Quality Assessment of Real-Time Location, Time and Temperature Data Generated by Active Radio Frequency Identification (RFID) Technology in Hospital Settings

Objective: RFID requirements specifications on test methods for the assessment of the performance of RFID systems in health care settings are limited. This paper proposes a systematic and detailed approach to analyze and assess the performance of an RFID system in the context of the particular healthcare setting

medication from the beginning to the end of the process of prescription-validation-dosing-preparation-administration [3] significantly improve hand hygiene compliance [4] and monitoring of hospitalized patients or detect and prevent them from bed falls [5][6], and to improve standardization and documentation of key steps within the blood collection and transfusion process in blood banks and hospitals [7].
RFID systems contain 2 basic components: tags and readers.
Tags are attached to people or objects and transfer their data to readers using radio waves. An active tag contains a battery and is able to collect and transfer data, such as temperature, humidity, location and its identity, without necessarily being activated by the electromagnetic field generated by a reader. In contrast, a passive tag does not contain a battery and is only able to transmit data while being activated by the electromagnetic field generated by the reader [8]. In comparison to other current identification technologies, RFID seems to be the most promising. For instance, barcodes on medical products have the disadvantage that they require active user interaction and that they must be read in a straight line [9]. Moreover, multiple barcodes on products, including those codes containing irrelevant information from previous steps in the process, might generate incorrect information when the wrong barcode is being scanned further in the process [10]. RFID has the advantage of transmitting data for multiple objects simultaneously, through physical barriers and from a distance. RFID tags can also hold more and more up-to-date information and can generate more accurate data in relation to information systems depending on hand-based data entry [10].
In general, new information and communication technologies might bring new risks to patient health or data quality issues through for example errors in computer software. Studies on pilot testing of RFID systems within hospital settings show, for instance, harmful electromagnetic interference on medical equipment and concerns about the potentials and obstacles to the adoption of these technologies [11][12][13][14][15]. Several factors within a hospital setting might influence the performance (needs) of a RFID system . First, environmental factors, like aluminium films, liquid, metal and concrete walls might influence the performance and quality of data generated by a RFID system [11]. When for example radio waves broadcasted by RFID tags are blocked by fluids, such as Red Blood Cell products (RBCs), the data may not reach the reader resulting in incomplete data. Second, RFID systems themselves might influence the performance of physical objects in their environment through for example electromagnetic interference (EMI) on medical equipment [12][13][14][15]. EMI subsequently might jeopardize the care processes and compromise patient safety. Finally, the specific aims of implementing a RFID technology set specific requirements on the data to be generated [11].
The aim of this study is to report on the results concerning the assessment of the accuracy and completeness of location, time and temperature data generated by an active RFID system used for the tracking and tracing of RBCs inside an academic hospital. Care Unit (ICU) of an academic hospital.

Figure 1:
A schematic overview of the detailed steps to be taken to perform RFID data accuracy assessments.

Figure2:
A schematic overview of the detailed steps to be taken to perform RFID data completeness assessments.
An assessment study on possible Electromagnetic Interference (EMI) by RFID on medical equipment within our academic hospital setting, representing phase E of our framework, showed that RFID technology might indeed induce potentially hazardous incidents in medical devices in a controlled trial setting on-site [12]. The last phase I of our framework concerns the actual evaluation of the added value of a RFID technology in the context of the aims set for introducing an RFID system. After the data validation study described in this paper, we evaluated whether the management of blood transfusion chain concerning RBCs complied to European and Dutch guidelines by tracking and tracing these RBCs with our RFID system inside our hospital.

Background
This study was performed within the Academic Medical Center (AMC), a large university hospital in Amsterdam, The Netherlands.

