.83 ± 7.25 Surgical Ward 2 22.70 ± 9.83 20.75 ± 6.80 Total 31.56 ± 10.88 18.74 ± 5.60
The nurses’ satisfaction with the closed-loop medication administration system is shown in Table 2.
Nurses’ satisfaction with the system (Items Strongly disagree Disagree Neutral Agree Strongly agree n % n % n % n % n % The system can facilitate your work and reduce your workload 5 7.14 11 15.71 21 30.00 19 27.14 14 20.00 The system can reduce check time and enhance work efficiency 3 4.29 9 12.86 22 31.43 19 27.14 17 24.29 The system can help to improve checking accuracy and reduce MAEs 2 2.86 0 0 10 14.29 34 48.57 24 34.29 The system can track MAEs to improve nursing quality 3 4.29 6 8.57 16 22.86 25 35.71 20 28.57 The degree of helpfulness of the system to your work 4 5.71 1 1.43 23 32.86 28 40.00 14 20.00
There were only 27 MEA alert logs from the repeated scans of 3,428 instances of medication administration during the observation period.
As Table 1 shows, there was a significant difference (
Table 2 shows that 60.00% (
The MAE alert rate was 1.22% during the one-year observation period in a previous study, suggesting that the closed-loop medication administration system contributed to improving patient safety by preventing potential MAEs.2 In the limited time since the implementation in the present study, there were only 27 MAE alert logs among 3,428 medication administrations during the observation period. The full impact of the closed-loop medication administration system on patient safety will be evaluated and analyzed in future studies.
Successful implementation and adoption of the closed-loop medication administration system are more likely to happen only when the developers and implementers understand the complexities and unpredictability of the nurses’ workflow. For instance, an inability to ensure the compatibility of the system with nursing workflow may lead to unintended consequences. Clearer guidance is required from hospitals on the use of the systems by physicians, nurses, and other medical staff. Without clear policies, well-developed systems that are well implemented and designed may hinder nursing workflow and impact patient safety and care. A key policy priority, therefore, is to plan for the long-term use of these systems.
This study had three limitations. First, the main weakness of this study is that data were only collected from four pilot wards. Second, the system has only been trialed for half a month, since the open tender process of the PDA was unsuccessful. Third, the project failed to follow the schedule. The MAE alert logs have been evaluated for half a month, which is insufficient for further analysis. The success of the pilot work has laid the foundation for further research and implementation of the system throughout the hospital.