A novel method for inferring RFID tag reader recordings into clinical events.

BACKGROUND

Nosocomial infections (NIs) are among the important indicators used for evaluating patients' safety and hospital performance during accreditation of hospitals. NI rate is higher in Intensive Care Units (ICUs) than in the general wards because patients require intense care involving both invasive and non-invasive clinical procedures. The emergence of Superbugs is motivating health providers to enhance infection control measures. Contact behavior between health caregivers and patients is one of the main causes of cross infections. In this technology driven era remote monitoring of patients and caregivers in the hospital setting can be performed reliably, and thus is in demand. Proximity sensing using radio frequency identification (RFID) technology can be helpful in capturing and keeping track on all contact history between health caregivers and patients for example.

OBJECTIVES

This study intended to extend the use of proximity sensing of radio frequency identification technology by proposing a model for inferring RFID tag reader recordings into clinical events. The aims of the study are twofold. The first aim is to set up a Contact History Inferential Model (CHIM) between health caregivers and patients. The second is to verify CHIM with real-time observation done at the ICU ward.

METHOD

A pre-study was conducted followed by two study phases. During the pre-study proximity sensing of RFID was tested, and deployment of the RFID in the Clinical Skill Center in one of the medical centers in Taiwan was done. We simulated clinical events and developed CHIM using variables such as duration of time, frequency, and identity (tag) numbers assigned to caregivers. All clinical proximity events are classified into close-in events, contact events and invasive events. During the first phase three observers were recruited to do real time recordings of all clinical events in the Clinical Skill Center with the deployed automated RFID interaction recording system. The observations were used to verify the CHIM recordings. In second phase the first author conducted 40 h of participatory observation in the ICU, and observed values that were used as golden standard to validate CHIM.

RESULTS

There were a total of 193 events to validate the CHIM in the second phase. The sensitivity, specificity, and accuracy of close-in events were 73.8%, 83.8%, and 81.6%; contact events were 81.4%, 78.8%, and 80.7%; and invasive events were 90.9%, 98.0%, and 97.5% respectively.

CONCLUSION

The results of the study indicated that proximity sensing of the RFID detects proximity events effectively, and the CHIM can infer proximity events accurately. RFID technology can be used for recording complete clinical contact history between caregivers and patients thus assisting in tracing cause of NIs. Since this model could infer the ICU activities accurately, we are convinced that the CHIM can also be applied in other wards and can be used for additional purposes.