RFID System
The selected active RFID system (Eureka RFID, Avonwood, UK) has a 125 kHz reader (68x10E-3 µT at 1m) that forces data transmission by any tag in its proximity. The active RFID tags have an operational frequency of 868 MHz at 2µW. RFID active tags containing a temperature sensor will be used for tracking and tracing of the RBCs.
On one side of the tag a copper pin is pointing out. This pin senses the temperature at its surface and guides the temperature to the actual sensor positioned inside the tag. Temperature data can be stored on the tag itself or immediately be broadcasted, in combination with other data-like a RBC's location, to the reader at predefined timestamps or when being inside the field of the reader that forces data transmission by the tag in its proximity. Within this validation study, tags were placed inside plastic containers that were attached to the blood bags surfaces with the copper pins placed towards the blood bags surfaces.
After data transmission by a tag, the receiver sent data concerning the tag's identification, its current location, the current time, the current temperature sensed by the temperature sensor and possible other data, like for example its battery status, through the local area network (LAN) to a central Oracle 10g database.
Every 8 minutes, a temperature sensor would additionally record temperature data in its memory for final storage in this database.

Data Requirements
The requirements concerning time, temperature and location data to be generated by the RFID system data were set with reference to the intra-hospital guidelines concerning the storage, transport and distribution conditions of RBCs and concerned the following: a.
The RFID tags should be able to record and broadcast temperatures with an accuracy of 0. 5  The specification of the required accuracy of the corresponding timestamps was to realize the generation of data about the whereabouts and conditions of RBCs as close to reality as possible.

Controlled Field Setting
Laboratory Setting: The RFID tags were tested on their generation of accurate location, time and temperature data in a laboratory setting first. Figure Table 1.   Figure 4.

Figure 4:
Median temperatures and their ranges recorded by 13 randomly selected RFID temperature tags concerning their speed of adjustment to changing temperature circumstances and their relation to the temperatures measured by a data logger and a quicksilver thermometer.

Data Completeness Tests in A Real-Life Setting
During the larger research study, 37 temperature tags containing a temperature sensor were randomly attached to RBCs and used for tracking and tracing of the RBCs.

Data Cleaning
Before the analysis on the completeness of the RFID generated datasets could be performed, the following data cleaning steps were carried out. First, bouncing effects were managed to avoid data overlapping and gaps in timelines. Bouncing effects are caused by two readers recording data while the tag is passing in one direction but transiently appearing at one and the other reader.
Second, because our research focused on the tracking of RBCs being transported between rooms and on monitoring temperature changes, only the last and first records of a dataset containing the same location and temperature data were maintained. All other data records were removed. When no change in temperature of an RBC had been recorded at a specific OR for instance, only the first and last record containing time, location temperature data and temperature data within the same temperature range at that specific location were maintained for the analyses. When the temperature of an RBC had increased or decreased, the first and

Laboratory Setting
The accuracy tests of time and location data were performed by having 2 test persons carry randomly chosen tags through a doorway covered with two readers. The results showed accurate data generated by the tags concerning their location and times. During these tests the 2 different rooms that were 'created' with the 2 readers could be discriminated from each other. Also, no bouncing effects or tag drops were experienced during these tests. The tags generated corresponding time stamp data with an accuracy of at least one hundredth of a second (hand-recording).

Accuracy of Temperature Data Recordings by Tags in A Laboratory Setting
Seventeen (17)

Accuracy of Real-Time Temperature Data Measurements and Broadcasts in A Laboratory Setting
Thirteen tags of a total of 37 were assessed on their accuracy concerning real-time generated and broadcasted temperature data in water baths varying in temperatures from 1.0°C to 37.0°C. The maximum difference between the temperature measured by the quicksilver thermometer and the temperature measured by tags was 1.0°C. On average the temperatures measured and broadcasted by the selected tags were higher than the temperatures measured by the quicksilver thermometer.

Controlled Field Setting
The accuracy of location and time stamp data was tested in a controlled field setting by two persons hand recording the location and time of 10 randomly chosen tags that were moved around the OR. The location and time data generated by 10 randomly chosen RFID tags recorded in the Oracle database corresponded to the data that was registered by hand with a maximum difference of 2 seconds. This difference resulted from the registration of location and time data by hand, which were less accurate than the RFID generated data. These two persons were not able to register time and location data as precise as the RFID system itself. Also no bouncing effects were seen.

Field Setting
Sixteen RFID temperature tags out of a total of 37, were assessed on their speed of adjustment to changing temperature circumstances concerning their real-time temperature data measurements and broadcasts. Figure 4

Discussion
Based on a literature review on current insights of RFID implementations in healthcare, we previously developed a framework describing a systematic approach of 9 phases that can be used for assessing the feasibility and impact of using a RFID technology on its health care environment and vice versa [11]. In this paper, the validation study on the accuracy and completeness of location, time and temperature data generated by an active RFID system to be used for the tracking and tracing of RBCs inside the AMC is described. The outcomes of this validation study allowed us to determine whether our RFID system was capable of generating data of such a quality that it could be used in analysing to what extent the transfusion blood chain within the AMC complies to European and Dutch guidelines and regulations on preservation of blood products. The tests concerning the accuracy of location and time data generated by active RFID tags showed minimum differences compared to hand-recorded time and location data.
The tests on the accuracy of temperature data generated by the tags revealed that both real-time recorded temperature data were on average 0.26ºC and 0.5ºC higher compared to the temperature data measured by data a logger and a quicksilver thermometer Then, the RFID technology used in this study might not work properly when being used for tracking and tracing of other objects, frozen plasmas or blood biosamples for instance, which need RFID tags that are able to resist temperatures between -30°C and -80°C [17]. In other circumstances, RFID tags might need to be able to resist electromagnetic irradiation produced by medical devices such as Computer Tomography or Magnetic Resonance Imagining for instance [18]. RFID systems differing in characteristics and varying environments of RFID implementation both might affect signal penetration and coverage and might cause variations in tag blocks or data transmission failures by tags. Third, when the use of RFID serves another purpose, unique identification and accurate real time location of medication and patients [19], or of clinicians and patients in a healthcare facility [20] for instances, the required data quality might differ from the requirements set for data accuracy and completeness in this validation study or be much easier to accomplish. When RFID would be used to track medical devices, an RFID tag would be attached to each piece of equipment and RFID readers would have to be strategically located throughout the area of operations where this equipment is used. In this way, any medical device could be traced merely on the basis of the location data generated by the tag.
These three scenarios exemplify that RFID system implementations should always be preceded by data accuracy and completeness tests within the context of the processes being supported. Our validation study illustrates the need of such principle. Another limitation of this study might be the data quality test method itself, because it is not a fully tested and proved method at this stage. As mentioned earlier the data quality assessment tests are part of our framework. This also accounts for the EMI-tests that were performed before these data quality tests took place.
For these EMI-tests we used a method already been applied in performance tests concerning technology comparable to RFID, i.e. mobile phones. Methods prescribing how to perform data quality tests concerning RFID systems themselves were yet not available at the time we conducted our pre-installation assessment study concerning the performance of our RFID system in the AMC setting.

Conclusion
this study shows that the tested RFID technology is capable of generating accurate location, temperature and time data in a specific health care setting. The completeness of RFID generated data may yet depend on the building structure, the objects RFID performance tests as prescribed in different phases as of our framework prove to be a solid foundation for assessing those environmental circumstances influencing the performance of RFID systems and vice versa systematically, at least in hospital settings. The RFID performance assessment tests concerning data quality that we present could at least act as a reference model for conducting data quality tests on RFID systems in other hospitals.

Authors' Contributions
RvdT, LWDP and MWJ designed the study, RvdT performed the measurements, the data analysis and drafted the manuscript. MWJ coordinated and supervised the designing of the study, the data analyses, and the drafting of the manuscript. All authors read and approved the final manuscript.

Funding
This research was funded by the Dutch Ministry of Health, Welfare and Sport and the Ministry of Economic Affairs, the Academic Medical Center, Capgemini, Geodan, Oracle and Intel.
They had no influence whatever on the content of this manuscript.